A method and system for predicting the life of a solid-state drive based on big data

Abstract

The present invention relates to the technical field of hard disk life prediction, and in particular to a method and system for predicting the life of a solid-state hard disk based on big data. The method comprises the following steps: obtaining solid-state hard disk structure data; performing operation simulation based on the solid-state hard disk structure data, thereby generating solid-state hard disk operation simulation data; using the solid-state hard disk operation simulation data to collect power fluctuation data of flash memory units in the hard disk on the solid-state hard disk structure data, and obtaining power fluctuation data of flash memory units in key areas of the solid-state hard disk; classifying solid-state hard disk operations on the solid-state hard disk operation simulation data, and generating write operations and erase operations. The present invention improves the accuracy of life prediction for solid-state hard disks through multi-dimensional data collection and analysis, combined with factors such as hard disk operation, power fluctuation, electron migration effect, and material fatigue.

Description

Technical Field

[0001] The present invention relates to the technical field of hard disk life prediction, and in particular to a method and system for predicting the life of a solid-state hard disk based on big data. Background Art

[0002] Solid-state drives (SSDs), as a new type of storage medium, are widely used in personal computers, servers, and data centers due to their high operating speed, low power consumption, and shock resistance. However, the life-limiting factor of SSDs is mainly determined by the number of erase and write times of the flash memory chip. With frequent data writing, the flash memory unit will gradually wear out, resulting in a decrease in storage performance or even failure. With the development of artificial intelligence and big data technologies, prediction methods based on machine learning have gradually become a hot topic of research. By collecting and analyzing the health data of SSDs on a large scale and building machine learning models such as support vector machines (SVMs), random forests, and neural networks, it is possible to more accurately identify the changing trend of SSD life and provide real-time life warning. However, the existing life prediction methods usually rely on basic indicators such as the number of writes (P / E cycles), remaining available space, temperature, and battery life of the hard disk, but they cannot fully reflect the internal physical degradation of the hard disk during long-term use, especially the performance degradation under large-scale high-frequency writing and high-load conditions. At the same time, conventional methods usually make predictions based on the overall working parameters of the hard disk (such as the number of writes and erases). These methods assume that each hard disk has a similar life under the same load and working conditions. However, in actual use, differences in workloads for different users or application scenarios (such as high-frequency writing, large data throughput, etc.) can lead to significant differences in hard drive life, resulting in lower accuracy in the prediction of SSD life. Summary of the invention

[0003] Based on this, it is necessary to provide a solid state drive life prediction method and system based on big data to solve at least one of the above technical problems.

[0004] To achieve the above purpose, a method for predicting the life of a solid state drive based on big data is provided, the method comprising the following steps:

[0005] Step S1: acquiring the solid state drive structure data; performing operation simulation based on the solid state drive structure data, thereby generating the solid state drive operation simulation data; using the solid state drive operation simulation data to collect the power fluctuation data of the flash memory unit in the hard disk on the solid state drive structure data, and obtaining the power fluctuation data of the flash memory unit in the key area of ​​the solid state drive;

[0006] Step S2: classifying SSD operation on SSD operation simulation data to generate write operation and erase operation; performing SSD abnormal change trend analysis on power supply fluctuation data of flash memory units in key areas of SSD based on write operation and erase operation to generate SSD abnormal change trend data; using SSD abnormal change trend data to identify electron migration effect on SSD to generate SSD electron transition identification data;

[0007] Step S3: acquiring the time series data of the solid state drive; scanning the internal storage medium of the solid state drive on the time series data of the solid state drive to obtain a data set of scanning the internal storage medium of the solid state drive; identifying the fatigue effect of the nano-material of the solid state drive on the data set of scanning the internal storage medium of the solid state drive to generate data of identifying the fatigue effect of the hard drive;

[0008] Step S4: Perform hard disk usage correlation analysis on the solid-state hard disk electronic transition identification data and the hard disk usage fatigue effect identification data to generate solid-state fatigue characteristic mapping data; perform hard disk real-time life prediction based on the solid-state fatigue characteristic mapping data to generate hard disk real-time life prediction data; adaptively write or warn the hard disk real-time life prediction data through a preset standard life warning threshold, thereby executing the solid-state hard disk life extension optimization operation.

[0009] The present invention can deeply understand the performance of each component inside the hard disk by obtaining the structural data of the SSD and performing operation simulation based on these data. This step is not only based on surface data, but also simulates the actual operation of the SSD through comprehensive simulation, thereby obtaining more accurate operation data. By collecting the power fluctuation data of the flash memory unit inside the hard disk, the electrical changes of the hard disk in actual use can be captured, which is very important for accurately identifying the early signs of hard disk performance degradation, especially for those problems that cannot be directly reflected by traditional SMART data. By distinguishing between write operations and erase operations, the impact of each operation on the hard disk can be finely analyzed. The different modes and frequencies of write and erase operations will directly affect the service life of the hard disk. This distinction can better understand its specific impact on the life of the SSD. Based on the operation data, the trend analysis of power fluctuations can capture the abnormal changes of the hard disk in real time. This makes the life prediction more accurate and can provide early warning before the hard disk has potential failures. Electron migration effects (such as electron transitions) are one of the key factors in the decline of hard disk life. The patent can effectively identify potential loss problems in advance by analyzing these microscopic electrical effects. By obtaining the hard disk usage timing data, the use history of the hard disk can be tracked and the life analysis can be performed over time. This time-series-based data analysis helps to capture the changing characteristics of the hard disk at different stages of use and improve the timeliness of the prediction. By scanning the internal storage medium of the hard disk, the nano-level material fatigue effect of the hard disk can be identified. This identification can not only find microscopic damage that cannot be detected by traditional methods, but also conduct a comprehensive analysis of factors such as the material properties and working environment of the hard disk. By combining the electronic transition identification data and the hard disk use fatigue effect identification data, correlation analysis can be performed to analyze the health status of the SSD at multiple levels. This comprehensive analysis method overcomes the defect that the traditional life prediction method only relies on a single indicator, and improves the accuracy and reliability of the prediction results. Based on the above analysis results, real-time life prediction can be performed. This can not only provide early warning for hard disk users, but also optimize the use strategy of the hard disk, such as automatically adjusting the write strategy or timely data backup to avoid data loss. Therefore, the present invention improves the accuracy of life prediction for solid-state hard disks through multi-dimensional data acquisition and analysis, combined with factors such as hard disk operation, power supply fluctuations, electronic migration effects and material fatigue.

[0010] Preferably, step S1 comprises the following steps:

[0011] Step S11: Acquire the solid state hard disk structure data;

[0012] Step S12: performing physical topology modeling on the SSD structure data to generate SSD topology structure modeling data; performing operation simulation on the SSD based on the SSD topology structure modeling data to generate SSD operation simulation data;

[0013] Step S13: marking the key area of ​​the SSD structure data using the SSD running simulation data to generate the SSD key area marking data;

[0014] Step S14: Based on the integrated current and voltage sensors, the power supply fluctuation data of the flash memory unit in the SSD is collected for the key area mark data of the SSD to obtain the power supply fluctuation data of the flash memory unit in the key area of ​​the SSD.

[0015] The present invention can deeply analyze the structure and operation characteristics of the solid-state hard disk through physical topology modeling and operation simulation, and provide a scientific basis for subsequent optimization and improvement. Using operation simulation data to mark the key areas of the hard disk helps to accurately identify the parts that have the greatest impact on performance and reliability, and provides support for targeted analysis and optimization. Through the integration of current and voltage sensors, accurate collection of flash memory unit power fluctuations is achieved, which helps to gain an in-depth understanding of power stability issues in key areas and provide data support for hard disk performance improvement and reliability assessment. Each step is interrelated, from structure to operation, from simulation to physical collection, forming a closed-loop analysis process, which lays a solid foundation for the design optimization, fault prediction and reliability improvement of solid-state hard disks. By systematically acquiring and analyzing key data, potential performance bottlenecks and reliability risks can be effectively identified, thereby optimizing hard disk design and improving its overall performance and life.

[0016] Preferably, running simulation on the solid state drive based on the solid state drive topology structure modeling data includes:

[0017] Setting operation simulation rules for the solid state drive based on the solid state drive topology structure modeling data to obtain the solid state drive operation simulation rules, wherein the solid state drive operation simulation rules include operation sequence, data block size, and operation frequency;

[0018] According to the preset operation sequence, data block size and operation frequency, the operation parameters of the solid state drive are configured to generate operation simulation rule configuration data;

[0019] The solid-state drive is simulated in a virtual environment using the simulation rule configuration data to generate a solid-state drive simulation data set; the solid-state drive simulation data set is preprocessed to generate solid-state drive simulation data, wherein the data preprocessing includes data cleaning, data denoising, data missing value filling and data standardization.

[0020] The present invention can fully simulate actual usage scenarios and cover multiple types of workloads by setting operation simulation rules (such as operation order, data block size and operation frequency). Flexible adjustment of rule parameters helps to simulate the performance and behavior of solid-state hard disks under different conditions, and provides a diversified test environment for hardware and algorithm optimization. Configuration data is generated based on the operation simulation rules to make the simulation more in line with the real application scenario and ensure the reliability and accuracy of the results. The systematic generation of configuration data helps to quickly build a virtual test environment and improve simulation efficiency. Operation simulation in a virtual environment can simulate a variety of usage conditions, reduce dependence on actual hardware, and save test resources and costs. Virtual simulation supports high-frequency iterative testing, which helps to quickly discover potential problems. The quality and reliability of simulation data sets are improved through preprocessing steps such as data cleaning, denoising, missing value filling and standardization. High-quality data sets provide a solid foundation for subsequent analysis, modeling and optimization, avoiding misjudgment due to noise or missing data. The simulation data set can be used to analyze the operation performance bottleneck of solid-state hard disks and help discover design defects or potential problems in operation. Provide reliable data support to guide the optimization of storage controller algorithms, improve operation efficiency, and improve the durability and stability of hard disks. By flexibly adjusting simulation rules and parameters, different workloads (such as random operation, large data stream transmission, etc.) can be simulated to make the test results more universal, providing a reference for optimized design for specific scenarios (such as enterprise-level storage, consumer-level storage, etc.).

[0021] Preferably, step S2 comprises the following steps:

[0022] Step S21: classifying SSD operation on SSD operation simulation data to generate SSD operation classification data, wherein the SSD operation classification data includes write operation and erase operation; performing write current fluctuation screening on power fluctuation data of flash memory units in key areas of the SSD based on the write operation to obtain write operation current fluctuation data;

[0023] Step S22: performing an erase voltage fluctuation screening on the power fluctuation data of the flash memory unit in the key area of ​​the solid state hard disk based on the erase operation to obtain the erase operation voltage fluctuation data; performing a fluctuation amplitude change analysis on the write operation current fluctuation data and the erase operation voltage fluctuation data to generate the operation amplitude change data and the operation current peak change data;

[0024] Step S23: performing abnormal fluctuation analysis on the operation amplitude change data and the operation current peak change data to perform abnormal change trend analysis on the solid state hard disk, and generating abnormal change trend data on the solid state hard disk;

[0025] Step S24: using the abnormal change trend data of the solid state hard disk to identify the electron migration effect of the solid state hard disk, and generating the electron transition identification data of the solid state hard disk.

