Solid-state drive energy consumption prediction method and system based on deep learning

Abstract

The present invention relates to the technical field of solid-state hard disk energy consumption analysis, and in particular to a method and system for predicting solid-state hard disk energy consumption based on deep learning. The method comprises the following steps: obtaining a solid-state hard disk drive operation log; performing a comprehensive hard disk energy consumption evaluation on the solid-state hard disk drive operation log to generate solid-state hard disk energy consumption data; obtaining multiple user request behavior data based on the solid-state hard disk drive operation log, and performing read and write energy consumption calculation based on the solid-state hard disk energy consumption data, thereby obtaining energy consumption data for each hard disk read and write operation behavior; performing multi-period hard disk read and write operation identification on multiple user request behavior data, and performing read and write logic implicit association learning, thereby generating user read and write logic association features; performing time window division on the solid-state hard disk drive operation log based on the user read and write logic association features, and constructing multiple potential behavior prediction scenarios. The present invention realizes accurate and efficient prediction of solid-state hard disk energy consumption.

Description

Technical Field

[0001] The present invention relates to the technical field of solid-state drive energy consumption analysis, and in particular to a method and system for predicting solid-state drive energy consumption based on deep learning. Background Art

[0002] With the continuous improvement of computing device performance and the increasing functionality, energy management of electronic products has become an important issue that needs to be addressed. As one of the key components of computing devices, the energy consumption characteristics of solid-state drives (SSDs) directly affect the energy efficiency and battery life of the entire system. For mobile devices such as laptops, the energy consumption of SSDs will directly affect battery life, and excessive energy consumption will cause the hard drive to overheat, which will in turn shorten the life of the SSD as a whole. Therefore, analyzing and predicting the energy consumption of SSDs has become an important part of improving SSD performance.

[0003] Traditional SSD energy consumption prediction methods mainly rely on empirical formulas of hardware parameters or simple test data. It is difficult to accurately capture the dynamic energy consumption characteristics of SSDs in complex working environments. There are often problems such as inaccurate energy consumption prediction and low analysis efficiency. With the rapid development of artificial intelligence technology, modeling and prediction methods based on deep learning have shown excellent performance in many fields. Compared with traditional energy consumption analysis methods, deep learning can automatically extract complex features in the SSD working process and build a more accurate energy consumption prediction model. However, the direct application of general deep learning models also has some limitations. Therefore, it is urgent to develop an intelligent energy consumption prediction method. Summary of the invention

[0004] In order to solve the above technical problems, the present invention proposes a solid state drive energy consumption prediction method and system based on deep learning to solve at least one of the above technical problems.

[0005] To achieve the above object, the present invention provides a method for predicting energy consumption of a solid state drive based on deep learning, comprising the following steps:

[0006] Preferably, step S1 comprises the following steps:

[0007] Step S1: obtaining a solid-state hard disk drive operation log; performing real-time operation status analysis on the solid-state hard disk drive operation log, and performing a comprehensive evaluation of the hard disk drive energy consumption to generate solid-state hard disk drive energy consumption data;

[0008] Step S2: obtaining multiple user request behavior data based on the solid state hard disk drive operation log, and performing read and write energy consumption calculation based on the solid state hard disk drive energy consumption data, thereby obtaining energy consumption data of each hard disk drive read and write operation behavior;

[0009] Step S3: Identify hard disk read and write operations in multiple time periods for multiple user request behavior data, and perform implicit association learning of read and write logic, thereby generating user read and write logic association features;

[0010] Step S4: dividing the solid-state hard disk drive operation log into time windows based on the user read-write logic association characteristics, and performing behavior evolution prediction to construct multiple potential behavior prediction scenarios;

[0011] Step S5: simulating the scene energy consumption of multiple potential behavior prediction scenarios according to the energy consumption data of each hard disk read and write operation behavior, and generating scene energy consumption data of all potential behavior prediction scenarios;

[0012] Step S6: Predict the aging state evolution of the scene energy consumption data of all potential behavior prediction scenes, re-estimate the hard disk aging energy consumption, and construct a hard disk scene energy consumption prediction curve.

[0013] The present invention can understand the current state of the solid-state hard disk by performing real-time analysis on the operation log, which is helpful for monitoring the operation of the hard disk. Energy consumption evaluation is helpful for understanding the energy consumption of the hard disk, and provides basic data for subsequent energy consumption calculation and prediction. Read and write energy consumption calculation is performed according to user request behavior data and hard disk energy consumption data, and the energy consumption of each read and write operation can be quantified, providing detailed energy consumption data for subsequent analysis. By identifying read and write operations and learning logical associations, the user's read and write behavior pattern can be revealed, providing a basis for subsequent behavior evolution prediction. Time window division and behavior evolution prediction based on logical association characteristics can construct multiple potential behavior prediction scenarios, providing a basis for scenario energy consumption simulation. Through aging state evolution prediction and energy consumption re-estimation, the energy consumption of the hard disk can be more accurately predicted, providing a reference for the long-term use of the hard disk. Constructing a hard disk scenario energy consumption prediction curve is helpful for understanding the energy consumption characteristics of the hard disk in different usage scenarios, and providing guidance for energy saving optimization and performance adjustment. Through scenario energy consumption simulation, energy consumption data of all potential behavior prediction scenarios can be generated, providing detailed data support for hard disk energy consumption prediction.

[0014] Preferably, step S1 comprises the following steps:

[0015] Step S11: obtaining a solid state hard disk drive operation log;

[0016] Step S12: Analyze the real-time operation status of the solid state hard disk drive operation log and extract the real-time operation status parameters;

[0017] Step S13: performing time stamp alignment processing on the real-time operating status parameters to obtain time-standardized operating status parameters;

[0018] Step S14: performing real-time temperature calculation on the time-standardized operating state parameters to generate real-time temperature data of the solid state hard disk;

[0019] Step S15: Calculating the power load based on the real-time operating status parameters to obtain power load data;

[0020] Step S16: Perform a comprehensive hard disk energy consumption evaluation on the power load data according to the real-time temperature data of the solid state hard disk to generate the solid state hard disk energy consumption data.

[0021] The present invention records the operating status and activity information of the solid-state hard disk by providing an original data source, thus providing a basis for subsequent analysis and prediction. The ability to monitor the status of the solid-state hard disk in real time helps to discover potential problems or abnormal situations. Extract key operating status parameters to provide data support for subsequent analysis and calculations. Ensure that the data is on the same timeline to facilitate subsequent time-related calculations and analysis. Obtaining real-time temperature data through temperature calculation can help monitor the temperature changes of the hard disk and provide important indicators for energy consumption evaluation. By performing power load calculation on the real-time operating status parameters, the power consumption of the hard disk can be understood, providing important data for energy consumption evaluation. Combine real-time temperature data and power load data to perform a comprehensive energy consumption evaluation, generate energy consumption data for the solid-state hard disk, and provide a basis for energy consumption prediction.

[0022] Preferably, the specific steps of step S2 are:

[0023] Step S21: performing user request behavior analysis on the solid state hard disk drive operation log to obtain a plurality of user request behavior data;

[0024] Step S22: classifying the hard disk read / write response types of the multiple user request behavior data, and extracting the hard disk read / write operation response data;

[0025] Step S23: performing time stamp sorting processing on the hard disk read and write operation response data to generate a read and write operation response timing sequence;

[0026] Step S24: performing timestamp matching on the read and write operation response timing sequence according to the time-standardized operating status parameter, thereby obtaining read and write operation response matching data;

[0027] Step S25: Calculate the read and write energy consumption of the read and write operation response matching data based on the energy consumption data of the solid state hard disk, so as to obtain energy consumption data of each hard disk read and write operation behavior.

