A behavior-aware and prediction-based storage optimization system and method

Abstract

The application discloses a storage optimization system and method based on behavior sensing and prediction, relates to the field of solid state disk writing modes, and adjusts the writing strategy of a storage medium by sensing and predicting the running state of a user equipment, and comprises successively connected multi-modal behavior data acquisition modules, a user behavior prediction engine, a dynamic writing strategy optimization module and a self-adaptive learning and feedback module, the output end of the self-adaptive learning and feedback module is connected back to the user behavior prediction engine, the multi-modal behavior data acquisition modules are used for acquiring the running state and behavior information of the user equipment, and are used for analyzing the location information, the network connection state and the running parameters of the user equipment, thereby forming historical behavior data; the user behavior prediction engine constructs a prediction model based on the historical behavior data, infers the operation trend of the user in the future, and forms a prediction result; and the dynamic writing strategy optimization module is used for regulating and controlling the writing strategy of the storage medium according to the prediction result.

Description

Technical Field

[0001] This invention relates to the field of solid-state drive writing methods, and in particular to a storage optimization system and method based on behavior awareness and prediction. Background Technology

[0002] With the increasing prevalence of SSDs in consumer electronics, mobile devices, and enterprise applications, users have increasingly higher demands for storage performance and device lifespan. Existing SSD optimization technologies mainly focus on storage cell characteristics, environmental parameters, or data characteristics, while neglecting the dynamic impact of user behavior on SSD write patterns. When users frequently operate on certain files within a specific time period, SSD write operations occur in concentrated bursts, leading to increased write amplification, performance degradation, and shortened lifespan.

[0003] Therefore, a new way is needed to analyze users' multi-dimensional operational behavior in real time and predict the actions users are about to take. Summary of the Invention

[0004] In view of the above-mentioned shortcomings in the current solid-state drive writing methods, the present invention provides a storage optimization system and method based on behavior perception and prediction, which can dynamically analyze user operating habits, automatically adjust writing strategies according to prediction results, effectively extend the life of SSDs, and significantly improve system response speed by avoiding unnecessary writing operations.

[0005] To achieve the above objectives, the embodiments of the present invention adopt the following technical solutions: A storage optimization system based on behavior perception and prediction adjusts the write strategy of the storage medium by perceiving and predicting the operating status of user devices. The system comprises a multimodal behavior data acquisition module, a user behavior prediction engine, a dynamic write strategy optimization module, and an adaptive learning and feedback module, connected sequentially. The output of the adaptive learning and feedback module is connected back to the user behavior prediction engine. The multimodal behavior data acquisition module acquires the operating status and behavior information of the user device, and analyzes it in conjunction with the user device's location information, network connection status, and operating parameters to form historical behavior data. The user behavior prediction engine constructs a prediction model based on the historical behavior data, infers the user's subsequent operation trends, and generates a prediction result. The dynamic write strategy optimization module adjusts the write strategy of the storage medium according to the prediction result.

[0006] According to one aspect of the present invention, the adaptive learning and feedback module is used to evaluate the execution result of the writing strategy by continuously monitoring the writing effect and operating status of the storage medium.

[0007] According to one aspect of the present invention, when the adaptive learning and feedback module detects a change in system performance or response, it adjusts the prediction model and control parameters through a feedback mechanism.

[0008] According to one aspect of the present invention, in multi-device application scenarios, the adaptive learning and feedback module performs collaborative analysis of the behavior patterns of each terminal through distributed learning, thereby improving the predictive model's capabilities without directly exchanging raw data.

[0009] According to one aspect of the present invention, the multimodal behavior data acquisition module extracts only necessary feature information during the data acquisition process and processes user data through a privacy protection mechanism.

[0010] According to one aspect of the present invention, the operating status and behavior information of the user device includes at least keyboard input frequency, mouse movement trajectory, application usage duration, and file access type.

[0011] According to one aspect of the present invention, the user behavior prediction engine identifies the type of operation that a user is about to perform and potential data access needs by analyzing historical behavior data, and further determines the data types of high-frequency access and the time intervals in which they occur, providing a basis for adjusting the writing strategy.