[0026] The present invention clarifies the key operation behavior characteristics of the solid-state hard disk by classifying the write operation and the erase operation of the simulation data. The screening method of the write current fluctuation and the erase voltage fluctuation can focus on the power supply change of the flash memory unit in the key area, and provide fine-grained data support for performance analysis. The amplitude and peak value changes of the write current fluctuation data and the erase voltage fluctuation data are analyzed to quantify the dynamic characteristics during the operation. The fluctuation analysis results provide refined performance indicators of the solid-state hard disk operation, which helps to find potential problems (such as excessive current or voltage fluctuations). Based on the operation amplitude and current peak change data, abnormal fluctuation analysis can be performed to effectively identify abnormal behavior or potential fault signs in operation. Abnormal change trend data provides a scientific basis for predicting the life of the hard disk, reliability assessment and fault warning. By using the abnormal change trend data for electron migration effect analysis, material migration problems caused by high current or high voltage can be effectively identified to avoid the failure of the flash memory unit. Providing a deep understanding of the electron migration effect helps to optimize the circuit design and operation strategy and improve the durability of the hard disk. Combining classification, screening, fluctuation analysis and trend identification, a closed-loop process from data to decision-making is formed, which provides data support for optimizing the performance and reliability of the solid-state hard disk. It can detect and avoid potential problems in advance during the design phase, and can be used for real-time monitoring and early warning during the operation phase to reduce the risk of failure during use. The generation of electronic transition recognition data can guide material and structural optimization, such as reducing the risk of failure caused by excessive current density. The data-driven feedback mechanism helps to continuously improve the manufacturing and design processes of solid-state drives and improve product quality.

[0027] Preferably, step S24 includes the following steps:

[0028] Step S241: defining the current-voltage signal of the abnormal change trend data of the solid state hard disk based on a preset time window to generate a current signal and a voltage signal; calculating the instantaneous value of the current signal and the voltage signal to obtain the instantaneous value of the current and the instantaneous value of the voltage;

[0029] Step S242: constructing a phase space trajectory through the instantaneous current value and the instantaneous voltage value to generate a current-voltage phase diagram of the solid state drive; performing trajectory overlap detection on the current-voltage phase diagram of the solid state drive to generate a phase trajectory overlap curve of the solid state drive;

[0030] Step S243: identifying the current-voltage mutation point of the solid-state hard disk current-voltage phase diagram according to the solid-state hard disk phase trajectory overlap curve, thereby obtaining the current-voltage mutation point;

[0031] Step S244: Use the current-voltage mutation point to mark the hard disk electronic transition mutation signal on the current signal and the voltage signal to obtain the hard disk electronic transition mutation marking signal; calculate the proportion of the hard disk electronic transition mutation marking signal to the current signal and the voltage signal to generate the solid state hard disk electronic transition identification data.

[0032] The present invention defines and extracts current signals and voltage signals based on a preset time window to ensure that the signals have time continuity and analytical properties. The instantaneous value of the signal is calculated to capture the key details in the dynamic changes, laying a data foundation for the subsequent trajectory construction. The accurate current and voltage signal characteristics are provided to help identify short-term dynamic behaviors and improve the detection sensitivity. The current-voltage phase space trajectory diagram of the solid-state hard disk is constructed to intuitively display the dynamic evolution path of the signal. Through trajectory overlap detection, a phase trajectory overlap curve is generated to reveal the dynamic relationship between current and voltage in key areas. The nonlinear change characteristics of the current-voltage signal are clearly identified, especially providing visualization support in abnormal or sudden changes. Based on the phase trajectory overlap curve, the mutation point in the current-voltage phase diagram is located with high precision. Mutation point identification can reveal the specific time and amplitude of the sharp change of current or voltage during the operation of the hard disk. The accurate identification of mutation points helps to discover potential electronic transition phenomena and risk signals in the operation of the hard disk. The signal is marked for electronic transition through mutation points to further refine the mutation signal characteristics. The proportion of current and voltage signals is calculated, the influence of electronic transition is quantified, and the final electronic transition identification data is generated. The electron transition mutation mark and proportion analysis provide a reliable quantitative basis for the material property research, circuit optimization design and fault prediction of hard disks. The full-process analysis method of trajectory construction, overlap detection and mutation mark deeply reveals the correlation between hard disk current and voltage. The mutation point and electron transition identification provide a scientific basis for detecting and preventing sudden abnormalities in hard disk operation.

[0033] Preferably, step S3 comprises the following steps:

[0034] Step S31: obtaining the solid state hard disk usage timing data;

[0035] Step S32: performing time segmentation on the SSD usage time series data to generate the SSD usage time series segmentation data; using the SSD usage time series segmentation data as a scanning time interval, and scanning the internal storage medium of the SSD using a scanning probe microscope technology according to the scanning time interval to obtain a SSD internal storage medium scanning data set;

[0036] Step S33: performing hard disk nano-material use fatigue effect identification on the internal storage medium scanning data set of the solid state hard disk to generate hard disk use fatigue effect identification data.

[0037] The present invention obtains the time series data of the solid state hard disk, reflects the time series characteristics of the hard disk during operation, and comprehensively covers different operation stages such as operation operation, idle state and data migration. The time series data is segmented and analyzed to convert the complex time series information into structured time series segmented data. Time series data analysis provides a comprehensive view of the dynamic operation state and provides an accurate time window for subsequent scanning operations. The time series segmented data of the solid state hard disk is used as the scanning time interval, and the scanning probe microscope technology is used to perform an in-depth scan of the storage medium inside the hard disk. The scanning technology has a nanometer resolution and can accurately capture the physical form and material changes of the storage medium. The scanning probe microscope technology ensures the high resolution and high precision of the storage medium scanning, and provides detailed basic data for analyzing the material changes inside the hard disk. The scanning data set is deeply analyzed to identify the fatigue effects of the nano-level materials of the hard disk under long-term use, such as microcracks, surface morphology changes and local degradation. The generated fatigue effect identification data can quantify the degree of material degradation and reveal the fatigue mechanism under different operating conditions. Fatigue effect identification provides a quantitative indicator for the study of material aging process, which helps to predict the risk of hard disk failure in advance. By combining timing segmentation and scanning results, the health status and operational reliability of the internal storage media of the solid-state drive are systematically evaluated. Early identification of fatigue effects can help design more efficient fault monitoring and early warning systems. Significantly improve the health monitoring capabilities of the hard disk, reduce the risk of sudden failures, and improve user data security. Based on fatigue effect identification data, the material selection of storage media can be optimized, circuit design can be improved, and the use strategy of operating operations (such as balanced erase and temperature control) can be adjusted. The data support provided will help develop more durable and efficient solid-state drive products. Extend the service life of the hard disk, improve the market competitiveness of the product, and reduce the user's maintenance costs.

[0038] Preferably, step S33 includes the following steps:

[0039] Step S331: performing abnormal point detection on the data set of the scanned storage medium inside the solid state drive to generate a data set of characteristics without abnormality; performing nanoscale strain distribution analysis on the data set of characteristics without abnormality to generate strain distribution characteristic data of the storage medium;

[0040] Step S332: performing crystal structure degradation detection on the internal storage medium scan data set of the solid state drive according to the storage medium strain distribution characteristic data to generate crystal degradation characteristic data, wherein the crystal structure degradation detection includes dislocation density detection and grain boundary migration detection;

[0041] Step S333: Perform charge transfer characteristic analysis on the solid state drive electronic transition identification data to generate solid state drive charge transfer characteristic data; perform fatigue degradation index calculation on the solid state drive internal storage medium scan data set using the crystal degradation characteristic data and the solid state drive charge transfer characteristic data to obtain hard drive use fatigue effect identification data, wherein the fatigue degradation index calculation formula is as follows:

[0042] ;

[0043] ;

[0044] ;

[0045] ;

[0046] In the formula, Expressed as fatigue degradation index, Expressed as the crystal degradation coefficient, Expressed as the charge transfer accumulation characteristic coefficient, Expressed as the environmental coupling factor coefficient, Expressed as a characteristic fluctuation intensity value, Expressed as the crystal degradation characteristic weight coefficient, Expressed as the charge transfer characteristic weight coefficient, Expressed as the weight coefficient of the environmental coupling factor, Expressed as the characteristic fluctuation intensity correction coefficient, Expressed as The local lattice distortion value of a crystal, is the initial value of the overall lattice parameter of the crystal, Expressed as the number of crystals, Expressed as time The charge distribution change function inside is Represents the time limit, Expressed as the standard deviation of the crystal degradation characteristics, Expressed as the standard deviation of the charge transfer accumulation characteristic.

[0047] The present invention generates a data set without abnormal characteristics by detecting abnormal points in the data set of the storage medium scan, effectively removing noise and interference information, and ensuring the purity and accuracy of the analysis data. Nanoscale strain distribution analysis provides the spatial characteristic distribution of the force and deformation inside the material, laying the foundation for crystal degradation detection. Accurately extract the microscopic physical properties of the storage medium to ensure the scientific nature of fatigue effect identification. Combined with dislocation density detection and grain boundary migration detection, the key characteristic data of material crystal degradation are revealed from the microstructural level. Dislocation density reflects the degree of defect accumulation inside the crystal, and grain boundary migration reveals the changes in the interaction between grains. Providing the degradation characteristics of the microscopic crystal structure provides a scientific basis for predicting the life and stability of the storage medium. Based on the solid-state hard disk electronic transition identification data, the charge migration characteristic analysis is performed to quantify the dynamic distribution of charge in the storage medium and its cumulative changes. The charge migration characteristic data reflects the trend of electrical performance changes of the material during long-term use. Provide quantitative data support for in-depth research on the mechanism of electrical performance degradation and optimize hard disk design and material selection. Fatigue degradation index Through a comprehensive multi-parameter evaluation of crystal degradation, charge migration, environmental coupling and characteristic fluctuation intensity, the fatigue degree of the storage medium is fully reflected. The parameter formula covers lattice distortion, dynamic charge distribution and its time accumulation characteristics, providing multi-dimensional quantitative indicators for fatigue analysis. The fatigue degradation index provides a comprehensive evaluation indicator of the health status of the storage medium, which helps to identify potential failure risks in advance. Combining microscopic crystal characteristics and electrical performance data, a multi-level and multi-angle fatigue effect evaluation is achieved. Through quantitative indicators, a reliable health status monitoring system is established to provide a scientific basis for real-time maintenance and fault warning of hard disks. Enhance the reliability monitoring and stability evaluation capabilities of solid-state hard disks and improve user data security.

[0048] Preferably, step S4 comprises the following steps:

[0049] Step S41: performing hard disk usage association analysis on the solid state hard disk electronic transition identification data and the hard disk usage fatigue effect identification data to generate solid state fatigue characteristic mapping data;

[0050] Step S42: performing real-time hard disk life prediction based on the solid-state fatigue characteristic mapping data to generate real-time hard disk life prediction data; comparing the real-time hard disk life prediction data with a preset standard life warning threshold, and performing life warning processing when the real-time hard disk life prediction data is less than the preset standard life warning threshold;

[0051] Step S43: When the real-time life prediction data of the hard disk is greater than or equal to the preset standard life warning threshold, the solid state drive is adaptively written based on the real-time life prediction data of the hard disk, thereby performing the solid state drive life extension optimization operation.