[0028] The present invention can obtain multiple user request behavior data by analyzing the solid-state hard disk drive operation log, providing a basis for subsequent analysis and prediction, and can extract the hard disk read and write operation response data by classifying the user request behavior data, providing key information for subsequent energy consumption calculation, and can generate a timing sequence of read and write operation responses through timestamp sorting processing, providing an accurate data basis for subsequent timestamp matching, and can perform timestamp matching on the read and write operation timing sequence through time-standardized operating status parameters to obtain read and write operation response matching data, providing associated information for energy consumption calculation. Based on the solid-state hard disk energy consumption data, energy consumption calculation is performed on the read and write operation response matching data to obtain energy consumption data for each hard disk read and write operation behavior, providing detailed energy consumption information for energy consumption prediction.

[0029] Preferably, step S3 specifically comprises the following steps:

[0030] Step S31: performing time series analysis on multiple user request behavior data to obtain time series request behavior data;

[0031] Step S32: performing user operation logic analysis on the time sequence request behavior data to obtain user behavior logic data;

[0032] Step S33: performing multi-period hard disk read and write operation identification based on the solid state hard disk drive operation log, and generating a multi-stage hard disk read and write track;

[0033] Step S34: performing implicit association learning of read-write logic on multi-stage hard disk read-write traces according to user behavior logic data, thereby generating user read-write logic association features.

[0034] The present invention can identify the patterns and trends of user request behaviors through time series analysis, which is helpful for understanding the regularities of user behaviors. It can serialize user request behavior data by time, and provide an orderly data basis for subsequent logical analysis of user behaviors. Through logical analysis, a logical model of user behaviors can be established to reveal the user's operating habits and behavior patterns. In-depth analysis of user operation logic can help to explore the regularities and motivations behind user behaviors, and provide a basis for subsequent hard disk operation identification. Through log analysis, it can accurately identify the hard disk's read and write operation modes, including read and write behaviors in different time periods, and generate multi-stage hard disk read and write trajectories, which is helpful for understanding the hard disk's usage and read and write modes in different time periods. Combined with user behavior logic and hard disk read and write trajectories, association learning is performed to reveal the potential association between user behavior and hard disk read and write, generate user read and write logic association features, and provide richer feature information for deep learning models for prediction and optimization of solid-state hard disk energy consumption.

[0035] Preferably, the specific steps of step S4 are:

[0036] Step S41: dividing the solid state hard disk drive operation log into time windows, thereby obtaining a plurality of time window operation cycle data;

[0037] Step S42: Perform user request frequency statistics on multiple time window operation cycle data to generate user request behavior frequency parameters for each cycle;

[0038] Step S43: performing behavior type distribution analysis on the user request behavior frequency parameters of each cycle to obtain user request behavior type distribution data;

[0039] Step S44: Based on the user read-write logic association characteristics, the user request behavior type distribution data is predicted to perform behavior evolution, and multiple potential behavior prediction scenarios are constructed.

[0040] The present invention divides the operation log data into multiple time periods through time window division, which is convenient for subsequent data processing and analysis. The data of each time window can provide operation characteristics in different time periods, which is helpful to capture the time correlation of data. The frequency distribution of user request behavior in each cycle can be understood through frequency statistics, which is helpful to understand user behavior habits and needs. User request behavior frequency parameters can be used as input features of deep learning models to provide information on behavior patterns. The distribution of different types of user requests can be revealed through behavior type distribution analysis, which is helpful to identify user behavior patterns and trends. Behavior type distribution data can be displayed in a visual form to help users understand user behavior characteristics more intuitively. Behavior evolution prediction can be used to infer the future development trend of user request behavior, providing a reference for system performance optimization. Constructing multiple potential behavior prediction scenarios can help the system make corresponding adjustments to different situations and improve the accuracy and robustness of the prediction.

[0041] Preferably, the specific steps of step S44 are:

[0042] Performing probability distribution calculation on the user request behavior type distribution data to generate behavior type probability distribution data;

[0043] Based on the user read-write logic association characteristics, the up-down dependency analysis between user behaviors is performed on multiple time window operation cycle data to obtain the up-down dependency relationship between behaviors;

[0044] Based on the up-down dependency relationship between behaviors, the convolution sequence of the behavior type probability distribution data is learned to obtain the distribution law of user behavior;

[0045] Conduct multi-time point behavior simulation based on user behavior distribution rules to obtain multiple behavior simulation trajectories;

[0046] Perform behavioral evolution prediction on multiple behavioral simulation trajectories to obtain multi-time point behavioral evolution prediction data;

[0047] Identify potential behavior scenarios based on multi-time point operation behavior prediction data and construct multiple potential behavior prediction scenarios.

[0048] The present invention can quantify the occurrence probability of different behavior types by calculating the probability distribution of behavior types, providing a mathematical basis for subsequent analysis. Generating behavior type probability distribution data helps to better understand user behavior patterns and the distribution laws of behavior types. Analyzing the upper and lower dependencies between user behaviors can reveal the correlation and influence degree between different behaviors, which helps to understand the dynamic evolution of user behaviors. Through dependency analysis, the patterns and laws between user behaviors can be mined to provide a basis for behavior analysis and prediction. Through convolutional sequence learning, the sequence characteristics of user behavior distribution can be captured to reveal the laws and trends of user behavior evolution. Obtaining the user behavior distribution law helps to establish a more accurate behavior model and provide a more reliable basis for energy consumption prediction. Conducting multi-time point behavior simulation can simulate the evolution of user behavior at different time points, help understand the dynamic changes of user behavior, generate multiple behavior simulation trajectories to help observe different behavior evolution scenarios, and provide diverse references for behavior evolution prediction. Evolution prediction through behavior simulation trajectories can predict the possible development trajectory of future user behavior and provide early warning for system optimization and adjustment. Multi-time point behavior evolution prediction data can help verify the accuracy and robustness of the model and provide support for decision-making. Identifying potential behavior prediction scenarios can help the system predict user behavior patterns in different situations and provide strategies for system adjustment and response. Constructing multiple potential behavior prediction scenarios can help formulate more comprehensive response strategies and improve the system's ability to respond to future changes.

[0049] Preferably, the specific steps of step S5 are:

[0050] Step S51: extracting a first potential behavior prediction scenario based on multiple potential behavior prediction scenarios;

[0051] Step S52: Based on the energy consumption data of each hard disk read and write operation behavior, a scenario energy consumption simulation is performed on the first potential behavior prediction scenario to obtain the operation behavior energy consumption data of the first scenario;

[0052] Step S53: performing energy consumption superposition calculation on the operation behavior energy consumption data of the first scene, thereby generating the energy consumption data of the first scene;

[0053] Step S54: Iteratively calculate all potential behavior prediction scenarios to generate scenario energy consumption data for all potential behavior prediction scenarios.

[0054] The present invention helps to specifically describe a specific behavior scenario by extracting the first potential behavior prediction scenario, providing a specific scenario for subsequent simulation and analysis. By extracting different potential behavior prediction scenarios, the potential behavior patterns can be classified and subdivided, which helps to refine the overall behavior prediction. The scene energy consumption simulation can simulate the energy consumption of the hard disk in a specific behavior scenario, providing real data basis for energy consumption prediction. Obtaining the operation behavior energy consumption data of the first scenario helps to establish an energy consumption model in the scenario, providing data support for subsequent energy consumption superposition calculation. The energy consumption superposition calculation can comprehensively consider the impact of different operation behaviors on energy consumption, and more accurately estimate the overall energy consumption in a specific scenario. Generating the first scene energy consumption data helps to analyze the energy consumption distribution and characteristics in the scenario, and provides a detailed data basis for overall energy consumption prediction. By iteratively calculating the scene energy consumption data of all potential behavior prediction scenarios, the impact of different scenarios on hard disk energy consumption can be fully considered, improving the comprehensiveness and accuracy of the prediction. Generating the scene energy consumption data of all potential behavior prediction scenarios helps to analyze the energy consumption characteristics and differences in different scenarios, and provides a variety of references for formulating energy consumption management strategies.