[0012] According to one aspect of the present invention, the dynamic write strategy optimization module selects the write location based on data access characteristics, and prioritizes the allocation of frequently accessed data to high-performance storage areas.

[0013] According to one aspect of the present invention, the dynamic write strategy optimization module dynamically adjusts the execution timing of internal maintenance operations based on the user's activity status, completing the processing within a period of less user operation, thereby reducing the impact on response performance.

[0014] A storage optimization method based on behavior awareness and prediction adjusts the write strategy of storage media by sensing and predicting the operating status of user devices, including the following steps: Acquire the operating status and behavior information of user equipment, and analyze it in conjunction with the user equipment's location information, network connection status and operating parameters to form historical behavior data; A predictive model is built based on historical behavioral data to infer the user's subsequent operational trends and generate prediction results; The writing strategy of the storage medium is adjusted based on the prediction results.

[0015] The advantages of this invention are as follows: Through multimodal behavior perception and prediction technology, intelligent optimization of the storage system is achieved. It can dynamically analyze user operating habits and automatically adjust write strategies based on prediction results, effectively extending the lifespan of SSDs. Simultaneously, by avoiding unnecessary write operations, it significantly improves system response speed, ensuring a smooth experience even during high-frequency operations. Frequently used data is automatically placed in high-speed storage areas, optimizing overall access efficiency and reducing device maintenance needs. Differential privacy technology is employed, collecting only necessary behavioral characteristics to protect user privacy and security. It can autonomously learn user habits, achieving continuous optimization without manual intervention, perfectly matching storage performance with user needs. Attached Figure Description

[0016] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0017] Figure 1 This is a schematic diagram of the system modules of a storage optimization system based on behavior perception and prediction according to the present invention; Figure 2 This is a schematic diagram of the structure of a storage optimization system based on behavior perception and prediction according to the present invention; Figure 3 This is a flowchart illustrating a storage optimization method based on behavior perception and prediction as described in this invention. Detailed Implementation

[0018] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0019] Example 1 like Figure 1As shown, a storage optimization system based on behavior perception and prediction monitors the operating status of user devices in real time through a system interface and a lightweight sensing mechanism to adjust the write strategy of the storage medium. It includes a multimodal behavior data acquisition module, a user behavior prediction engine, a dynamic write strategy optimization module, and an adaptive learning and feedback module. The output of the multimodal behavior data acquisition module is connected to the input of the user behavior prediction engine, the output of the user behavior prediction engine is connected to the input of the dynamic write strategy optimization module, the output of the dynamic write strategy optimization module is connected to the adaptive learning and feedback module, and the output of the adaptive learning and feedback module is connected back to the user behavior prediction engine.

[0020] The multimodal behavior data acquisition module is used to acquire the operating status and behavior information of user devices. It also combines this information with contextual data such as the user device's location, network connection status, and device operating parameters for comprehensive analysis to form historical behavior data. The operating status and behavior information of the user device includes at least multi-dimensional behavior information such as keyboard input frequency, mouse movement trajectory, application usage duration, and file access type. During data acquisition, the multimodal behavior data acquisition module extracts only necessary feature information and processes user data through privacy protection mechanisms, thereby minimizing the impact on user privacy while ensuring system functionality.

[0021] The user behavior prediction engine constructs a prediction model based on historical behavior data to infer user operation trends in the near future and generate prediction results. Through analysis of historical behavior data, the engine can identify the types of operations a user is about to perform and potential data access needs, and further determine the data types accessed frequently and their likely time intervals, thus providing a basis for adjusting subsequent write strategies.

[0022] The dynamic write strategy optimization module is used to adjust the write strategy of the storage medium based on the prediction results. This module can select the write location based on data access characteristics, prioritizing the allocation of frequently accessed data to higher-performance storage areas. Simultaneously, the module dynamically adjusts the execution timing of internal maintenance operations based on user activity levels, completing processing during periods of low user activity to minimize the impact on response performance. Furthermore, differentiated write strategies can be employed for different types of data to improve overall storage efficiency and reduce unnecessary resource consumption.