[0052] The present invention generates solid fatigue characteristic mapping data by correlating and analyzing solid state hard disk electronic transition identification data with fatigue effect identification data, and establishes a correlation relationship between the working state and service life of the hard disk. The characteristic mapping data reveals the performance degradation mode and key fatigue characteristics of the hard disk under different working conditions. It provides a deep-level analysis basis for the service life and performance change of the hard disk, which is helpful to understand the dynamic correlation between fatigue characteristics and electronic transition. Real-time life prediction is performed based on solid fatigue characteristic mapping data, and real-time hard disk life prediction data is generated. By comparing with the preset standard life warning threshold, the potential hard disk failure risk is discovered in advance. When the predicted data is lower than the standard warning threshold, the life warning processing is triggered to provide a reliable risk prompt for the user. The hard disk life is accurately predicted, the equipment safety is improved, and the risk of accidental data loss is reduced. For hard disks whose real-time life prediction data is within the normal range, the adaptive writing technology is used to optimize the write distribution to reduce the overuse and fatigue accumulation of the local storage area. Adaptive writing can dynamically adjust the data storage strategy and delay the degradation process of the hard disk. The service life of the hard disk is extended, and the performance stability and reliability of the hard disk are significantly improved. Combined with real-time life prediction and life extension optimization operations, closed-loop control of hard disk life management is achieved. By continuously optimizing the hard disk write strategy, the generation of write hot spots is reduced and unnecessary hard disk aging is avoided. The resource utilization efficiency of the hard disk is optimized, and the frequency and cost of hard disk replacement are reduced. Life warning processing provides users with sufficient time for data migration and backup, reducing the risk of accidental data loss. Providing real-time life prediction data helps users manage the use of hard disk devices more scientifically, improves user experience, and enhances data security.

[0053] Preferably, performing real-time hard disk life prediction based on solid-state fatigue characteristic mapping data includes:

[0054] Extract key characteristic indicators from the solid-state fatigue characteristic mapping data to obtain hard disk fatigue characteristic indicators; divide the solid-state fatigue characteristic mapping data into data sets based on the hard disk fatigue characteristic indicators to generate a model training set and a model test set;

[0055] The model training set is trained using the long short-term memory neural network algorithm to generate a hard disk real-time life prediction pre-model; the hard disk real-time life prediction pre-model is optimized and iterated using the model test set to generate a hard disk real-time life prediction model;

[0056] The solid fatigue characteristic mapping data is imported into the hard disk real-time life prediction model to perform hard disk real-time life prediction and generate hard disk real-time life prediction data.

[0057] The present invention can accurately characterize the fatigue characteristics of the hard disk by extracting key characteristic indicators from the solid fatigue characteristic mapping data, making the prediction model more targeted and accurate, thereby improving the real-time prediction ability of the hard disk life. By dividing the data set to generate the model training set and the test set, the efficient allocation and full utilization of data resources are achieved, and the scientificity and reliability of the model training and testing are ensured. By processing the data using the long short-term memory neural network (LSTM) algorithm, the characteristics and trends in the time series can be fully captured, thereby improving the model's adaptability to the changes in the hard disk fatigue characteristics over time. The prediction pre-model is optimized and iterated through the model test set, and the prediction accuracy and robustness of the model are continuously improved to ensure the reliability of the final life prediction model in practical applications. The real-time hard disk life prediction model can quickly process the input data, generate real-time life prediction results, provide users with timely decision support, and help prevent data loss and system downtime caused by hard disk failure. Based on the real-time life prediction data, the hard disk usage plan can be optimized, and the hard disk that is about to fail can be replaced in time, thereby improving the system's operating efficiency and the utilization rate of hard disk resources. By predicting the hard drive life and taking maintenance measures in advance, you can reduce the hard drive failure rate, extend the overall life of the storage system, and reduce maintenance costs and resource waste.

[0058] In this specification, a solid state drive life prediction system based on big data is provided, which is used to execute the solid state drive life prediction method based on big data. The solid state drive life prediction system based on big data includes:

[0059] The power data acquisition module is used to obtain the solid-state hard disk structure data; perform operation simulation based on the solid-state hard disk structure data to generate the solid-state hard disk operation simulation data; use the solid-state hard disk operation simulation data to collect the power fluctuation data of the flash memory unit in the hard disk on the solid-state hard disk structure data to obtain the power fluctuation data of the flash memory unit in the key area of ​​the solid-state hard disk;

[0060] The electronic transition analysis module is used to classify the SSD operation based on the SSD operation simulation data, and generate write operation and erase operation; based on the write operation and erase operation, the SSD abnormal change trend analysis is performed on the power supply fluctuation data of the flash memory unit in the key area of ​​the SSD, and the SSD abnormal change trend data is generated; the SSD abnormal change trend data is used to identify the electronic migration effect of the SSD, and the SSD electronic transition identification data is generated;

[0061] The nanomaterial fatigue analysis module is used to obtain the time series data of the solid state drive; scan the internal storage medium of the solid state drive for the time series data of the solid state drive, and obtain the internal storage medium scan data set of the solid state drive; identify the fatigue effect of the nano-level material of the solid state drive for the internal storage medium scan data set of the solid state drive, and generate the hard disk fatigue effect identification data;

[0062] The real-time life prediction module is used to perform hard disk usage correlation analysis on the solid-state hard disk electronic transition identification data and the hard disk usage fatigue effect identification data to generate solid-state fatigue characteristic mapping data; perform hard disk real-time life prediction based on the solid-state fatigue characteristic mapping data to generate hard disk real-time life prediction data; adaptively write or warn the hard disk real-time life prediction data through a preset standard life warning threshold, thereby executing solid-state hard disk life extension optimization operations.

[0063] The beneficial effect of the present invention is that by collecting and running simulation of the solid-state hard disk structure data, the power supply fluctuation of the flash memory unit in the hard disk can be monitored in real time, especially the fluctuation of the key area. This makes it possible to more accurately capture the power supply problems that affect the performance and life of the hard disk, and provides accurate data support for subsequent analysis. By classifying the write and erase operations of the solid-state hard disk and analyzing the abnormal trend of the power supply fluctuation data of the flash memory unit, the electronic migration effect (electronic transition) that causes the hard disk performance degradation and failure can be identified. This identification can warn the hard disk failure in advance, avoid data loss or hard disk damage caused by electronic migration, thereby improving the reliability of the hard disk. The analysis of the timing data used by the solid-state hard disk can deeply identify the nano-level fatigue effect of the hard disk storage medium. By scanning the internal storage medium of the hard disk, potential fatigue damage can be identified, which provides an accurate basis for early fault warning and helps prevent the hard disk performance degradation or complete failure caused by material fatigue. By performing adaptive write optimization or early warning based on real-time life prediction, targeted extension measures can be taken before the hard disk life is about to expire. Through reasonable write strategies and optimization operations, the wear and aging of the hard disk can be effectively delayed, thereby improving the actual service life of the solid-state hard disk. Therefore, the present invention improves the accuracy of life prediction for solid-state drives through multi-dimensional data collection and analysis, combined with factors such as hard drive operation, power supply fluctuations, electron migration effects, and material fatigue. BRIEF DESCRIPTION OF THE DRAWINGS

[0064] Figure 1 A schematic diagram of the steps of a solid state drive life prediction method based on big data;

[0065] Figure 2 for Figure 1 Detailed implementation steps of step S2 in the flowchart;

[0066] Figure 3 for Figure 1 Detailed implementation steps of step S3 in FIG.

[0067] The realization of the purpose, functional features and advantages of the present invention will be further explained in conjunction with embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0068] The technical method of the present invention is described clearly and completely below in conjunction with the accompanying drawings. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by technicians in this field without creative work are within the scope of protection of the present invention.

[0069] In addition, the accompanying drawings are only schematic illustrations of the present invention and are not necessarily drawn to scale. The same reference numerals in the figures represent the same or similar parts, and their repeated description will be omitted. Some of the block diagrams shown in the accompanying drawings are functional entities and do not necessarily correspond to physically or logically independent entities. The functional entities can be implemented in software form, or implemented in one or more hardware modules or integrated circuits, or implemented in different networks and / or processor methods and / or microcontroller methods.

[0070] It should be understood that, although the terms "first", "second", etc. may be used herein to describe various units, these units should not be limited by these terms. These terms are used only to distinguish one unit from another unit. For example, without departing from the scope of the exemplary embodiments, the first unit may be referred to as the second unit, and similarly the second unit may be referred to as the first unit. The term "and / or" used herein includes any and all combinations of one or more of the listed associated items.

[0071] To achieve this, please refer to Figures 1 to 3 , a solid state drive life prediction method based on big data, the method comprising the following steps:

[0072] Step S1: acquiring the solid state drive structure data; performing operation simulation based on the solid state drive structure data, thereby generating the solid state drive operation simulation data; using the solid state drive operation simulation data to collect the power fluctuation data of the flash memory unit in the hard disk on the solid state drive structure data, and obtaining the power fluctuation data of the flash memory unit in the key area of ​​the solid state drive;

[0073] Step S2: classifying SSD operation on SSD operation simulation data to generate write operation and erase operation; performing SSD abnormal change trend analysis on power supply fluctuation data of flash memory units in key areas of SSD based on write operation and erase operation to generate SSD abnormal change trend data; using SSD abnormal change trend data to identify electron migration effect on SSD to generate SSD electron transition identification data;

[0074] Step S3: acquiring the time series data of the solid state drive; scanning the internal storage medium of the solid state drive on the time series data of the solid state drive to obtain a data set of scanning the internal storage medium of the solid state drive; identifying the fatigue effect of the nano-material of the solid state drive on the data set of scanning the internal storage medium of the solid state drive to generate data of identifying the fatigue effect of the hard drive;

[0075] Step S4: Perform hard disk usage correlation analysis on the solid-state hard disk electronic transition identification data and the hard disk usage fatigue effect identification data to generate solid-state fatigue characteristic mapping data; perform hard disk real-time life prediction based on the solid-state fatigue characteristic mapping data to generate hard disk real-time life prediction data; adaptively write or warn the hard disk real-time life prediction data through a preset standard life warning threshold, thereby executing the solid-state hard disk life extension optimization operation.

[0076] The present invention can deeply understand the performance of each component inside the hard disk by obtaining the structural data of the SSD and performing operation simulation based on these data. This step is not only based on surface data, but also simulates the actual operation of the SSD through comprehensive simulation, thereby obtaining more accurate operation data. By collecting the power fluctuation data of the flash memory unit inside the hard disk, the electrical changes of the hard disk in actual use can be captured, which is very important for accurately identifying the early signs of hard disk performance degradation, especially for those problems that cannot be directly reflected by traditional SMART data. By distinguishing between write operations and erase operations, the impact of each operation on the hard disk can be finely analyzed. The different modes and frequencies of write and erase operations will directly affect the service life of the hard disk. This distinction can better understand its specific impact on the life of the SSD. Based on the operation data, the trend analysis of power fluctuations can capture the abnormal changes of the hard disk in real time. This makes the life prediction more accurate and can provide early warning before the hard disk has potential failures. Electron migration effects (such as electron transitions) are one of the key factors in the decline of hard disk life. The patent can effectively identify potential loss problems in advance by analyzing these microscopic electrical effects. By obtaining the hard disk usage timing data, the use history of the hard disk can be tracked and the life analysis can be performed over time. This time-series-based data analysis helps to capture the changing characteristics of the hard disk at different stages of use and improve the timeliness of the prediction. By scanning the internal storage medium of the hard disk, the nano-level material fatigue effect of the hard disk can be identified. This identification can not only find microscopic damage that cannot be detected by traditional methods, but also conduct a comprehensive analysis of factors such as the material properties and working environment of the hard disk. By combining the electronic transition identification data and the hard disk use fatigue effect identification data, correlation analysis can be performed to analyze the health status of the SSD at multiple levels. This comprehensive analysis method overcomes the defect that the traditional life prediction method only relies on a single indicator, and improves the accuracy and reliability of the prediction results. Based on the above analysis results, real-time life prediction can be performed. This can not only provide early warning for hard disk users, but also optimize the use strategy of the hard disk, such as automatically adjusting the write strategy or timely data backup to avoid data loss. Therefore, the present invention improves the accuracy of life prediction for solid-state hard disks through multi-dimensional data acquisition and analysis, combined with factors such as hard disk operation, power supply fluctuations, electronic migration effects and material fatigue.