[0055] Preferably, the specific steps of step S6 are:

[0056] Step S61: performing a time series temperature change analysis on the real-time temperature data of the solid state hard disk to generate temperature change data of the solid state hard disk;

[0057] Step S62: performing load operation fluctuation analysis on the power load data to obtain load operation fluctuation data;

[0058] Step S63: performing aging state evolution prediction based on the solid state hard disk temperature change data and load operation fluctuation data, thereby generating hard disk aging state evolution prediction data;

[0059] Step S64: re-estimating the hard disk aging energy consumption of the scene energy consumption data of all potential behavior prediction scenes based on the hard disk aging state evolution prediction data to obtain the scene energy consumption data under the aging state;

[0060] Step S65: Perform time series energy consumption fitting on the scene energy consumption data in the aging state to construct a hard disk scene energy consumption prediction curve.

[0061] The present invention can help understand the temperature variation law of the hard disk under different loads by analyzing the real-time temperature data of the solid-state hard disk, and provide basic data for the subsequent prediction of the aging state evolution. The generation of solid-state hard disk temperature variation data is helpful to predict the temperature trend of the hard disk under different load conditions, and provide a basis for the prediction of the aging state evolution. The analysis of the fluctuation of power load data can reveal the load variation law of the hard disk during operation, and provide necessary information for the subsequent prediction of the aging state evolution. The load operation fluctuation data is helpful to understand the dynamic characteristics of the hard disk load, and provide data support for the subsequent prediction of the aging state evolution. By combining the temperature variation data and the load fluctuation data for the aging state evolution prediction, the aging trend of the hard disk in different usage scenarios can be predicted more accurately, and the hard disk aging can be generated. The state evolution prediction data helps to understand the possible state changes of the hard disk during long-term use, and provides a basis for subsequent energy consumption re-estimation. By re-estimating the hard disk aging energy consumption based on the energy consumption data of the potential behavior prediction scenario, the impact of hard disk aging on energy consumption can be more realistically considered to improve the accuracy of the prediction. Obtaining the scene energy consumption data under the aging state helps to adjust the energy consumption prediction model and better reflect the actual energy consumption of the hard disk in different aging states. By performing time-series energy consumption fitting on the scene energy consumption data under the aging state, a model of the energy consumption change of the hard disk in different aging states can be established to provide a more accurate curve for prediction. Constructing the hard disk scene energy consumption prediction curve helps to intuitively display the energy consumption characteristics of the hard disk under different usage conditions, and provides visual support for energy consumption management and optimization.

[0062] In this specification, a solid-state drive energy consumption prediction system based on deep learning is provided, which is used to execute the solid-state drive energy consumption prediction method based on deep learning as described above, including:

[0063] The hard disk energy consumption evaluation module is used to obtain the solid-state hard disk drive operation log; perform real-time operation status analysis on the solid-state hard disk drive operation log, and perform a comprehensive evaluation of the hard disk energy consumption to generate solid-state hard disk energy consumption data;

[0064] A read / write energy consumption calculation module is used to obtain multiple user request behavior data based on the solid-state hard disk drive operation log, and perform read / write energy consumption calculation based on the solid-state hard disk drive energy consumption data, thereby obtaining energy consumption data for each hard disk drive read / write operation behavior;

[0065] The read-write logic association module is used to identify hard disk read-write operations in multiple time periods based on multiple user request behavior data, and to learn implicit associations of read-write logic, thereby generating user read-write logic association features;

[0066] The behavior prediction scenario module is used to divide the SSD operation log into time windows based on the user read-write logic association characteristics, and to predict the behavior evolution and construct multiple potential behavior prediction scenarios;

[0067] The scene energy consumption module is used to simulate the scene energy consumption of multiple potential behavior prediction scenes according to the energy consumption data of each hard disk read and write operation behavior, and generate the scene energy consumption data of all potential behavior prediction scenes;

[0068] The aging evolution prediction module is used to predict the aging state evolution of the scene energy consumption data of all potential behavior prediction scenarios, re-estimate the hard disk aging energy consumption, and construct the hard disk scene energy consumption prediction curve.

[0069] The present invention can timely understand the operation status of the hard disk by real-time analysis of the hard disk operation log, which is helpful for monitoring the energy consumption of the hard disk. By comprehensively analyzing the hard disk operation log, accurate hard disk energy consumption data can be generated to provide a basis for subsequent energy consumption calculations. Extracting user request behavior data based on the hard disk operation log is helpful for understanding the user's read and write operations on the hard disk. Read and write energy consumption calculations based on the hard disk energy consumption data can evaluate the energy consumption of each read and write operation, providing data support for subsequent behavior predictions. Identifying the hard disk read and write operations in multiple user request behavior data is helpful for understanding user behavior patterns. By learning the implicit correlation features of the read and write logic, the logical connection between user behaviors can be discovered to provide a basis for behavior predictions. Write logic association features to divide time windows and predict behavior evolution, which helps to understand the development trend of user behavior. Constructing multiple potential behavior prediction scenarios can provide different usage scenarios for subsequent energy consumption simulation. By simulating potential behavior prediction scenarios based on the energy consumption data of each read and write operation, the energy consumption of the hard disk under different scenarios can be evaluated. The energy consumption data of all potential behavior prediction scenarios is generated to provide basic data for the subsequent aging state evolution prediction. The aging state evolution prediction of the scenario energy consumption data can predict the energy consumption state of the hard disk under long-term use and provide a reference for energy consumption management. Re-estimation of hard disk aging energy consumption helps to adjust the energy consumption prediction model and construct the hard disk scenario energy consumption prediction curve to more accurately reflect the energy consumption of the hard disk. BRIEF DESCRIPTION OF THE DRAWINGS

[0070] Figure 1 A schematic diagram of the steps of a method for predicting energy consumption of a solid-state hard disk based on deep learning according to the present invention;

[0071] Figure 2 Detailed implementation flow chart of step S1;

[0072] Figure 3 Detailed implementation flow chart of step S2;

[0073] Figure 4 Detailed implementation flow chart of step S3. DETAILED DESCRIPTION

[0074] It should be understood that the specific embodiments described herein are only used to explain the present invention, and are not used to limit the present invention.

[0075] The present application example provides a method and system for predicting energy consumption of a solid-state hard disk based on deep learning. The execution subject of the method and system for predicting energy consumption of a solid-state hard disk based on deep learning includes but is not limited to: mechanical equipment, data processing platform, cloud server node, network upload device, etc. equipped with the system can be regarded as the general computing node of the present application, and the data processing platform includes but is not limited to: at least one of an audio and image management system, an information management system, and a cloud data management system.

[0076] See also Figures 1 to 4 The present invention provides a solid state drive energy consumption prediction method based on deep learning, and the solid state drive energy consumption prediction method based on deep learning comprises the following steps:

[0077] Step S1: obtaining a solid-state hard disk drive operation log; performing real-time operation status analysis on the solid-state hard disk drive operation log, and performing a comprehensive evaluation of the hard disk drive energy consumption to generate solid-state hard disk drive energy consumption data;

[0078] Step S2: obtaining multiple user request behavior data based on the solid state hard disk drive operation log, and performing read and write energy consumption calculation based on the solid state hard disk drive energy consumption data, thereby obtaining energy consumption data of each hard disk drive read and write operation behavior;

[0079] Step S3: Identify hard disk read and write operations in multiple time periods for multiple user request behavior data, and perform implicit association learning of read and write logic, thereby generating user read and write logic association features;

[0080] Step S4: dividing the solid-state hard disk drive operation log into time windows based on the user read-write logic association characteristics, and performing behavior evolution prediction to construct multiple potential behavior prediction scenarios;

[0081] Step S5: simulating the scene energy consumption of multiple potential behavior prediction scenarios according to the energy consumption data of each hard disk read and write operation behavior, and generating scene energy consumption data of all potential behavior prediction scenarios;

[0082] Step S6: Predict the aging state evolution of the scene energy consumption data of all potential behavior prediction scenes, re-estimate the hard disk aging energy consumption, and construct a hard disk scene energy consumption prediction curve.