[0023] The adaptive learning and feedback module continuously monitors the write performance and operational status of the storage medium to evaluate the execution results of the write strategy. When the adaptive learning and feedback module detects changes in system performance or response, it adjusts the prediction model and control parameters through a feedback mechanism, enabling the system to gradually adapt to different users' habits. The adaptive learning and feedback module also establishes a correlation between user behavior characteristics and system performance, continuously optimizing prediction accuracy and control strategies through long-term data accumulation. In multi-device application scenarios, distributed learning can be used to collaboratively analyze the behavior patterns of each terminal, improving the prediction model's capabilities without directly exchanging raw data, thereby further enhancing the overall system performance and adaptability.

[0024] The advantages of this invention are as follows: Through multimodal behavior perception and prediction technology, intelligent optimization of the storage system is achieved. It can dynamically analyze user operating habits and automatically adjust write strategies based on prediction results, effectively extending the lifespan of SSDs. Simultaneously, by avoiding unnecessary write operations, it significantly improves system response speed, ensuring a smooth experience even during high-frequency operations. Frequently used data is automatically placed in high-speed storage areas, optimizing overall access efficiency and reducing device maintenance needs. Differential privacy technology is employed, collecting only necessary behavioral characteristics to protect user privacy and security. It can autonomously learn user habits, achieving continuous optimization without manual intervention, perfectly matching storage performance with user needs.

[0025] Example 2 like Figure 2 As shown, a storage optimization system based on behavior perception and prediction is applied to a personal computer terminal equipped with a solid-state drive (SSD), an Intel i7 series processor, 16GB of memory, and a Windows 11 operating system. The system includes a multimodal behavior data acquisition module, a user behavior prediction engine, a dynamic write strategy optimization module, and an adaptive learning and feedback module. These modules interact with each other via a system bus.

[0026] The multimodal behavior data acquisition module is used to collect user behavior and device status in real time, with a sampling period of 1 second. The collected data includes: keyboard input frequency f_key (unit: times / second, range 0–15 times / second), mouse movement distance d_mouse and number of clicks, application foreground runtime t_app (e.g., Word, browser, video player, etc.), file access type and access frequency type_file (e.g., .docx, .mp4, .png), and device context information state_device, including GPS positioning (Wi-Fi positioning is used indoors), network status (Wi-Fi / wired / offline), battery level (0–100%), and device temperature (30–85℃). After feature extraction, the collected data forms a behavior vector B(t)=[f_key, d_mouse, t_app, type_file, state_device]. A differential privacy mechanism (ε=1.0) is used to add noise to the behavior data, retaining only statistical features for subsequent analysis.

[0027] The user behavior prediction engine comprises a single-layer LSTM network with 64 hidden layers for extracting time-series features, and a Transformer encoder with two attention heads for modeling long-range dependencies. Its input is a sequence of user behavior over the past 30 minutes, with a time window of 1800 sampling points. The output includes predictions of user actions over the next 5–30 minutes, such as a probability of 0.82 for opening a document editing software, 0.10 for playing a video, and 0.08 for being idle, as well as predictions of file writing behavior, with high-frequency writing types being .docx and .xlsx, and an estimated probability of writing within the next 10 minutes of 0.76.

[0028] The dynamic write strategy optimization module adjusts the SSD control strategy based on prediction results, including write location optimization, garbage collection scheduling, and write amplification control. In write location optimization, data predicted to be frequently accessed is written to the SSD's high-speed SLC cache area, while low-frequency data is written to the QLC area, thereby improving access speed. In garbage collection scheduling, garbage collection is triggered when the predicted user activity level for the next 10 minutes is below a threshold of 0.2; otherwise, execution is delayed to reduce system lag. In write amplification control, small block writes (4KB) are used for document files to reduce write amplification, while sequential large block writes (1MB) are used for video files to improve throughput efficiency.