[0077] In the embodiment of the present invention, reference Figure 1 FIG. 1 is a schematic diagram of a process flow of a solid state drive life prediction method based on big data according to the present invention. In this example, the solid state drive life prediction method based on big data includes the following steps:

[0078] Step S1: acquiring the solid state drive structure data; performing operation simulation based on the solid state drive structure data, thereby generating the solid state drive operation simulation data; using the solid state drive operation simulation data to collect the power fluctuation data of the flash memory unit in the hard disk on the solid state drive structure data, and obtaining the power fluctuation data of the flash memory unit in the key area of ​​the solid state drive;

[0079] In an embodiment of the present invention, physical modeling is performed through hardware documents, technical specifications provided by the manufacturer, or disassembly of the solid-state hard disk. Perform detailed structural analysis on various parts of the solid-state hard disk (such as flash memory chip, controller, power module, data bus, cache area, etc.). Obtain information such as the circuit diagram, chip layout, storage unit type (such as NAND flash memory, SLC / TLC / QLC) and data transmission path of the solid-state hard disk. Extract and establish a functional model of the hard disk controller, including its data processing, error correction algorithm, cache management, and its interaction with the flash memory unit. Use hardware design tools (such as CAD, AltiumDesigner) to model the circuit of the controller and extract key signals and interface information. The acquired solid-state hard disk structure data is usually saved in multiple formats, such as CAD files, BOM (bill of materials), circuit diagrams, etc. It can be converted into a digital model suitable for simulation software processing. Select a suitable simulation platform, such as COMSOL Multiphysics, ANSYS, Simulink, or a dedicated solid-state hard disk simulation tool. Input the simulation model according to the hard disk structure data. Set the working environment of the solid-state drive, including: input voltage, current; data operation load, processor activity; temperature changes, heat dissipation conditions; flash memory operation mode (read, write, erase). Use the power simulation module to simulate the power consumption of each part of the hard disk (especially the controller and flash memory unit) to evaluate power demand and power fluctuations. Simulate the I / O performance of the hard disk and simulate the speed, delay, reliability, etc. of data flow transmission in the hard disk. Generate the operation simulation data of the solid-state drive, including the performance data, power consumption data, temperature change and other information of the hard disk under different loads. These data can be presented in the form of charts, heat maps, etc. Define and locate key acquisition points in the simulation model, mainly focusing on the power input end of the flash memory unit. Focus on monitoring areas that are sensitive to power fluctuations, such as the power management circuit between the power module and the flash memory, and the controller output voltage. According to the hard disk simulation model, real-time acquisition of power supply voltage and current fluctuation data, especially the changes in the power fluctuation of the flash memory unit during operations such as reading, writing and erasing. Analyze the amplitude, frequency characteristics, noise interference and other information of power supply fluctuations through spectrum analysis, time domain analysis and other methods. Use the acquisition module of the simulation tool (such as ANSYS's signal analysis tool and Simulink's signal processing tool) to collect power fluctuation data with high precision. Filter the data to remove noise and interference to ensure that the collected data reflects the actual power fluctuation situation. The obtained flash memory unit power fluctuation data includes voltage and current fluctuation curves, spectrum diagrams, etc. The output data can be used to further analyze the stability, life and failure mode of the flash memory unit. Integrate the power fluctuation data collected under different simulation conditions to build a comprehensive fluctuation data set.Generate complete power fluctuation data for key areas (such as flash memory chips, controller interfaces, etc.), involving different fluctuation characteristics of multiple operating modes. Use statistical analysis methods (such as mean, variance, standard deviation, etc.) to summarize the fluctuation data and extract key features. Compare the power fluctuation differences under different workloads to evaluate the stability of the power supply. Compare the simulation data with the actual hardware test data to ensure the accuracy of the simulation results. Generate a complete power fluctuation data report for flash memory cells in key areas of the SSD. The data format can be Excel, CSV, MATLAB data files, etc., including detailed information and analysis results of the power fluctuation.

[0080] Step S2: classifying SSD operation on SSD operation simulation data to generate write operation and erase operation; performing SSD abnormal change trend analysis on power supply fluctuation data of flash memory units in key areas of SSD based on write operation and erase operation to generate SSD abnormal change trend data; using SSD abnormal change trend data to identify electron migration effect on SSD to generate SSD electron transition identification data;

[0081] In an embodiment of the present invention, different types of operations are identified by extracting operation logs from simulation data of a solid-state drive. Simulation data generally includes operations such as read, write, erase, and refresh. Operation data related to write and erase are extracted from a log file or a database using techniques such as regular expressions or timestamp matching. Write Operation: Identify all operations involving data writing to a flash memory cell, and record information such as the amount of data, operation time, and target flash memory cell of the operation. Erase Operation: Identify all records involving flash memory cell erase operations, and identify parameters such as erased data blocks and erase duration. Mark each operation with parameters such as timestamp, operation type (write or erase), affected flash memory area, and data transfer rate. Classify these operations and store them as separate data sets for subsequent analysis. The generated output is a time series data set of write operations and erase operations, including detailed information of each operation (such as operation type, timestamp, target flash memory cell, etc.). The timestamps of write operations and erase operations are synchronized with the power supply fluctuation data of the corresponding area. Specifically, the real-time data of power supply voltage and current are recorded during each write or erase operation to ensure that the data is synchronized. The time domain and frequency domain characteristics of power supply fluctuations when the operation occurs are obtained through time series analysis methods (such as sliding window method, Fourier transform, etc.). The changes in power supply fluctuations under normal operation and after write / erase operations are compared to identify abnormal change trends. For example, the erase operation will cause instantaneous changes in the power supply voltage, while the write operation will cause current fluctuations. Statistical methods (such as standard deviation, coefficient of variation) and machine learning methods (such as anomaly detection algorithms) are used to detect anomalies in power supply fluctuation data and extract data points with abnormal fluctuations. According to the change trend of power supply fluctuations after write and erase operations, a power supply fluctuation model is established to evaluate the fluctuation amplitude, fluctuation period, amplitude change rate, etc. Time series prediction models (such as ARIMA, LSTM neural network) are used to predict future power supply fluctuations and identify whether there is a potential abnormal change trend. Generate abnormal change trend data for solid-state drives, including trend analysis of the impact of each operation on power supply fluctuations. The data output format can include charts, trend curves, timestamps of abnormal fluctuation events, etc. Electromigration effect refers to the material migration phenomenon caused by the flow of electrons in flash memory cells under long-term high current load. Excessive electron migration can cause flash memory cells to be damaged or fail. Determining the electron migration effect is usually achieved by analyzing abnormal changes in parameters such as current density, temperature changes, and power supply fluctuations. Use the power supply fluctuation information in the simulation data to calculate the current density and temperature increase. Excessive current density or high temperature is a potential indication of electron migration. Combine the power supply fluctuation data with the temperature data in the SSD working environment to analyze the impact of temperature changes on current conduction and electron migration.Compare normal and abnormal power fluctuation data, use machine learning algorithms (such as support vector machine SVM, decision tree) to analyze and identify abnormal current fluctuations and temperature fluctuations, and determine whether there is electron migration. You can determine whether the electron migration effect occurs by calculating the power density (Power Density) and current density (Current Density) of the flash memory unit and comparing them with the set threshold. Combine multiple dimensions such as power fluctuation data, temperature data, and current data to construct a multidimensional feature space for identifying electron migration effects. Use data clustering or anomaly detection methods, such as K-means clustering or DBSCAN, to automatically identify abnormal patterns related to electron migration. Generate solid-state drive electron transition identification data, including relevant indicators of the occurrence of electron migration effects (such as current density, temperature, power density, etc.), timestamps and areas where electron migration potentially occurs, and identified abnormal pattern data.

[0082] Step S3: acquiring the time series data of the solid state drive; scanning the internal storage medium of the solid state drive on the time series data of the solid state drive to obtain a data set of scanning the internal storage medium of the solid state drive; identifying the fatigue effect of the nano-material of the solid state drive on the data set of scanning the internal storage medium of the solid state drive to generate data of identifying the fatigue effect of the hard drive;

[0083] In an embodiment of the present invention, the real-time performance data of the hard disk is obtained by using a hard disk health monitoring tool (such as a SMART monitoring system). These data generally include the number of hard disk operations, bad track monitoring, usage time, temperature, power consumption, etc. The hard disk monitoring software is configured to regularly collect time series data, and the records include: the number of writes and reads; temperature and humidity data; real-time changes in current and voltage; the health status of each storage block, erase and write cycles, etc. The collected data can be stored in CSV, JSON format, or stored using a time series database (such as InfluxDB) for subsequent analysis. To ensure high-frequency data collection, it is usually necessary to ensure that updated data is obtained within a time interval of one second or less. Generate a solid-state hard disk usage time series data set, including timestamps, operation operations, temperature, voltage, power consumption, and other information, which is suitable for subsequent analysis. Evaluate the health status of each flash memory block based on the write / read operation frequency in the time series data and the usage of each storage unit. Use indicators such as the number of erases, bad tracks, and reallocated block counts in the SMART data to evaluate the fatigue of the storage medium inside the hard disk. Use the "bad block scan" function of the hard disk to check whether there are bad blocks or areas of the storage medium that can no longer be written. Perform a multi-dimensional scan of the storage blocks to detect whether the storage media has signs of degradation after long-term use. For example, the number of writes and erases of each storage block can be used to evaluate the degree of damage to each area. Combined with information such as temperature and power in the time series data, identify whether there is material damage caused by thermal failure or current overload. Monitor the write amplification effect of the hard disk to analyze whether the frequently written areas are excessively worn. Generate a data set of internal storage media scans of the solid-state drive, covering detailed information such as the health status, damage level, bad block distribution, and erase and write times of each storage unit of the hard disk. The flash memory unit in the hard disk uses NAND flash memory technology. The materials of the flash memory unit (such as silicon oxide or high dielectric material) will produce fatigue effects after experiencing multiple erase and write cycles. When analyzing it, pay attention to the following aspects: As the erase and write cycles increase, the material will gradually age, and the charge storage capacity of the storage unit will decrease, resulting in an increase in the write error rate. The long-term high current and high voltage will cause material damage, which will then produce electromigration effects or thermal failure phenomena. Temperature changes accelerate physical damage to materials, so it is necessary to analyze the relationship between temperature changes and flash memory cell degradation. Use data mining algorithms (such as cluster analysis, decision trees, support vector machines (SVMs), etc.) to analyze storage media scan data and identify different fatigue modes. For example, storage cells at different fatigue stages can be divided according to the number of writes and erases to the storage cell and the health status of the storage block. Compare the write and erase operations with the damage to the storage cell, build a prediction model for fatigue effects, and predict future failures.Extract features related to fatigue effects, such as the number of erases, voltage fluctuations, temperature fluctuations, etc., and construct a feature space for fatigue effect identification. Use supervised learning methods (such as random forests, neural networks) to train models to automatically identify and predict the fatigue effects of storage media. Generate data for identifying fatigue effects on solid-state drives, including the fatigue status of storage units, material fatigue in different areas, and prediction data for future failures. The data can be in the form of classification labels (such as "healthy", "mild fatigue", "severe fatigue"), numerical features (such as the number of writes, current density, temperature rise, etc.), and prediction results.