[0083] The present invention can understand the current state of the solid-state hard disk by performing real-time analysis on the operation log, which is helpful for monitoring the operation of the hard disk. Energy consumption evaluation is helpful for understanding the energy consumption of the hard disk, and provides basic data for subsequent energy consumption calculation and prediction. Read and write energy consumption calculation is performed according to user request behavior data and hard disk energy consumption data, and the energy consumption of each read and write operation can be quantified, providing detailed energy consumption data for subsequent analysis. By identifying read and write operations and learning logical associations, the user's read and write behavior pattern can be revealed, providing a basis for subsequent behavior evolution prediction. Time window division and behavior evolution prediction based on logical association characteristics can construct multiple potential behavior prediction scenarios, providing a basis for scenario energy consumption simulation. Through aging state evolution prediction and energy consumption re-estimation, the energy consumption of the hard disk can be more accurately predicted, providing a reference for the long-term use of the hard disk. Constructing a hard disk scenario energy consumption prediction curve is helpful for understanding the energy consumption characteristics of the hard disk in different usage scenarios, and providing guidance for energy saving optimization and performance adjustment. Through scenario energy consumption simulation, energy consumption data of all potential behavior prediction scenarios can be generated, providing detailed data support for hard disk energy consumption prediction.

[0084] In the embodiment of the present invention, refer to Figure 1 , is a schematic flow chart of a step flow of a method for predicting energy consumption of a solid-state hard disk based on deep learning of the present invention. In this example, the steps of the method for predicting energy consumption of a solid-state hard disk based on deep learning include:

[0085] Step S1: obtaining a solid-state hard disk drive operation log; performing real-time operation status analysis on the solid-state hard disk drive operation log, and performing a comprehensive evaluation of the hard disk drive energy consumption to generate solid-state hard disk drive energy consumption data;

[0086] In this embodiment, the data source of the operation log of the solid-state hard disk drive is determined, including the built-in logging function of the drive or the use of system monitoring tools (such as SMART, CrystalDiskInfo), and the relevant tools or drive settings are configured to ensure real-time recording of the operation log, including read and write operations, temperature, power and other information, and the operation log of the solid-state hard disk is collected regularly or in real time. This can be achieved by using monitoring tools or writing scripts, and the collected operation log data is stored in a structured format (such as CSV, JSON or database) for subsequent analysis, use data analysis tools to load the saved SSD operation log data, clean the data, process missing values ​​and outliers, ensure data accuracy, convert the timestamp field to a date and time format that can be used for analysis, calculate key operating status indicators, such as current read and write rates, temperature monitoring, and power usage, use visualization tools to generate real-time status charts to help identify changes in operating status, define energy consumption models based on the working characteristics of the SSD, including static energy consumption and dynamic energy consumption, use the relationship between power and time, calculate the energy consumption of each operation, and summarize the total energy consumption, organize the energy consumption calculation results into a table, record the total energy consumption, number of read and write operations, and average power for each time period, and save the generated SSD energy consumption data in a structured format (such as CSV or Excel) for subsequent analysis and reporting.

[0087] Step S2: obtaining multiple user request behavior data based on the solid state hard disk drive operation log, and performing read and write energy consumption calculation based on the solid state hard disk drive energy consumption data, thereby obtaining energy consumption data of each hard disk drive read and write operation behavior;

[0088] In this embodiment, relevant user request behaviors are screened out from the operation log, focusing on the following fields: timestamp: the specific time when the request occurs, request type: distinguish whether it is a read request or a write request, data volume: the size of the data involved in each request (usually in bytes), handle missing values ​​and abnormal values ​​in the log data, ensure the accuracy and completeness of the data, convert the extracted data into a structured format (such as a data frame) for subsequent analysis, organize the extracted user request behavior data into a table, record the detailed information of each request, including timestamp, request type and data volume, determine the method for calculating the energy consumption of hard disk read and write operations, usually including: static energy consumption: the energy consumption of the hard disk in standby mode, dynamic energy consumption: the energy consumption of the hard disk when performing read and write operations, usually related to the data volume and request type, for each user request behavior, add a fixed static energy consumption value during each request, need to determine the power value and operation time according to the hard disk specification or previous energy consumption data, and record the energy consumption calculation result of each request into a data structure, including the timestamp, request type, data volume and calculated energy consumption value of the request.

[0089] Step S3: Identify hard disk read and write operations in multiple time periods for multiple user request behavior data, and perform implicit association learning of read and write logic, thereby generating user read and write logic association features;

[0090] In this embodiment, a data analysis tool is used to load the previously generated multiple user request behavior data to ensure that the data contains information such as timestamp, request type (read / write) and data volume. The entire time range is divided into multiple time periods, such as hours, minutes or other time granularities, so as to analyze the read and write operation behaviors in different time periods. For each time period, the number of read and write operations and the amount of data in it are counted to form a read and write operation data set in the time period. A statistical table is generated to record the number of read requests and write requests in each time period and their total data volume. The read and write operation data of each time period are analyzed to identify significant operation patterns, such as peak time periods, valley time periods, etc. A suitable implicit association learning algorithm is selected, association rule learning: such as Apriori algorithm or FP-Growth algorithm, to identify the association between read and write operations. Sequential pattern mining: such as GSP or SPADE algorithm, to analyze the sequential relationship between operations. Use the selected algorithm to perform association learning on the prepared data set to generate association rules or patterns between read and write operations. Evaluate the generated association rules, filter out rules with high support and confidence, and ensure that they are practical. Organize the filtered rules into user read-write logic association features, which may include: combinations of associated operations, time intervals and frequencies between operations.

[0091] Step S4: dividing the solid-state hard disk drive operation log into time windows based on the user read-write logic association characteristics, and performing behavior evolution prediction to construct multiple potential behavior prediction scenarios;

[0092] In this embodiment, the size of the time window is determined according to the characteristics of user behavior and the analysis target. For example, an hour, day or a custom time period can be selected. Based on the determined time window, the operation log of the solid-state hard disk drive is divided. The data in each time window will be used for subsequent behavior analysis. The operation log in each time window is counted to record the number of read and write operations, data volume, request type and other information in the time period. The statistical results in each time window are organized into a structured data set. According to the data characteristics and prediction targets, a suitable behavior evolution prediction model is selected, for example: Time series model: such as ARIMA, SARIMA Or LSTM, suitable for processing time series data, regression models: such as linear regression or support vector regression, suitable for predicting related numerical values, extract features from the divided time window data set, prepare training data sets, features can include: number of read and write operations, data volume, user read and write logical association features, etc., use the prepared training data set to train the selected prediction model to capture the evolution pattern of user behavior, use the trained model to predict future user behavior, generate expected data of read and write operations in future time windows, evaluate the accuracy and reliability of the prediction, use indicators such as mean square error (MSE) or mean absolute error (MAE) for evaluation, based on the prediction results, identify multiple potential user behavior prediction scenarios, for example, user game behavior scenarios, movie watching behavior scenarios, video editing behavior scenarios, define key features for each potential behavior prediction scenario, including: expected number of read and write operations, expected data volume and type, related user behavior patterns.

[0093] Step S5: simulating the scene energy consumption of multiple potential behavior prediction scenarios according to the energy consumption data of each hard disk read and write operation behavior, and generating scene energy consumption data of all potential behavior prediction scenarios;

[0094] In this embodiment, the calculation model for determining energy consumption simulation generally includes: Static energy consumption: adding a fixed static energy consumption to each scenario. Dynamic energy consumption: calculating the dynamic energy consumption based on the operation type (read / write), data volume and energy consumption data within each potential scenario. For each potential behavior prediction scenario, perform the following steps: Add the static energy consumption to the total energy consumption of the current scenario. Calculate the corresponding dynamic energy consumption based on the number of read and write operations and the amount of data in the scenario, and add the calculated dynamic energy consumption to the static energy consumption to obtain the total energy consumption of the scenario. Organize the energy consumption data of all potential behavior prediction scenarios into an easy-to-understand format, ensuring that it contains information such as scenario identifier, total energy consumption, static energy consumption and dynamic energy consumption. Save the generated scenario energy consumption data in a structured format (such as CSV, Excel or a database) for subsequent analysis and reporting.