[0029] The adaptive learning and feedback module is used to evaluate system performance and optimize strategies, including real-time monitoring of write latency, system response time, and SSD write amplification factor. When performance degradation is detected, such as an increase in write latency exceeding 20%, model parameters are automatically adjusted, for example, updating LSTM weights or adjusting the prediction time window from 30 minutes to 20 minutes. Simultaneously, a mapping relationship between behavior and performance is established: Performance = f(Behavior, Strategy). In multi-device scenarios, model gradients are shared through federated learning without transmitting raw data, achieving cross-device optimization.

[0030] The system using this embodiment can dynamically optimize the SSD write strategy based on user behavior, thereby improving system response performance and reducing write amplification of storage devices.

[0031] Example 3 like Figure 3 As shown, a storage optimization method based on behavior perception and prediction is implemented based on the storage optimization system based on behavior perception and prediction described in Embodiment 1. This method adjusts the write strategy of the storage medium by perceiving and predicting the operating status of user devices. It includes a multimodal behavior data acquisition module, a user behavior prediction engine, a dynamic write strategy optimization module, and an adaptive learning and feedback module. The output of the multimodal behavior data acquisition module is connected to the input of the user behavior prediction engine, the output of the user behavior prediction engine is connected to the input of the dynamic write strategy optimization module, the output of the dynamic write strategy optimization module is connected to the adaptive learning and feedback module, and the output of the adaptive learning and feedback module is connected back to the user behavior prediction engine.

[0032] It also includes the following steps: S1: Acquire the operating status and behavior information of user equipment, and analyze it in conjunction with the user equipment's location information, network connection status and operating parameters to form historical behavior data.

[0033] The multimodal behavior data acquisition module is used to acquire the operating status and behavior information of user devices. It also combines this information with contextual data such as the user device's location, network connection status, and device operating parameters for comprehensive analysis to form historical behavior data. The operating status and behavior information of the user device includes at least multi-dimensional behavior information such as keyboard input frequency, mouse movement trajectory, application usage duration, and file access type. During data acquisition, the multimodal behavior data acquisition module extracts only necessary feature information and processes user data through privacy protection mechanisms, thereby minimizing the impact on user privacy while ensuring system functionality.

[0034] S2: Based on historical behavior data, a predictive model is built to infer the user's subsequent operational trends and generate prediction results.

[0035] The user behavior prediction engine constructs a prediction model based on historical behavior data to infer user operation trends in the near future and generate prediction results. Through analysis of historical behavior data, the engine can identify the types of operations a user is about to perform and potential data access needs, and further determine the data types accessed frequently and their likely time intervals, thus providing a basis for adjusting subsequent write strategies.

[0036] S3: Adjust the writing strategy of the storage medium based on the prediction results.

[0037] The dynamic write strategy optimization module is used to adjust the write strategy of the storage medium based on the prediction results. This module can select the write location based on data access characteristics, prioritizing the allocation of frequently accessed data to higher-performance storage areas. Simultaneously, the module dynamically adjusts the execution timing of internal maintenance operations based on user activity levels, completing processing during periods of low user activity to minimize the impact on response performance. Furthermore, differentiated write strategies can be employed for different types of data to improve overall storage efficiency and reduce unnecessary resource consumption.

[0038] The adaptive learning and feedback module continuously monitors the write performance and operational status of the storage medium to evaluate the execution results of the write strategy. When the adaptive learning and feedback module detects changes in system performance or response, it adjusts the prediction model and control parameters through a feedback mechanism, enabling the system to gradually adapt to different users' habits. The adaptive learning and feedback module also establishes a correlation between user behavior characteristics and system performance, continuously optimizing prediction accuracy and control strategies through long-term data accumulation. In multi-device application scenarios, distributed learning can be used to collaboratively analyze the behavior patterns of each terminal, improving the prediction model's capabilities without directly exchanging raw data, thereby further enhancing the overall system performance and adaptability.