[0084] Step S4: Perform hard disk usage correlation analysis on the solid-state hard disk electronic transition identification data and the hard disk usage fatigue effect identification data to generate solid-state fatigue characteristic mapping data; perform hard disk real-time life prediction based on the solid-state fatigue characteristic mapping data to generate hard disk real-time life prediction data; adaptively write or warn the hard disk real-time life prediction data through a preset standard life warning threshold, thereby executing the solid-state hard disk life extension optimization operation.

[0085] In an embodiment of the present invention, the electronic transition identification data includes characteristic information such as current density, temperature, and power density related to the electronic migration effect. The hard disk fatigue effect identification data includes the number of erases and writes of the storage unit, the write amplification effect, the degree of damage, etc. Ensure that the two data sets have the same timestamp or can be aligned according to the time series to ensure the time series association between the data. Extract the key features from the electronic transition data and the fatigue effect identification data, such as the severity of the electronic migration effect, the damage rate of the storage medium, and the temperature fluctuation. Standardize these features to eliminate the influence of the dimension so that different data sets can be effectively combined and analyzed. Use statistical analysis methods (such as Pearson correlation coefficient and mutual information) to evaluate the correlation between the electronic transition effect and the hard disk fatigue effect. Use multiple regression analysis or machine learning methods (such as random forests, support vector machines SVM) to build a model to explore the complex nonlinear relationship between the electronic migration effect and the hard disk fatigue effect. According to the analysis results, generate fatigue characteristic mapping data of the solid-state hard disk, match the fatigue degree of each hard disk with the severity of the electronic transition effect, and form a mapping diagram. Generate solid fatigue characteristic mapping data, including the fatigue degree of different storage units of the hard disk, the strength of the electronic transition effect, and the correlation mapping between the two. Based on the solid fatigue characteristic mapping data, build a hard disk life prediction model. Machine learning algorithms, such as gradient boosting tree (GBDT), long short-term memory neural network (LSTM) and other time series prediction models, can be used to predict the remaining life of the hard disk in combination with the hard disk usage data, fatigue characteristic data and electronic transition identification data. Feature input includes: fatigue degree of each storage unit; degree of electronic transition effect; number of write / erase operations; environmental factors such as temperature change and power fluctuation. Ensure that new electronic transition identification data and fatigue effect data are collected in real time during the use of the hard disk, and use real-time data to update the life prediction model. During the entire use cycle of the hard disk, by regularly acquiring data, continuously perform life prediction and calibrate the prediction model to reflect the current health status of the hard disk. Calculate the remaining life of the hard disk through the prediction model, and output the actual remaining working time of the hard disk (in hours, days, etc.). Output the current health percentage of the hard disk, for example, "health is 85%", which means that the remaining life of the hard disk is 85%. Generate real-time hard disk life prediction data, including the predicted remaining life, health status, and performance degradation trends during future use. Set standard life warning thresholds, such as triggering a warning when the remaining life of the hard disk is less than 30%, and performing life extension optimization operations when the remaining life is less than 10%. Set different warning thresholds based on the hard disk's working environment, usage scenarios, and manufacturer's standards. For example, set a lower threshold for enterprise-level hard disks and a higher threshold for consumer-level hard disks. When the hard disk life approaches the threshold, execute an adaptive write optimization strategy. The optimization strategy includes: migrating frequently written storage units to less used areas to reduce write pressure in thermally damaged areas.Reduce unnecessary writes by compressing or merging data operations. Adjust the temperature through the hard disk's temperature control system to avoid excessive temperature accelerating material fatigue. When the hard disk life approaches the preset threshold, the hard disk health warning notification is triggered to prompt the user or management system to perform data backup or hardware replacement in advance. Through the integrated alarm system, reminders are sent to system administrators, users or cloud services when the remaining life of the hard disk is short. Generate execution records of hard disk life extension optimization operations, including optimization operation type (such as data migration, write optimization, etc.), operation time, result evaluation, etc. In addition, life warning notification data is also generated, including health warnings and warning times.

[0086] Preferably, step S1 comprises the following steps:

[0087] Step S11: Acquire the solid state hard disk structure data;

[0088] Step S12: performing physical topology modeling on the SSD structure data to generate SSD topology structure modeling data; performing operation simulation on the SSD based on the SSD topology structure modeling data to generate SSD operation simulation data;

[0089] Step S13: marking the key area of ​​the SSD structure data using the SSD running simulation data to generate the SSD key area marking data;

[0090] Step S14: Based on the integrated current and voltage sensors, the power supply fluctuation data of the flash memory unit in the SSD is collected for the key area mark data of the SSD to obtain the power supply fluctuation data of the flash memory unit in the key area of ​​the SSD.

[0091] In an embodiment of the present invention, by using professional hardware detection tools, such tools can directly read the firmware information of the solid-state drive, and parse out basic parameters such as chip model and firmware version number. At the same time, the official technical documents of the solid-state drive of this model are consulted to obtain more detailed physical structure details such as internal circuit layout diagram and pin definition of each chip, so as to obtain the solid-state drive structure data in an all-round way. Use professional electronic design automation (EDA) software to enter the collected solid-state drive structure data. According to the actual physical connection relationship of each component, the topological structure such as bus connection and power line direction between the main control and the flash memory chip is drawn in the software to generate the solid-state drive topological structure modeling data. For example, if the solid-state drive adopts the NVMe protocol, it is accurately modeled according to the interface connection method under the NVMe standard specification to ensure the accuracy of the topological structure. Based on the generated topological structure modeling data, the corresponding working parameters are set in the simulation software to simulate the operating state of the solid-state drive under different operating load scenarios. For example, gradually increase the number of operation requests from hundreds of requests per second to thousands of requests, observe the changes in key indicators such as bandwidth utilization and signal delay of the data transmission link, and then generate solid-state drive operation simulation data. These data accurately reflect the performance of the SSD under various working conditions. Analyze the SSD operation simulation data, focusing on those areas where obvious performance bottlenecks or abnormal signal fluctuations occur during high-load operation. For example, when the simulation finds that the data transmission delay of a group of flash memory chips is far higher than the average level, or the voltage drop of a certain power line is too large, the corresponding flash memory chip group, power line and its surrounding related circuits are marked in the SSD structure data as key areas, thereby generating SSD key area marking data. The marking process can be implemented with the help of automated scripts, and the performance thresholds and abnormal judgment rules are written into the script to quickly locate the key areas. Miniaturized and highly sensitive current and voltage sensors are precisely installed near the key areas of the SSD. These sensors establish communication connections with the main control chip of the SSD through customized interface circuits to transmit the collected data in real time. In different working modes such as normal operation and standby of the SSD, the sensors continuously monitor the power supply of the flash memory units in the key areas, accurately capture the current and voltage fluctuations at the microsecond level, and record these fluctuation data one by one, and finally obtain the power fluctuation data of the flash memory units in the key areas of the SSD.

[0092] Preferably, running simulation on the solid state drive based on the solid state drive topology structure modeling data includes:

[0093] Setting operation simulation rules for the solid state drive based on the solid state drive topology structure modeling data to obtain the solid state drive operation simulation rules, wherein the solid state drive operation simulation rules include operation sequence, data block size, and operation frequency;

[0094] According to the preset operation sequence, data block size and operation frequency, the operation parameters of the solid state drive are configured to generate operation simulation rule configuration data;

[0095] The solid-state drive is simulated in a virtual environment using the simulation rule configuration data to generate a solid-state drive simulation data set; the solid-state drive simulation data set is preprocessed to generate solid-state drive simulation data, wherein the data preprocessing includes data cleaning, data denoising, data missing value filling and data standardization.

[0096] In the embodiment of the present invention, there are two modes of sequential operation and random operation through the common operation sequence. In the sequential operation mode, for example, it is set to start from the starting logical address of the solid-state hard disk and read or write data blocks in sequence, which simulates continuous data transmission scenarios such as video playback and large file copying, and the operation sequence is simple and linear. In the random operation scenario, the random number generation algorithm can be used to set the logical address of each operation operation to be randomly distributed in the entire storage area, simulating the working conditions of frequent database queries and multi-task small file operations. Assume that the proportion of sequential operation scenarios is set to 40% and the proportion of random operation scenarios is set to 60%, so as to be close to the actual complex usage. According to common file systems and application data processing units, multiple data block sizes are selected. For example, the small data block is set to 4KB, which is compatible with the default cluster size of the Windows NTFS file system and is often used for ordinary document operation simulation; the medium data block is set to 64KB, which is suitable for simulating database page operation; the large data block is set to 1MB, corresponding to the large file transfer scenario. In the operation simulation rules, the operation operations of these data block sizes are allocated in different proportions, for example, the small data block operation accounts for 30%, the medium data block accounts for 40%, and the large data block accounts for 30%. Set the operating frequency according to the different performance levels and target application scenarios of the SSD. For consumer-grade entry-level SSDs, the operating frequency can be set to 200-500 operating operations per second to simulate daily office use; for high-performance enterprise-level SSDs, the operating frequency is increased to 1000-2000 times per second to simulate the high-concurrency operation scenario of the server. In summary, when setting the operating frequency, start with low-frequency operation, and then increase it in steps to observe the response of the SSD at different stages. Use professional storage performance testing software, such as CrystalDiskMark, to enter the above-set operating sequence, data block size, and operating frequency parameters into the corresponding configuration fields of the software in sequence. For example, in CrystalDiskMark, fill in the corresponding proportional values ​​under the "Sequential Run" and "Random Run" options, check the selected 4KB, 64KB, and 1MB options in the "Data Block Size" drop-down menu, and enter the set range of operations per second in the "Operation Frequency" input box. After completing the parameter configuration, the software automatically integrates and generates the operation simulation rule configuration data. Open the virtual environment simulation module built into the test software, load the previously generated operation simulation rule configuration data, and let the software simulate a closed SSD operation test environment. In this virtual environment, the software continuously initiates operation requests according to the set rules, reads or writes data from the virtual storage space of the SSD, and records the response time, data throughput, error rate and other indicators of each operation in real time, accumulates massive data, and finally generates a SSD operation simulation data set containing various operation performance parameters. The scale of the data set reaches thousands or even tens of thousands of operation records.Traverse the SSD operation simulation data set and remove data records with obvious errors, such as negative operation response time and data throughput far exceeding the physical limit. Use conditional judgment statements to set a reasonable value range, and delete records that do not meet the range directly to ensure the basic rationality of the data. In the SSD operation simulation data set, due to slight interference in the test environment, some data will have small fluctuation noise. Use filtering algorithms, such as moving average filtering, set the filter window size to 5, calculate the average value of the data in the window, and use the average value to replace the window center data point to smooth out high-frequency noise interference. Check whether there are missing values ​​in the data set. If so, for numerical missing data, use the mean filling method to calculate the mean of the column data and replace the missing values ​​with the mean; for categorical missing data, fill according to the mode of the column data to ensure data integrity. Use the Z-score standardization method to process the data of each indicator in the data set, unify data of different magnitudes into a similar range, avoid the impact of data magnitude differences on subsequent analysis, and finally generate high-quality SSD operation simulation data.