[0095] Step S6: Predict the aging state evolution of the scene energy consumption data of all potential behavior prediction scenes, re-estimate the hard disk aging energy consumption, and construct a hard disk scene energy consumption prediction curve.

[0096] In this embodiment, the model used to predict the evolution of aging state is determined. Common methods include: linear regression model: suitable for simple aging trend prediction, time series analysis: such as ARIMA or Exponential Smoothing, used to capture the time dependency of the aging process. Machine learning models such as random forests or neural networks can handle complex nonlinear relationships. Collect historical data related to hard disk aging, which may include: hard disk temperature changes, usage frequency, number of read and write operations, etc. Combine these data with scene energy consumption data to form a training data set. Features include scene energy consumption and aging status indicators. Use the prepared training data set to train the selected model and learn how to predict the hard disk aging status from scene energy consumption data. Use the trained model to predict the aging status evolution of future scene energy consumption data. Output the energy consumption value of each scene at different aging stages. Determine the method for re-estimating aging energy consumption, which usually includes: Considering the impact of aging on hard disk performance and energy consumption. For example, aging may lead to higher energy consumption and reduced read and write efficiency. Based on the aging status prediction results, re-estimate the energy consumption data of each potential behavior prediction scenario. The aging factor can be obtained based on historical data and model predictions to reflect the energy consumption changes of the hard disk under different aging states. Organize the re-estimated energy consumption data into an easy-to-understand format, ensuring that it contains scene identifiers, aging status, and corresponding energy consumption values. Use data visualization tools such as Matplotlib to visualize the energy consumption of each scene. or Tableau) to build energy consumption prediction curves to show the changing trends of scene energy consumption under different aging conditions, analyze the prediction curves, identify potential energy consumption peaks and valleys, help understand the performance of hard disks under different usage and aging conditions, and save the generated hard disk scene energy consumption prediction curves and their data as image files or structured formats (such as CSV) for subsequent reporting and decision support.

[0097] In this embodiment, refer to Figure 2 , is a flowchart of detailed implementation steps of step S1. In this embodiment, the detailed implementation steps of step S1 include:

[0098] Step S11: obtaining a solid state hard disk drive operation log;

[0099] Step S12: Analyze the real-time operation status of the solid state hard disk drive operation log and extract the real-time operation status parameters;

[0100] Step S13: performing time stamp alignment processing on the real-time operating status parameters to obtain time-standardized operating status parameters;

[0101] Step S14: performing real-time temperature calculation on the time-standardized operating state parameters to generate real-time temperature data of the solid state hard disk;

[0102] Step S15: Calculating the power load based on the real-time operating status parameters to obtain power load data;

[0103] Step S16: Perform a comprehensive hard disk energy consumption evaluation on the power load data according to the real-time temperature data of the solid state hard disk to generate the solid state hard disk energy consumption data.

[0104] In this embodiment, the source of the operation log of the solid-state drive is determined, usually through the monitoring tool of the operating system, the software provided by the solid-state drive manufacturer, or the use of a third-party monitoring tool (such as CrystalDiskInfo), the system is configured to enable the SMART (self-monitoring, analysis and reporting technology) function of the solid-state drive to ensure that the operation status and performance data can be recorded, and the operation log is started to be recorded, usually including timestamps, read and write operations, temperature, service life and other information, and the log data is exported regularly or read in real time and saved in CSV or text file format for subsequent analysis. The acquired solid-state drive operation log is loaded using a data analysis library (such as Pandas), the log data is cleaned, invalid or duplicate records are removed, and the accuracy of the data is ensured, and key operation status parameters are extracted from the log, such as: read and write speed: extract read and write speed parameters, I / O operation count: calculate the number of input and output operations in a specific time period, error count: extract any error or warning information, temperature parameters and hard disk function rate load parameters, ensure that the timestamp format extracted from different sources (such as sensors and logs) is consistent, use time parsing methods (such as pd.to_datetime()) for conversion, align real-time operating status parameters according to timestamps, ensure that data at different time points can be compared with each other, use interpolation methods to fill missing data, ensure the continuity of time series, extract temperature-related data from logs, if there is no temperature data in the log, you can use a formula or model to estimate the temperature (such as calculated based on load), if there is sensor data, update the temperature data in real time, calculate the current temperature value, use the output of the temperature sensor to reflect the operating temperature of the SSD in real time, extract voltage and current data according to operating status parameters (such as load), calculate the power load of the SSD in real time, consider the impact of read and write operations and temperature on power, determine the energy consumption calculation formula, usually energy (watt-hours) equals power (watts) multiplied by time (hours), calculate energy consumption data based on real-time power load data and timestamps, generate an energy consumption assessment report for the SSD, analyze the generated energy consumption data, identify energy consumption trends, and present the results in the form of charts or reports to facilitate decision-making and optimization.

[0105] In this embodiment, refer to Figure 3, is a flowchart of detailed implementation steps of step S2. In this embodiment, the detailed implementation steps of step S2 include:

[0106] Step S21: performing user request behavior analysis on the solid state hard disk drive operation log to obtain a plurality of user request behavior data;

[0107] Step S22: classifying the hard disk read and write response types of the multiple user request behavior data, and extracting the hard disk read and write operation response data;

[0108] Step S23: performing time stamp sorting processing on the hard disk read and write operation response data to generate a read and write operation response timing sequence;

[0109] Step S24: performing timestamp matching on the read and write operation response timing sequence according to the time-standardized operating status parameter, thereby obtaining read and write operation response matching data;

[0110] Step S25: Calculate the read and write energy consumption of the read and write operation response matching data based on the energy consumption data of the solid state hard disk, thereby obtaining energy consumption data of each hard disk read and write operation behavior.

[0111] In this embodiment, entries related to user requests are extracted from the log, usually including records of read and write operations, and key fields of the user request are determined, such as: timestamp: the time when the request occurred, operation type: read or write, data block size: the amount of data requested, and invalid or duplicate records are cleaned to ensure data accuracy. The extracted user request data is organized into a structured format (such as a data frame), and classification criteria are determined, such as classification by request type (read / write) and response time. According to the defined criteria, the response time and type of hard disk read and write operations are extracted from the user request behavior data, and conditional statements or machine learning classification algorithms (such as decision trees and random forests) are used to classify read and write operations, and a hard disk read and write response data set is generated. The classified hard disk read and write operation response data is loaded using a data analysis library (such as Pandas), and the timestamp field is ensured to be in a unified format. pd.to_datetime() is used. The function is used to convert the read and write operation response data, sort the read and write operation response data according to the timestamp, generate a response timing sequence arranged in chronological order, use the data analysis tool to load the time-standardized operating status parameter data (such as temperature, power load, etc.), use a suitable method (such as the nearest neighbor method) to match the time-standardized operating status parameters with the read and write operation response timing sequence, find the corresponding operating status parameters by timestamp, create a new data set, including the read and write operation responses and the corresponding operating status parameters, select a suitable energy consumption calculation model, usually based on the product of power and time, for each matching read and write operation, use the energy consumption calculation formula (such as energy = power × time) to calculate the corresponding energy consumption data, and record the calculated energy consumption data of each read and write operation in the result data set to form a complete energy consumption history record.

[0112] In this embodiment, refer to Figure 4 , is a flowchart of detailed implementation steps of step S3. In this embodiment, the detailed implementation steps of step S3 include:

[0113] Step S31: performing time series analysis on multiple user request behavior data to obtain time series request behavior data;

[0114] Step S32: performing user operation logic analysis on the time sequence request behavior data to obtain user behavior logic data;

[0115] Step S33: performing multi-period hard disk read and write operation identification based on the solid state hard disk drive operation log, and generating a multi-stage hard disk read and write track;

[0116] Step S34: performing implicit association learning of read-write logic on multi-stage hard disk read-write traces according to user behavior logic data, thereby generating user read-write logic association features.