[0039] The advantages of this invention are as follows: Through multimodal behavior perception and prediction technology, intelligent optimization of the storage system is achieved. It can dynamically analyze user operating habits and automatically adjust write strategies based on prediction results, effectively extending the lifespan of SSDs. Simultaneously, by avoiding unnecessary write operations, it significantly improves system response speed, ensuring a smooth experience even during high-frequency operations. Frequently used data is automatically placed in high-speed storage areas, optimizing overall access efficiency and reducing device maintenance needs. Differential privacy technology is employed, collecting only necessary behavioral characteristics to protect user privacy and security. It can autonomously learn user habits, achieving continuous optimization without manual intervention, perfectly matching storage performance with user needs.

[0040] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A storage optimization system based on behavior perception and prediction, which adjusts the write strategy of storage media by perceiving and predicting the operating status of user equipment, characterized in that, The system comprises a multimodal behavior data acquisition module, a user behavior prediction engine, a dynamic write strategy optimization module, and an adaptive learning and feedback module, connected in sequence. The output of the adaptive learning and feedback module is connected back to the user behavior prediction engine. The multimodal behavior data acquisition module acquires the operating status and behavior information of the user device, and analyzes it in conjunction with the user device's location information, network connection status, and operating parameters to form historical behavior data. The user behavior prediction engine builds a prediction model based on the historical behavior data, infers the user's subsequent operation trends, and forms a prediction result. The dynamic write strategy optimization module adjusts the write strategy of the storage medium according to the prediction result.

2. The storage optimization system based on behavior perception and prediction according to claim 1, characterized in that, The adaptive learning and feedback module is used to evaluate the execution results of the writing strategy by continuously monitoring the writing effect and operating status of the storage medium.

3. The storage optimization system based on behavior perception and prediction according to claim 2, characterized in that, When the adaptive learning and feedback module detects changes in system performance or response, it adjusts the prediction model and control parameters through a feedback mechanism.

4. The storage optimization system based on behavior perception and prediction according to claim 1, characterized in that, In multi-device application scenarios, the adaptive learning and feedback module performs collaborative analysis of the behavior patterns of each terminal through distributed learning, thereby improving the predictive model's capabilities without directly exchanging raw data.

5. The storage optimization system based on behavior perception and prediction according to claim 1, characterized in that, During the data acquisition process, the multimodal behavior data acquisition module extracts only the necessary feature information and processes the user data through a privacy protection mechanism.

6. The storage optimization system based on behavior perception and prediction according to claim 1, characterized in that, The operating status and behavior information of the user device includes at least keyboard input frequency, mouse movement trajectory, application usage duration, and file access type.

7. The storage optimization system based on behavior perception and prediction according to claim 1, characterized in that, The user behavior prediction engine analyzes historical behavior data to identify the types of operations a user is about to perform and potential data access needs. It further determines the data types accessed frequently and the time intervals in which they occur, providing a basis for adjusting the writing strategy.

8. The storage optimization system based on behavior perception and prediction according to any one of claims 1 to 7, characterized in that, The dynamic write strategy optimization module selects the write location based on data access characteristics, prioritizing the allocation of frequently accessed data to high-performance storage areas.

9. The storage optimization system based on behavior perception and prediction according to claim 8, characterized in that, The dynamic write strategy optimization module dynamically adjusts the execution timing of internal maintenance operations based on user activity levels, completing processing during periods with fewer user operations to reduce the impact on response performance.

10. A storage optimization method based on behavior perception and prediction, implemented based on the storage optimization system based on behavior perception and prediction as described in claim 1, wherein the method adjusts the write strategy of the storage medium by perceiving and predicting the operating status of the user equipment, characterized in that... Includes the following steps: Acquire the operating status and behavior information of user equipment, and analyze it in conjunction with the user equipment's location information, network connection status and operating parameters to form historical behavior data; A predictive model is built based on historical behavioral data to infer the user's subsequent operational trends and generate prediction results; The writing strategy of the storage medium is adjusted based on the prediction results.

Images