[0097] As an example of the present invention, refer to Figure 2 As shown, in this example, step S2 includes:

[0098] Step S21: classifying SSD operation on SSD operation simulation data to generate SSD operation classification data, wherein the SSD operation classification data includes write operation and erase operation; performing write current fluctuation screening on power fluctuation data of flash memory units in key areas of the SSD based on the write operation to obtain write operation current fluctuation data;

[0099] Step S22: performing an erase voltage fluctuation screening on the power fluctuation data of the flash memory unit in the key area of ​​the solid state hard disk based on the erase operation to obtain the erase operation voltage fluctuation data; performing a fluctuation amplitude change analysis on the write operation current fluctuation data and the erase operation voltage fluctuation data to generate the operation amplitude change data and the operation current peak change data;

[0100] Step S23: performing abnormal fluctuation analysis on the operation amplitude change data and the operation current peak change data to perform abnormal change trend analysis on the solid state hard disk, and generating abnormal change trend data on the solid state hard disk;

[0101] Step S24: using the abnormal change trend data of the solid state hard disk to identify the electron migration effect of the solid state hard disk, and generating the electron transition identification data of the solid state hard disk.

[0102] In an embodiment of the present invention, by carefully reviewing the SSD operation simulation data, the operations in the data are distinguished according to the operation instruction code and the related operation identification field. For example, when a specific write instruction code "0x01" appears in the data record, the corresponding data is classified as a write operation; if an erase instruction code "0x02" is encountered, the record is marked as an erase operation, thereby arranging and generating SSD operation classification data. Focus on the power fluctuation data of the flash memory unit in the key area of ​​the SSD corresponding to the write operation. Since the current fluctuation characteristics are more obvious during the write operation, a screening threshold for the current fluctuation amplitude is set, such as setting the threshold to ±50mA. By traversing the data, only those data records whose current fluctuation amplitude exceeds the threshold are retained. These records accurately reflect the abnormal current fluctuation during the write process, and then the write operation current fluctuation data is obtained. For the erase operation, the voltage part in the power fluctuation data of the flash memory unit in the key area of ​​the SSD is checked. Considering that the erase operation has a high requirement for voltage stability, a suitable voltage fluctuation screening interval is set, such as a fluctuation of 2V. Compare the data one by one, filter out the voltage data outside this fluctuation interval, and summarize them to obtain the erase operation voltage fluctuation data. Compare and analyze the write operation current fluctuation data and the erase operation voltage fluctuation data together. For the write current fluctuation, calculate the difference between the current fluctuation amplitudes at adjacent time points, and count the change trends of these differences over time; similarly, for the erase voltage fluctuation, calculate the change in the voltage fluctuation amplitude in each period. Through these calculations, generate the operation amplitude change data, and extract the current peak value in each fluctuation, observe its increase and decrease, and form the operation current peak change data. Set the abnormal judgment criteria in the operation amplitude change data and the operation current peak change data. If the change in a certain period of the operation amplitude change data exceeds 3 times the standard deviation of the preset mean, or the peak change rate suddenly jumps by more than 50% in the operation current peak change data, it is determined that an abnormal fluctuation occurs. Taking the identified abnormal fluctuation as the starting point, trace back and deduce backward along the time axis to observe whether the same type of abnormal fluctuations are gradually increasing or decreasing, count the frequency changes of abnormal fluctuations, integrate this information to generate the abnormal change trend data of the solid-state hard disk, and intuitively show the development trend of potential problems of the solid-state hard disk. The electron migration effect usually causes regular abnormal fluctuations in current and voltage, and this fluctuation will become more and more significant over time. Use the abnormal change trend data of the solid-state drive to check the long-term and high-frequency abnormal fluctuation characteristics. When the data shows that the current fluctuation continues to increase and the voltage stability continues to deteriorate, and it conforms to the theoretical fluctuation model of the problem caused by electron migration, mark the corresponding area and generate solid-state drive electron transition identification data that can accurately locate the area affected by electron migration, providing a key basis for subsequent targeted repair or optimization.

[0103] Preferably, step S24 includes the following steps:

[0104] Step S241: defining the current-voltage signal of the abnormal change trend data of the solid state hard disk based on a preset time window to generate a current signal and a voltage signal; calculating the instantaneous value of the current signal and the voltage signal to obtain the instantaneous value of the current and the instantaneous value of the voltage;

[0105] Step S242: constructing a phase space trajectory through the instantaneous current value and the instantaneous voltage value to generate a current-voltage phase diagram of the solid state drive; performing trajectory overlap detection on the current-voltage phase diagram of the solid state drive to generate a phase trajectory overlap curve of the solid state drive;

[0106] Step S243: identifying the current-voltage mutation point of the solid-state hard disk current-voltage phase diagram according to the solid-state hard disk phase trajectory overlap curve, thereby obtaining the current-voltage mutation point;

[0107] Step S244: Use the current-voltage mutation point to mark the hard disk electronic transition mutation signal on the current signal and the voltage signal to obtain the hard disk electronic transition mutation marking signal; calculate the proportion of the hard disk electronic transition mutation marking signal to the current signal and the voltage signal to generate the solid state hard disk electronic transition identification data.

[0108] In the embodiment of the present invention, by determining a suitable time window size, for example, 10 milliseconds is selected as a time window. The selection of this time window needs to comprehensively consider the response time of the internal circuit of the solid-state drive and the data acquisition frequency, so as to capture sufficient signal details without introducing too much redundant data. Based on the abnormal change trend data of the solid-state drive, in each time window, the data reflecting the current change is extracted and integrated to form a current signal; similarly, the corresponding voltage change data is extracted to form a voltage signal.

[0109] Assuming that the abnormal change trend data of the solid-state drive is stored in the form of a time series, the data index and slice operation are used to accurately intercept the required data segments according to the predetermined time window start and end time to generate current and voltage signals. For discrete current signal and voltage signal data points, numerical calculation methods are used to calculate the instantaneous value. A common method is to use the central difference formula, with the instantaneous value of current as the horizontal axis and the instantaneous value of voltage as the vertical axis, to combine the instantaneous values ​​of current and voltage corresponding to each time point into coordinate points. By connecting these coordinate points in sequence, a phase space trajectory is constructed on a two-dimensional plane to form a current-voltage phase diagram of the solid-state drive. For example, if the instantaneous value of current is i1 and the instantaneous value of voltage is v1 at a certain moment, a point (i1, v1) is marked at the coordinates. As time goes on, many such points outline the trajectory curve. Image analysis algorithms are used, such as contour detection-based methods. First, the trajectory in the current-voltage phase diagram of the solid-state drive is contour extracted, and the trajectory contours generated in different time periods are compared. When it is found that some contours are highly similar in spatial position and shape, these overlapping areas are marked, and the boundary lines of the overlapping areas are extracted, and finally the SSD phase trajectory overlap curve is generated. When the SSD phase trajectory overlap curve is viewed, when the trajectory curve has sudden bends, jumps and other discontinuities, the corresponding is the mutation point. Combined with the coordinate values ​​of the phase diagram, the current and voltage values ​​at these bends and jumps are accurately located to determine the current-voltage mutation point. At the same time, a slope threshold can be set. For example, when the slope of a certain section of the trajectory curve changes by more than 5, it can be used to assist in determining whether it is a mutation point. In the current signal and voltage signal data sequence, the time index position corresponding to the current-voltage mutation point is found, and specific marking codes are added to these positions, such as "1" for mutation points and "0" for normal points, so as to complete the hard disk electronic transition mutation signal marking and generate the hard disk electronic transition mutation marking signal. The number of "1" marked in the hard disk electronic transition mutation marking signal is counted, and divided by the total number of data points of the current signal and voltage signal respectively, to obtain the proportion of mutation signals in the current signal and voltage signal. These percentage data are collated together with key information such as the specific values ​​of current and voltage corresponding to the mutation point, the time of occurrence, etc., to generate solid-state hard drive electron transition identification data that can clearly reflect the electron transition situation.

[0110] As an example of the present invention, refer to Figure 3 As shown, in this example, step S3 includes:

[0111] Step S31: obtaining the solid state hard disk usage timing data;

[0112] Step S32: performing time segmentation on the SSD usage time series data to generate the SSD usage time series segmentation data; using the SSD usage time series segmentation data as a scanning time interval, and scanning the internal storage medium of the SSD using a scanning probe microscope technology according to the scanning time interval to obtain a SSD internal storage medium scanning data set;

[0113] Step S33: performing hard disk nano-material use fatigue effect identification on the internal storage medium scanning data set of the solid state hard disk to generate hard disk use fatigue effect identification data.

[0114] In the embodiment of the present invention, a communication connection is established with the main control of the solid-state drive through a professional data reading tool, and detailed data such as the operation timestamp, the time when the erase operation occurs, and the start and end time of the idle period are read according to the established communication protocol. In addition, the management software supporting some solid-state drives can also export the device operation log, and extract and integrate the solid-state drive usage time series data covering the complete operation process from the first use of the solid-state drive to the current moment from these log files. According to the working characteristics and typical application scenarios of the solid-state drive, reasonable segmentation rules are formulated. For example, considering the difference between intensive operation applications and intermittent use scenarios, the continuous high-intensity operation period is divided into a section, and the duration is set to range from a few minutes to tens of minutes; the long idle period is segmented separately, and the idle time exceeds half an hour and forms a section; the daily scattered operation period is classified into a section according to the frequency of operation, and every few hundred operations are classified into a section. Use an automated script to traverse the solid-state drive usage time series data according to the set rules, cut and mark each period, and generate the solid-state drive usage time series segmentation data. The duration of each period in the generated SSD time series segmented data is directly converted into the scanning time interval of the scanning probe microscope (SPM). The parameters of the SPM device are adjusted to ensure that its scanning resolution can meet the needs of nano-level observation. Generally, the lateral resolution is set to be better than 1 nanometer and the vertical resolution is set to 0.1 nanometer. In each scanning time interval, the SPM probe is accurately positioned on the surface of the internal storage medium of the SSD, and the atomic force, friction force, electrical characteristics and other microscopic information of the surface of the medium are scanned point by point. As the time interval advances, a large amount of scanning data is accumulated, and finally the internal storage medium scanning data set of the SSD is summarized. In-depth research on the basic material properties of the internal storage medium of the SSD, such as the NAND flash memory unit in the flash memory chip, to understand its initial atomic arrangement, electrical performance and other key parameters. Compare the data obtained at different time periods in the scanning data set to observe the changes of the material's atomic spacing, surface roughness, and electrical conductivity over time. For example, if it is found that the atomic spacing gradually increases and the surface becomes rough, it is likely that the material is under repeated stress and shows signs of fatigue. Combining materials theory with SSD industry experience, fatigue thresholds are set for different material performance indicators. For example, for electrical conductivity, if the decrease exceeds 10% of the initial value, and the atomic spacing increases by more than 0.5 nanometers relative to the initial value, fatigue effect is determined to occur. The data set is scanned by traversing the internal storage medium of the solid-state drive, and the data points and time periods that meet the fatigue characteristics are selected according to the set threshold, and the hard disk fatigue effect identification data is compiled to accurately reflect the fatigue of nano-level materials in various areas of the hard disk.