[0117] In this embodiment, a data analysis tool (such as Pandas) is used to load the organized user request behavior data. Ensure that the timestamps in the request behavior data are in a unified format and use pd.to_datetime() to convert them for time series analysis. Sort the request behavior data by timestamp to generate time series data and record the user requests at each time point. Determine a suitable logic analysis model (such as state machine, decision tree, association rule analysis) to capture the logical structure of user behavior. Extract features such as request interval, request order and frequency from the time series request behavior data to identify the general behavior pattern of users. Build a user operation logic model based on the extracted features to describe the user's behavior transition under different states. Output user behavior logic data, including a description of each logic state and the probability of state transition. Divide time into multiple time periods (such as hours, days, and weeks) according to requirements to prepare for subsequent analysis. Identify the read and write operations in each time period from the operation log and record information such as the operation type, time, and size. Integrate the identified read and write operations into a multi-stage hard disk read and write track to record the operation sequence of each time period. Use a data analysis tool to load the previously generated user behavior logic data. Select an appropriate association learning algorithm, for example: Apriori algorithm: used to discover frequent item sets and association rules. FP-Growth: used to efficiently discover frequent patterns. Combine user behavior logic data with multi-stage hard disk read and write traces to build a data set containing user behavior and corresponding read and write operations. Apply the selected association learning algorithm to identify the association features between user operations and hard disk read and write operations. Extract implicit logical association rules, for example: a certain user operation usually leads to specific read and write behaviors. When a user copies a file, he usually pastes or sends a file.

[0118] In this embodiment, step S4 includes the following steps:

[0119] Step S41: dividing the solid state hard disk drive operation log into time windows, thereby obtaining a plurality of time window operation cycle data;

[0120] Step S42: Perform user request frequency statistics on multiple time window operation cycle data to generate user request behavior frequency parameters for each cycle;

[0121] Step S43: performing behavior type distribution analysis on the user request behavior frequency parameters of each cycle to obtain user request behavior type distribution data;

[0122] Step S44: Based on the user read-write logic association characteristics, the user request behavior type distribution data is predicted to perform behavior evolution, and multiple potential behavior prediction scenarios are constructed.

[0123] In this embodiment, according to the analysis requirements, the length of the time window is defined (for example, 1 hour, 1 day, 1 week), and the time granularity is determined. The operation log data is divided according to the defined time windows. The pd.Grouper or a custom grouping method is used to distribute the data to each time window. The log data in each time window is summarized to generate a corresponding time window operation cycle data set, which includes the start and end time of each time window and related event statistics. The time window operation cycle data generated previously is loaded using a data analysis tool. For each time window, the total number of user requests is counted. The groupby and count methods are used to perform statistics to generate the user request frequency of each cycle. The user request frequency parameter of each cycle is calculated, which is usually expressed as the number of requests per hour or day. The generated user request behavior frequency parameter is saved in a structured format for subsequent analysis. The user request behavior frequency parameter generated previously is loaded using a data analysis tool. Different behavior types are defined according to the characteristics of the user request, such as: read request: a request for a read operation, write request: a request for a write operation, and mixed request: a request that includes both read and write operations. For the user request frequency parameter of each time window, the frequency distribution of different behavior types is calculated. Value_counts() is used. The method counts the number of requests of each type, organizes the analysis results into a structured format, records the frequency and proportion of different request types in each cycle, saves the user request behavior type distribution data into a structured format (such as CSV or Excel) for subsequent analysis, uses data analysis tools to load the previously generated user read-write logic association feature data, and selects a suitable prediction model, such as: time series prediction model: such as ARIMA or LSTM, suitable for processing time series data, machine learning model: such as random forest or support vector machine (SVM), for classification and regression tasks, uses user request behavior type distribution data and user read-write logic association features as input, trains the selected prediction model to generate predictions of user request behavior, runs the model to generate multiple potential user request behavior prediction scenarios, considers different behavior patterns and request frequencies, and saves the generated potential behavior prediction scenarios into a structured format (such as CSV or JSON) for subsequent analysis and application.

[0124] In this embodiment, the specific steps of step S44 are:

[0125] Performing probability distribution calculation on the user request behavior type distribution data to generate behavior type probability distribution data;

[0126] Based on the user read-write logic association characteristics, the up-down dependency analysis between user behaviors is performed on multiple time window operation cycle data to obtain the up-down dependency relationship between behaviors;

[0127] Based on the up-down dependency relationship between behaviors, the convolution sequence of the behavior type probability distribution data is learned to obtain the distribution law of user behavior;

[0128] Conduct multi-time point behavior simulation based on user behavior distribution rules to obtain multiple behavior simulation trajectories;

[0129] Perform behavioral evolution prediction on multiple behavioral simulation trajectories to obtain multi-time point behavioral evolution prediction data;

[0130] Identify potential behavior scenarios based on multi-time point operation behavior prediction data and construct multiple potential behavior prediction scenarios.

[0131] In this embodiment, a data analysis tool (such as Pandas) is used to load the previously generated user request behavior type distribution data, the number of requests in each time window is summed to obtain the total number of requests, the number of requests for each behavior type is normalized, the probability of each behavior type is calculated, the calculation results are organized into a structured format, the probability of each behavior type is recorded, and the generated behavior type probability distribution data is saved in a structured format (such as CSV or Excel) for subsequent analysis. A Bayesian network, Markov chain or other dependency analysis model is used to analyze the up-down dependency relationship between user request behaviors, extract the conditional probability and transition probability between behaviors, describe the dependency relationship between different request behaviors, record the analysis results, and form a data set of up-down dependency relationships between behaviors for subsequent use. The behavior type probability distribution data is standardized to ensure that the data is suitable for convolution operations, a suitable convolutional neural network (CNN) model is selected, appropriate convolution layers and pooling layers are designed, up-down dependencies between behaviors are used as input features, the convolution model is trained, the distribution law of user behaviors is identified, the distribution law of user behaviors is extracted from the trained model, relevant features and patterns are recorded, and a suitable simulation method is selected, such as Monte Carlo-based Method or random process model, conduct behavior simulation, take the generated user behavior distribution law as input, conduct multi-time point simulation, run the simulation process, generate user behavior simulation trajectories at multiple time points, record the behavior data at each simulation time point, use data analysis tools to load the multiple behavior simulation trajectory data generated before, select a suitable behavior evolution prediction model, such as: Long Short-Term Memory Network (LSTM): for time series prediction, Recurrent Neural Network (RNN): for processing sequence data, use behavior simulation trajectory data to train evolution prediction model, learn the law of behavior evolution, use the trained model to predict future behavior, generate multi-time point behavior evolution prediction data, save the generated multi-time point behavior evolution prediction data in a structured format for subsequent analysis and application, select a suitable scenario recognition method, such as cluster analysis (such as K-means, DBSCAN) or anomaly detection algorithm, input the generated multi-time point behavior evolution prediction data into the scenario recognition algorithm, run the scenario recognition algorithm, identify potential behavior scenarios and abnormal behaviors, record the identified potential behavior scenarios, and generate corresponding reports.

[0132] In this embodiment, the specific steps of step S5 are:

[0133] Step S51: extracting a first potential behavior prediction scenario based on multiple potential behavior prediction scenarios;

[0134] Step S52: Based on the energy consumption data of each hard disk read and write operation behavior, a scenario energy consumption simulation is performed on the first potential behavior prediction scenario to obtain the operation behavior energy consumption data of the first scenario;

[0135] Step S53: performing energy consumption superposition calculation on the operation behavior energy consumption data of the first scene, thereby generating the energy consumption data of the first scene;

[0136] Step S54: Iteratively calculate all potential behavior prediction scenarios to generate scenario energy consumption data for all potential behavior prediction scenarios.