[0115] Preferably, step S33 includes the following steps:

[0116] Step S331: performing abnormal point detection on the data set of the scanned storage medium inside the solid state drive to generate a data set of characteristics without abnormality; performing nanoscale strain distribution analysis on the data set of characteristics without abnormality to generate strain distribution characteristic data of the storage medium;

[0117] Step S332: performing crystal structure degradation detection on the internal storage medium scan data set of the solid state drive according to the storage medium strain distribution characteristic data to generate crystal degradation characteristic data, wherein the crystal structure degradation detection includes dislocation density detection and grain boundary migration detection;

[0118] Step S333: Perform charge transfer characteristic analysis on the solid state drive electronic transition identification data to generate solid state drive charge transfer characteristic data; perform fatigue degradation index calculation on the solid state drive internal storage medium scan data set using the crystal degradation characteristic data and the solid state drive charge transfer characteristic data to obtain hard drive use fatigue effect identification data, wherein the fatigue degradation index calculation formula is as follows:

[0119] ;

[0120] ;

[0121] ;

[0122] ;

[0123] In the formula, Expressed as fatigue degradation index, Expressed as the crystal degradation coefficient, Expressed as the charge transfer accumulation characteristic coefficient, Expressed as the environmental coupling factor coefficient, Expressed as a characteristic fluctuation intensity value, Expressed as the crystal degradation characteristic weight coefficient, Expressed as the charge transfer characteristic weight coefficient, Expressed as the weight coefficient of the environmental coupling factor, Expressed as the characteristic fluctuation intensity correction coefficient, Expressed as The local lattice distortion value of a crystal, is the initial value of the overall lattice parameter of the crystal, Expressed as the number of crystals, Expressed as time The charge distribution change function inside is Represents the time limit, Expressed as the standard deviation of the crystal degradation characteristics, Expressed as the standard deviation of the charge transfer accumulation characteristic.

[0124] In an embodiment of the present invention, an abnormal point is identified by using a statistical analysis method. First, the mean and standard deviation of each physical quantity (such as atomic force, electrical signal strength, etc.) in the scanned data set are calculated. A threshold is set, usually the mean plus 3 times the standard deviation is taken as the upper limit, and the mean minus 3 times the standard deviation is taken as the lower limit. Traverse each data point in the data set, and if the value of a point exceeds this threshold range, it is determined to be an abnormal point and removed. After such screening, what remains is a data set without abnormal characteristics. For example, for data reflecting the interatomic force, if its average value is 10nN and the standard deviation is 2nN, then data points greater than 16nN or less than 4nN will be removed. The strain distribution is determined by analyzing the change of atomic coordinates in the data set over time. Using the digital image correlation algorithm, the atomic images of the medium surface obtained at different time points are compared and matched, and the relative displacement between adjacent atoms is calculated. According to Hooke's law, the displacement is converted into a strain value. The area covered by the entire data set is calculated point by point, and finally the strain distribution characteristic data of the storage medium is obtained, which intuitively presents the force deformation of the storage medium at the nanoscale. Based on the strain distribution characteristic data, search for local strain concentration areas, which are often accompanied by crystal dislocations. The dislocation density is quantified by calculating the total length of the dislocation line within a unit volume. For example, the line integral method is used to accumulate the length of the suspected dislocation line along the trajectory, and then divide it by the corresponding scanning area volume to obtain the dislocation density value. Observe the position changes of grain boundaries in different scanning periods. Compare the grain boundary contours in the previous and subsequent scanning data, use image recognition technology to extract the grain boundary coordinates, and calculate the average distance and direction of grain boundary movement to determine the grain boundary migration situation. Combine the dislocation density and grain boundary migration information to generate crystal degradation characteristic data. Analyze the charge distribution and changes at the mutation point in the electron transition identification data. According to the law of conservation of charge, track the flow direction and aggregation area of ​​electrons before and after the mutation, and use Gauss's law to calculate the local charge density changes. Integrate the data of multiple mutation points to construct a dynamic model of charge migration over time, thereby obtaining the charge migration characteristic data of the solid-state hard disk, which reflects the dynamic transfer characteristics of charge during electron migration. According to the formula First, extract the local lattice distortion value of each crystal from the crystal degradation characteristic data , add these values ​​and divide by the initial value of the overall lattice parameter of the crystal , and obtain the crystal degradation characteristic coefficient. According to the formula , the charge distribution change function in the charge transfer characteristic data of the solid state drive Derivative, take its absolute value and then in the time interval Integrate the above to calculate the charge migration accumulation characteristic coefficient. Calculate the standard deviation of the crystal degradation characteristic coefficient and the charge migration accumulation characteristic coefficient respectively. and , add up to get the characteristic fluctuation intensity value Finally, the above calculated , And the pre-set environmental coupling factor coefficient Substitute into the fatigue degradation index calculation formula ,in , , , It is a weight coefficient determined based on experiments and experience. The fatigue degradation index is calculated through this formula, and the hard disk fatigue effect identification data is generated to accurately evaluate the fatigue degradation degree of the internal storage medium of the solid state drive.

[0125] Preferably, step S4 comprises the following steps:

[0126] Step S41: performing hard disk usage association analysis on the solid state hard disk electronic transition identification data and the hard disk usage fatigue effect identification data to generate solid state fatigue characteristic mapping data;

[0127] Step S42: performing real-time hard disk life prediction based on the solid-state fatigue characteristic mapping data to generate real-time hard disk life prediction data; comparing the real-time hard disk life prediction data with a preset standard life warning threshold, and performing life warning processing when the real-time hard disk life prediction data is less than the preset standard life warning threshold;

[0128] Step S43: When the real-time life prediction data of the hard disk is greater than or equal to the preset standard life warning threshold, the solid state drive is adaptively written based on the real-time life prediction data of the hard disk, thereby performing the solid state drive life extension optimization operation.

[0129] In the embodiment of the present invention, the linear correlation between the two sets of data is measured by using the Pearson correlation coefficient method. The correlation coefficient of the corresponding data points of the electronic transition identification data and the fatigue effect identification data is calculated point by point, and the formula is: ,in , are the elements in the two sets of data, , is the mean, is the number of data points. The coefficient ranges from -1 to 1. The closer the absolute value is to 1, the stronger the correlation is. Based on the calculated correlation coefficient, a two-dimensional mapping table is constructed. The horizontal axis represents the key indicators in the electronic transition identification data, such as the electronic transition frequency; the vertical axis represents the important parameters in the hard disk fatigue effect identification data, such as the degree of crystal degradation. Fill in the corresponding correlation coefficient in the table cell to clearly present the correlation between the two and generate solid-state fatigue characteristic mapping data. Based on the solid-state fatigue characteristic mapping data, a suitable machine learning algorithm is selected to construct a life prediction model, such as support vector regression (SVR). The features in the mapping data are used as input, and the known corresponding solid-state hard disk remaining life data (which can be accumulated through a large number of previous experiments and actual use cases) is used as output to train the SVR model. During the training process, the model performance is optimized by adjusting the kernel function parameters (such as adjusting the bandwidth parameters when the radial basis function kernel is selected) until the model reaches a high accuracy on the validation set. The currently acquired solid-state fatigue characteristic mapping data is input into the trained model in real time to obtain the real-time hard disk life prediction data, which is presented in intuitive forms such as the remaining usable time and the remaining number of runs. Compare the real-time life prediction data of the hard disk with the preset standard life warning threshold. If the prediction data is less than the warning threshold, the life warning mechanism is triggered. The system pop-up window can prompt the user "The remaining life of your solid-state drive is insufficient, please back up the data in time", and record detailed warning information to the system log for subsequent tracing. Dynamically adjust the write strategy of the solid-state drive based on the real-time life prediction data of the hard disk. If the prediction data shows that the remaining life is sufficient but close to the warning threshold, appropriately reduce the write frequency, such as from 100 write operations per second to 50 write operations per second. At the same time, adjust the write data block size, give priority to using larger data blocks for writing, and reduce the additional loss caused by frequent small data block writing. Implement the above adaptive write strategy at the driver level of the solid-state drive. By updating the write parameter configuration table of the driver, the adjusted write frequency, data block size and other parameters are applied to the actual write operation. Before each write operation, the driver first queries the current real-time life prediction data, and optimizes the write process accordingly, thereby extending the service life of the solid-state drive.

[0130] In this specification, a solid state drive life prediction system based on big data is provided, which is used to execute the solid state drive life prediction method based on big data. The solid state drive life prediction system based on big data includes:

[0131] The power data acquisition module is used to obtain the solid-state hard disk structure data; perform operation simulation based on the solid-state hard disk structure data to generate the solid-state hard disk operation simulation data; use the solid-state hard disk operation simulation data to collect the power fluctuation data of the flash memory unit in the hard disk on the solid-state hard disk structure data to obtain the power fluctuation data of the flash memory unit in the key area of ​​the solid-state hard disk;

[0132] The electronic transition analysis module is used to classify the SSD operation based on the SSD operation simulation data, and generate write operation and erase operation; based on the write operation and erase operation, the SSD abnormal change trend analysis is performed on the power supply fluctuation data of the flash memory unit in the key area of ​​the SSD, and the SSD abnormal change trend data is generated; the SSD abnormal change trend data is used to identify the electronic migration effect of the SSD, and the SSD electronic transition identification data is generated;

[0133] The nanomaterial fatigue analysis module is used to obtain the time series data of the solid state drive; scan the internal storage medium of the solid state drive for the time series data of the solid state drive, and obtain the internal storage medium scan data set of the solid state drive; identify the fatigue effect of the nano-level material of the solid state drive for the internal storage medium scan data set of the solid state drive, and generate the hard disk fatigue effect identification data;

[0134] The real-time life prediction module is used to perform hard disk usage correlation analysis on the solid-state hard disk electronic transition identification data and the hard disk usage fatigue effect identification data to generate solid-state fatigue characteristic mapping data; perform hard disk real-time life prediction based on the solid-state fatigue characteristic mapping data to generate hard disk real-time life prediction data; adaptively write or warn the hard disk real-time life prediction data through a preset standard life warning threshold, thereby executing solid-state hard disk life extension optimization operations.

[0135] The beneficial effect of the present invention is that by collecting and running simulation of the solid-state hard disk structure data, the power supply fluctuation of the flash memory unit in the hard disk can be monitored in real time, especially the fluctuation of the key area. This makes it possible to more accurately capture the power supply problems that affect the performance and life of the hard disk, and provides accurate data support for subsequent analysis. By classifying the write and erase operations of the solid-state hard disk and analyzing the abnormal trend of the power supply fluctuation data of the flash memory unit, the electronic migration effect (electronic transition) that may cause the hard disk performance degradation and failure can be identified. This identification can warn the hard disk failure in advance, avoid data loss or hard disk damage caused by electronic migration, and thus improve the reliability of the hard disk. The analysis of the timing data used by the solid-state hard disk can deeply identify the nano-level fatigue effect of the hard disk storage medium. By scanning the internal storage medium of the hard disk, potential fatigue damage can be identified, which provides an accurate basis for early fault warning and helps prevent the hard disk performance degradation or complete failure caused by material fatigue. By performing adaptive write optimization or early warning based on real-time life prediction, targeted extension measures can be taken before the hard disk life is about to expire. Through reasonable write strategies and optimization operations, the wear and aging of the hard disk can be effectively delayed, thereby improving the actual service life of the solid-state hard disk. Therefore, the present invention improves the accuracy of life prediction for solid-state drives through multi-dimensional data collection and analysis, combined with factors such as hard drive operation, power supply fluctuations, electron migration effects, and material fatigue.

[0136] Therefore, the embodiments should be regarded as illustrative and non-restrictive from all points, and the scope of the present invention is limited by the appended claims rather than the above description, and it is therefore intended that all changes falling within the meaning and range of equivalent elements of the application documents are included in the present invention.

[0137] The above description is only a specific embodiment of the present invention, so that those skilled in the art can understand or implement the present invention. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to the embodiments shown herein, but should conform to the widest scope consistent with the principles and novel features invented herein.

Claims

1. A method for predicting the life of a solid-state drive based on big data, characterized in that: The following steps are involved: Step S1: acquiring the solid state drive structure data; performing operation simulation based on the solid state drive structure data, thereby generating the solid state drive operation simulation data; using the solid state drive operation simulation data to collect the power fluctuation data of the flash memory unit in the hard disk on the solid state drive structure data, and obtaining the power fluctuation data of the flash memory unit in the key area of ​​the solid state drive; Step S2: classifying SSD operation on SSD operation simulation data to generate write operation and erase operation; performing SSD abnormal change trend analysis on SSD key area flash memory unit power supply fluctuation data based on write operation and erase operation to generate SSD abnormal change trend data; using SSD abnormal change trend data to identify electron migration effect on SSD to generate SSD electron transition identification data; Step S2 includes the following steps: Step S21: classifying SSD operation on SSD operation simulation data to generate SSD operation classification data, wherein the SSD operation classification data includes write operation and erase operation; performing write current fluctuation screening on power fluctuation data of flash memory units in key areas of the SSD based on the write operation to obtain write operation current fluctuation data; Step S22: performing an erase voltage fluctuation screening on the power fluctuation data of the flash memory unit in the key area of ​​the solid state hard disk based on the erase operation to obtain the erase operation voltage fluctuation data; performing a fluctuation amplitude change analysis on the write operation current fluctuation data and the erase operation voltage fluctuation data to generate the operation amplitude change data and the operation current peak change data; Step S23: performing abnormal fluctuation analysis on the operation amplitude change data and the operation current peak change data to perform abnormal change trend analysis on the solid state hard disk, and generating abnormal change trend data on the solid state hard disk; Step S24: using the abnormal change trend data of the solid state hard disk to identify the electron migration effect of the solid state hard disk, and generating the solid state hard disk electron transition identification data; Step S24 includes the following steps: Step S241: defining the current-voltage signal of the abnormal change trend data of the solid state hard disk based on a preset time window to generate a current signal and a voltage signal; calculating the instantaneous value of the current signal and the voltage signal to obtain the instantaneous value of the current and the instantaneous value of the voltage; Step S242: constructing a phase space trajectory through the instantaneous current value and the instantaneous voltage value to generate a current-voltage phase diagram of the solid state drive; performing trajectory overlap detection on the current-voltage phase diagram of the solid state drive to generate a phase trajectory overlap curve of the solid state drive; Step S243: identifying the current-voltage mutation point of the solid-state hard disk current-voltage phase diagram according to the solid-state hard disk phase trajectory overlap curve, thereby obtaining the current-voltage mutation point; Step S244: using the current-voltage mutation point to mark the hard disk electronic transition mutation signal on the current signal and the voltage signal, and obtaining the hard disk electronic transition mutation mutation marking signal; calculating the proportion of the hard disk electronic transition mutation marking signal to the current signal and the voltage signal, and generating the solid state hard disk electronic transition identification data; Step S3: acquiring the time series data of the solid state drive; scanning the internal storage medium of the solid state drive on the time series data of the solid state drive to obtain a data set of scanning the internal storage medium of the solid state drive; identifying the fatigue effect of the nano-material of the solid state drive on the data set of scanning the internal storage medium of the solid state drive to generate data of identifying the fatigue effect of the solid state drive; Step S4: Perform hard disk usage correlation analysis on the solid-state hard disk electronic transition identification data and the hard disk usage fatigue effect identification data to generate solid-state fatigue characteristic mapping data; perform hard disk real-time life prediction based on the solid-state fatigue characteristic mapping data to generate hard disk real-time life prediction data; adaptively write or warn the hard disk real-time life prediction data through a preset standard life warning threshold, thereby executing the solid-state hard disk life extension optimization operation.

2. The method for predicting the life of a solid state drive based on big data according to claim 1, characterized in that: Step S1 includes the following steps: Step S11: Acquire the solid state hard disk structure data; Step S12: performing physical topology modeling on the SSD structure data to generate SSD topology structure modeling data; performing operation simulation on the SSD based on the SSD topology structure modeling data to generate SSD operation simulation data; Step S13: marking the key area of ​​the SSD structure data using the SSD running simulation data to generate the SSD key area marking data; Step S14: Based on the integrated current and voltage sensors, the power supply fluctuation data of the flash memory unit in the SSD is collected for the key area mark data of the SSD to obtain the power supply fluctuation data of the flash memory unit in the key area of ​​the SSD.

3. The method for predicting the life of a solid state drive based on big data according to claim 2, characterized in that: The simulation of running the SSD based on the SSD topology modeling data includes: Setting operation simulation rules for the solid state drive based on the solid state drive topology structure modeling data to obtain the solid state drive operation simulation rules, wherein the solid state drive operation simulation rules include operation sequence, data block size, and operation frequency; According to the preset operation sequence, data block size and operation frequency, the operation parameters of the solid state drive are configured to generate operation simulation rule configuration data; The solid-state drive is simulated in a virtual environment using the simulation rule configuration data to generate a solid-state drive simulation data set; the solid-state drive simulation data set is preprocessed to generate solid-state drive simulation data, wherein the data preprocessing includes data cleaning, data denoising, data missing value filling and data standardization.

4. The method for predicting the life of a solid state drive based on big data according to claim 1, characterized in that: Step S3 includes the following steps: Step S31: obtaining the solid state hard disk usage timing data; Step S32: performing time segmentation on the SSD usage time series data to generate the SSD usage time series segmentation data; using the SSD usage time series segmentation data as a scanning time interval, and scanning the internal storage medium of the SSD using a scanning probe microscope technology according to the scanning time interval to obtain a SSD internal storage medium scanning data set; Step S33: performing hard disk nano-material use fatigue effect identification on the internal storage medium scanning data set of the solid state hard disk to generate hard disk use fatigue effect identification data.

5. The method for predicting the life of a solid state drive based on big data according to claim 4, characterized in that: Step S33 includes the following steps: Step S331: performing abnormal point detection on the data set of the scanned storage medium inside the solid state drive to generate a data set of characteristics without abnormality; performing nanoscale strain distribution analysis on the data set of characteristics without abnormality to generate strain distribution characteristic data of the storage medium; Step S332: performing crystal structure degradation detection on the internal storage medium scan data set of the solid state drive according to the storage medium strain distribution characteristic data to generate crystal degradation characteristic data, wherein the crystal structure degradation detection includes dislocation density detection and grain boundary migration detection; Step S333: Perform charge transfer characteristic analysis on the solid state drive electronic transition identification data to generate solid state drive charge transfer characteristic data; perform fatigue degradation index calculation on the solid state drive internal storage medium scan data set using the crystal degradation characteristic data and the solid state drive charge transfer characteristic data to obtain hard drive use fatigue effect identification data, wherein the fatigue degradation index calculation formula is as follows: ； ； ； ； In the formula, Expressed as fatigue degradation index, Expressed as the crystal degradation coefficient, Expressed as the charge transfer accumulation characteristic coefficient, Expressed as the environmental coupling factor coefficient, Expressed as a characteristic fluctuation intensity value, Expressed as the crystal degradation characteristic weight coefficient, Expressed as the charge transfer characteristic weight coefficient, Expressed as the weight coefficient of the environmental coupling factor, Expressed as the characteristic fluctuation intensity correction coefficient, Expressed as The local lattice distortion value of a crystal, is represented as the initial value of the overall lattice parameter of the crystal, Expressed as the number of crystals, Expressed as time The charge distribution change function inside is Represents the time limit, Expressed as the standard deviation of the crystal degradation characteristics, Expressed as the standard deviation of the charge transfer accumulation characteristic.

6. The method for predicting the life of a solid state drive based on big data according to claim 1, characterized in that: Step S4 includes the following steps: Step S41: performing hard disk usage association analysis on the solid state hard disk electronic transition identification data and the hard disk usage fatigue effect identification data to generate solid state fatigue characteristic mapping data; Step S42: performing real-time hard disk life prediction based on the solid-state fatigue characteristic mapping data to generate real-time hard disk life prediction data; comparing the real-time hard disk life prediction data with a preset standard life warning threshold, and performing life warning processing when the real-time hard disk life prediction data is less than the preset standard life warning threshold; Step S43: When the real-time life prediction data of the hard disk is greater than or equal to the preset standard life warning threshold, the solid state drive is adaptively written based on the real-time life prediction data of the hard disk, thereby performing the solid state drive life extension optimization operation.

7. The method for predicting the life of a solid state drive based on big data according to claim 6, characterized in that: Real-time hard disk life prediction based on solid-state fatigue property mapping data includes: Extract key characteristic indicators from the solid-state fatigue characteristic mapping data to obtain hard disk fatigue characteristic indicators; divide the solid-state fatigue characteristic mapping data into data sets based on the hard disk fatigue characteristic indicators to generate a model training set and a model test set; The model training set is trained using the long short-term memory neural network algorithm to generate a hard disk real-time life prediction pre-model; the hard disk real-time life prediction pre-model is optimized and iterated using the model test set to generate a hard disk real-time life prediction model; The solid fatigue characteristic mapping data is imported into the hard disk real-time life prediction model to perform hard disk real-time life prediction and generate hard disk real-time life prediction data.

8. A solid state drive life prediction system based on big data, characterized in that: Used to execute the solid state drive life prediction method based on big data as claimed in claim 1, the solid state drive life prediction system based on big data comprises: The power data acquisition module is used to obtain the solid-state hard disk structure data; perform operation simulation based on the solid-state hard disk structure data to generate the solid-state hard disk operation simulation data; use the solid-state hard disk operation simulation data to collect the power fluctuation data of the flash memory unit in the hard disk on the solid-state hard disk structure data to obtain the power fluctuation data of the flash memory unit in the key area of ​​the solid-state hard disk; The electronic transition analysis module is used to classify the SSD operation based on the SSD operation simulation data, and generate write operation and erase operation; based on the write operation and erase operation, the SSD abnormal change trend analysis is performed on the power supply fluctuation data of the flash memory unit in the key area of ​​the SSD, and the SSD abnormal change trend data is generated; the SSD abnormal change trend data is used to identify the electronic migration effect of the SSD, and the SSD electronic transition identification data is generated; The nanomaterial fatigue analysis module is used to obtain the time series data of the solid state drive; scan the internal storage medium of the solid state drive for the time series data of the solid state drive, and obtain the internal storage medium scan data set of the solid state drive; identify the fatigue effect of the nano-level material of the solid state drive for the internal storage medium scan data set of the solid state drive, and generate the hard disk fatigue effect identification data; The real-time life prediction module is used to perform hard disk usage correlation analysis on the solid-state hard disk electronic transition identification data and the hard disk usage fatigue effect identification data to generate solid-state fatigue characteristic mapping data; perform hard disk real-time life prediction based on the solid-state fatigue characteristic mapping data to generate hard disk real-time life prediction data; adaptively write or warn the hard disk real-time life prediction data through a preset standard life warning threshold, thereby executing solid-state hard disk life extension optimization operations.

Images