[0137] In this embodiment, a data analysis tool (such as Pandas) is used to load multiple potential behavior prediction scenario data generated previously, and the first potential behavior prediction scenario is extracted from the loaded data. The detailed information of the extracted first scenario (such as behavior type, timestamp, frequency, etc.) is usually saved as a variable through index or conditional selection (such as dataframe.iloc[0]) for subsequent use. The energy consumption data of each hard disk read and write operation behavior generated previously is loaded using the data analysis tool, and the calculation method of energy consumption simulation is determined, which may include: behavior-based energy consumption calculation: using energy consumption data and predicted operation behavior, time weighted method: energy consumption calculation is performed according to the time weight of each operation, each operation in the first potential behavior prediction scenario is traversed, simulation calculation is performed according to the energy consumption data of hard disk read and write operations, the energy consumption of each operation is recorded, and the operation behavior energy consumption data of the first scenario is generated. The obtained operation behavior energy consumption data of the first scenario is saved in a structured format for subsequent analysis. The operation behavior energy consumption data of the first scenario generated previously is loaded using the data analysis tool, and the energy consumption of all operation behaviors in the first scenario is superimposed and calculated. The calculation result is the total energy consumption of the scenario, and the generated energy consumption data of the first scenario is saved in a structured format (such as CSV or JSON) for subsequent analysis and reporting, use data analysis tools to load all previously generated potential behavior prediction scenario data, create a data frame or list to store the energy consumption data of each scenario, and use a loop to traverse each potential behavior prediction scenario: Extract scenario: extract detailed information of the current scenario, simulate energy consumption: perform energy consumption simulation based on hard disk read and write operation energy consumption data, overlay calculation: perform overlay calculation on the simulated energy consumption data to generate the total energy consumption of the scenario, and record the energy consumption data of each scenario in the previously initialized result storage.

[0138] In this embodiment, the specific steps of step S6 are:

[0139] Step S61: performing a time series temperature change analysis on the real-time temperature data of the solid state hard disk to generate temperature change data of the solid state hard disk;

[0140] Step S62: performing load operation fluctuation analysis on the power load data to obtain load operation fluctuation data;

[0141] Step S63: performing aging state evolution prediction based on the solid state hard disk temperature change data and load operation fluctuation data, thereby generating hard disk aging state evolution prediction data;

[0142] Step S64: re-estimating the hard disk aging energy consumption of the scene energy consumption data of all potential behavior prediction scenes based on the hard disk aging state evolution prediction data to obtain the scene energy consumption data under the aging state;

[0143] Step S65: Perform time series energy consumption fitting on the scene energy consumption data in the aging state to construct a hard disk scene energy consumption prediction curve.

[0144] In this embodiment, a data analysis tool (such as Pandas) is used to load the real-time temperature data of the solid-state drive to ensure that the data format is consistent and the timestamp field of the temperature data is in a unified format. The pd.to_datetime() function is used for conversion, and basic statistical indicators of the temperature data, such as the mean, maximum, minimum, and standard deviation, are calculated. Moving average or smoothing technology is used to analyze the temperature change trend, identify temperature fluctuations and change patterns, organize the analysis results into a structured format, record the time series data of temperature changes, and save the generated solid-state drive temperature change data into a structured format (such as CSV or Excel) for subsequent analysis. The power load data is loaded using a data analysis tool to ensure that the data format is consistent and the timestamp field of the power load data is in a unified format. The pd.to_datetime() function is used for conversion. Function is used for conversion to calculate the volatility of power load. Standard deviation, variance and other indicators are usually used to measure the degree of fluctuation of load data. By drawing a time series diagram of load data, the changing trend and fluctuation of load are observed. The analysis results are organized into a structured format. The fluctuation data of power load is recorded. A suitable prediction model is selected, for example: regression model: such as linear regression or polynomial regression, time series model: such as ARIMA or LSTM, which is suitable for processing time series data. Temperature change data and load fluctuation data are used as input to train the selected prediction model to predict the aging state of the hard disk. The trained model is used to predict the future aging state, and the hard disk aging state evolution prediction data is output. The scene energy consumption data of all potential behavior prediction scenarios generated previously are loaded using data analysis tools, and the calculation method of energy consumption re-estimation is determined. It is usually based on the influencing factors of aging state (such as response time, energy efficiency, etc.). Each potential behavior prediction scenario is traversed, and the scene energy consumption is re-estimated according to the aging state evolution prediction data. The scene energy consumption data under the aging state is generated and saved in a structured format (such as CSV). or Excel) for subsequent analysis. Use data analysis tools to load the generated scene energy consumption data under aging status and select a suitable fitting model, such as: polynomial regression: used to fit complex energy consumption change curves; spline regression: used to handle the fitting of nonlinear relationships. Use the energy consumption data under aging status to train the fitting model, generate the energy consumption prediction curve, visualize the fitting results, and generate the hard disk scene energy consumption prediction curve.

[0145] In this embodiment, a solid-state hard disk energy consumption prediction system based on deep learning is provided, which is used to execute the solid-state hard disk energy consumption prediction method based on deep learning as described above, including:

[0146] The hard disk energy consumption evaluation module is used to obtain the solid-state hard disk drive operation log; perform real-time operation status analysis on the solid-state hard disk drive operation log, and perform a comprehensive evaluation of the hard disk energy consumption to generate solid-state hard disk energy consumption data;

[0147] A read / write energy consumption calculation module is used to obtain multiple user request behavior data based on the solid-state hard disk drive operation log, and perform read / write energy consumption calculation based on the solid-state hard disk drive energy consumption data, thereby obtaining energy consumption data for each hard disk drive read / write operation behavior;

[0148] The read-write logic association module is used to identify hard disk read-write operations in multiple time periods based on multiple user request behavior data, and to learn implicit associations of read-write logic, thereby generating user read-write logic association features;

[0149] The behavior prediction scenario module is used to divide the SSD operation log into time windows based on the user read-write logic association characteristics, and to predict the behavior evolution and construct multiple potential behavior prediction scenarios;

[0150] The scene energy consumption module is used to simulate the scene energy consumption of multiple potential behavior prediction scenes according to the energy consumption data of each hard disk read and write operation behavior, and generate the scene energy consumption data of all potential behavior prediction scenes;

[0151] The aging evolution prediction module is used to predict the aging state evolution of the scene energy consumption data of all potential behavior prediction scenarios, re-estimate the hard disk aging energy consumption, and construct the hard disk scene energy consumption prediction curve.

[0152] The present invention can timely understand the operation status of the hard disk by real-time analysis of the hard disk operation log, which is helpful for monitoring the energy consumption of the hard disk. By comprehensively analyzing the hard disk operation log, accurate hard disk energy consumption data can be generated to provide a basis for subsequent energy consumption calculations. Extracting user request behavior data based on the hard disk operation log is helpful for understanding the user's read and write operations on the hard disk. Read and write energy consumption calculations based on the hard disk energy consumption data can evaluate the energy consumption of each read and write operation, providing data support for subsequent behavior predictions. Identifying the hard disk read and write operations in multiple user request behavior data is helpful for understanding user behavior patterns. By learning the implicit correlation features of the read and write logic, the logical connection between user behaviors can be discovered to provide a basis for behavior predictions. Write logic association features to divide time windows and predict behavior evolution, which helps to understand the development trend of user behavior. Constructing multiple potential behavior prediction scenarios can provide different usage scenarios for subsequent energy consumption simulation. By simulating potential behavior prediction scenarios based on the energy consumption data of each read and write operation, the energy consumption of the hard disk under different scenarios can be evaluated. The energy consumption data of all potential behavior prediction scenarios is generated to provide basic data for the subsequent aging state evolution prediction. The aging state evolution prediction of the scenario energy consumption data can predict the energy consumption state of the hard disk under long-term use and provide a reference for energy consumption management. Re-estimation of hard disk aging energy consumption helps to adjust the energy consumption prediction model and construct the hard disk scenario energy consumption prediction curve to more accurately reflect the energy consumption of the hard disk.

[0153] 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.

[0154] The above 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 will conform to the widest scope consistent with the principles and novel features invented herein.

Claims

1. A method for predicting energy consumption of solid-state drives based on deep learning, characterized in that: The following steps are involved: Step S1: obtaining a solid-state hard disk drive operation log; performing real-time operation status analysis on the solid-state hard disk drive operation log, and performing a comprehensive evaluation of the hard disk drive energy consumption to generate solid-state hard disk drive energy consumption data; Step S2: obtaining multiple user request behavior data based on the solid state hard disk drive operation log, and performing read and write energy consumption calculation based on the solid state hard disk drive energy consumption data, thereby obtaining energy consumption data of each hard disk drive read and write operation behavior; Step S3: Identify hard disk read and write operations in multiple time periods for multiple user request behavior data, and perform implicit association learning of read and write logic, thereby generating user read and write logic association features; Step S4: divide the SSD operation log into time windows based on the user read-write logic association characteristics, and perform behavior evolution prediction to construct multiple potential behavior prediction scenarios; the specific steps of step S4 are: Step S41: dividing the solid state hard disk drive operation log into time windows, thereby obtaining a plurality of time window operation cycle data; Step S42: Perform user request frequency statistics on multiple time window operation cycle data to generate user request behavior frequency parameters for each cycle; Step S43: performing behavior type distribution analysis on the user request behavior frequency parameters of each cycle to obtain user request behavior type distribution data; Step S44: predicting the behavior evolution of the user request behavior type distribution data based on the user read-write logic association characteristics, and constructing multiple potential behavior prediction scenarios; The specific steps of step S44 are: Performing probability distribution calculation on the user request behavior type distribution data to generate behavior type probability distribution data; Based on the user read-write logic association characteristics, the up-down dependency analysis between user behaviors is performed on multiple time window operation cycle data to obtain the up-down dependency relationship between behaviors; Based on the up-down dependency relationship between behaviors, the convolution sequence of the behavior type probability distribution data is learned to obtain the distribution law of user behavior; Conduct multi-time point behavior simulation based on user behavior distribution rules to obtain multiple behavior simulation trajectories; Perform behavioral evolution prediction on multiple behavioral simulation trajectories to obtain multi-time point behavioral evolution prediction data; Identify potential behavior scenarios for multi-time point operation behavior prediction data and construct multiple potential behavior prediction scenarios; Step S5: simulating the scene energy consumption of multiple potential behavior prediction scenarios according to the energy consumption data of each hard disk read and write operation behavior, and generating scene energy consumption data of all potential behavior prediction scenarios; Step S6: Predict the aging state evolution of the scene energy consumption data of all potential behavior prediction scenes, re-estimate the hard disk aging energy consumption, and construct a hard disk scene energy consumption prediction curve.

2. The method for predicting energy consumption of a solid-state drive based on deep learning according to claim 1, characterized in that: The specific steps of step S1 are: Step S11: obtaining a solid state hard disk drive operation log; Step S12: Analyze the real-time operation status of the solid state hard disk drive operation log and extract the real-time operation status parameters; Step S13: performing time stamp alignment processing on the real-time operating status parameters to obtain time-standardized operating status parameters; Step S14: performing real-time temperature calculation on the time-standardized operating state parameters to generate real-time temperature data of the solid state hard disk; Step S15: Calculating the power load based on the real-time operating status parameters to obtain power load data; Step S16: Perform a comprehensive hard disk energy consumption evaluation on the power load data according to the real-time temperature data of the solid state hard disk to generate the solid state hard disk energy consumption data.

3. The method for predicting energy consumption of a solid-state drive based on deep learning according to claim 1, characterized in that: The specific steps of step S2 are: Step S21: performing user request behavior analysis on the solid state hard disk drive operation log to obtain a plurality of user request behavior data; Step S22: classifying the hard disk read and write response types of the multiple user request behavior data, and extracting the hard disk read and write operation response data; Step S23: performing time stamp sorting processing on the hard disk read and write operation response data to generate a read and write operation response timing sequence; Step S24: performing timestamp matching on the read and write operation response timing sequence according to the time-standardized operating status parameter, thereby obtaining read and write operation response matching data; Step S25: Calculate the read and write energy consumption of the read and write operation response matching data based on the energy consumption data of the solid state hard disk, thereby obtaining energy consumption data of each hard disk read and write operation behavior.

4. The method for predicting energy consumption of a solid-state drive based on deep learning according to claim 1, characterized in that: The specific steps of step S3 are: Step S31: performing time series analysis on multiple user request behavior data to obtain time series request behavior data; Step S32: performing user operation logic analysis on the time sequence request behavior data to obtain user behavior logic data; Step S33: performing multi-period hard disk read and write operation identification based on the solid state hard disk drive operation log, and generating a multi-stage hard disk read and write track; Step S34: performing implicit association learning of read-write logic on multi-stage hard disk read-write traces according to user behavior logic data, thereby generating user read-write logic association features.

5. The method for predicting energy consumption of a solid state drive based on deep learning according to claim 1, characterized in that: The specific steps of step S5 are: Step S51: extracting a first potential behavior prediction scenario based on multiple potential behavior prediction scenarios; Step S52: Based on the energy consumption data of each hard disk read and write operation behavior, a scenario energy consumption simulation is performed on the first potential behavior prediction scenario to obtain the operation behavior energy consumption data of the first scenario; Step S53: performing energy consumption superposition calculation on the operation behavior energy consumption data of the first scene, thereby generating the energy consumption data of the first scene; Step S54: Iteratively calculate all potential behavior prediction scenarios to generate scenario energy consumption data for all potential behavior prediction scenarios.

6. The method for predicting energy consumption of a solid-state drive based on deep learning according to claim 1, characterized in that: The specific steps of step S6 are: Step S61: performing a time series temperature change analysis on the real-time temperature data of the solid state hard disk to generate temperature change data of the solid state hard disk; Step S62: performing load operation fluctuation analysis on the power load data to obtain load operation fluctuation data; Step S63: performing aging state evolution prediction based on the solid state hard disk temperature change data and load operation fluctuation data, thereby generating hard disk aging state evolution prediction data; Step S64: re-estimating the hard disk aging energy consumption of the scene energy consumption data of all potential behavior prediction scenes based on the hard disk aging state evolution prediction data to obtain the scene energy consumption data under the aging state; Step S65: Perform time series energy consumption fitting on the scene energy consumption data in the aging state to construct a hard disk scene energy consumption prediction curve.

7. A solid state drive energy consumption prediction system based on deep learning, characterized in that: The method for predicting energy consumption of a solid-state drive based on deep learning as claimed in claim 1 comprises: The hard disk energy consumption evaluation module is used to obtain the solid-state hard disk drive operation log; perform real-time operation status analysis on the solid-state hard disk drive operation log, and perform a comprehensive evaluation of the hard disk energy consumption to generate solid-state hard disk energy consumption data; A read / write energy consumption calculation module is used to obtain multiple user request behavior data based on the solid-state hard disk drive operation log, and perform read / write energy consumption calculation based on the solid-state hard disk drive energy consumption data, thereby obtaining energy consumption data for each hard disk drive read / write operation behavior; The read-write logic association module is used to identify hard disk read-write operations in multiple time periods based on multiple user request behavior data, and to learn implicit associations of read-write logic, thereby generating user read-write logic association features; The behavior prediction scenario module is used to divide the SSD operation log into time windows based on the user read-write logic association characteristics, and to predict the behavior evolution and construct multiple potential behavior prediction scenarios; The scene energy consumption module is used to simulate the scene energy consumption of multiple potential behavior prediction scenes according to the energy consumption data of each hard disk read and write operation behavior, and generate the scene energy consumption data of all potential behavior prediction scenes; The aging evolution prediction module is used to predict the aging state evolution of the scene energy consumption data of all potential behavior prediction scenarios, re-estimate the hard disk aging energy consumption, and construct the hard disk scene energy consumption prediction curve.

Images