Data information storage method based on artificial intelligence

Abstract

The invention discloses a data information storage method based on artificial intelligence, and relates to the technical field of data storage, and the method comprises the following steps: distributing a basic compression ratio for each type of data according to the types and expected access modes of different data; in the storage process after data compression, the access condition of each kind of data is tracked and collected in real time through a data access monitoring tool, and a dynamic analysis set is established according to the real-time monitored data access condition to reflect the change of the data access frequency. According to the invention, through intelligent compression ratio dynamic regulation and control, the decompression burden is reduced during high-frequency access, and the system response speed is improved; the compression ratio is increased during low-frequency access, utilization of storage resources is maximized, the problems of storage space waste and performance bottleneck caused by unified compression ratio are solved, efficient management of the storage resources is achieved, efficient storage and rapid access of data are guaranteed, and the overall performance and user experience of the storage system are further improved.

Description

Technical Field

[0001] The present invention relates to the technical field of data storage, and particularly to a data information storage method based on artificial intelligence. Background Art

[0002] Data information storage based on artificial intelligence refers to introducing artificial intelligence technology on the basis of traditional data storage systems to intelligently optimize processes such as data classification, compression, retrieval, and scheduling, thereby improving the efficiency, flexibility, and security of data storage. This method can not only automatically select the optimal storage strategy and medium (such as hot / cold storage, edge / cloud storage) according to the data type, access frequency, and importance level, but also perform semantic analysis and label annotation on structured and unstructured data with the help of deep learning algorithms to achieve efficient intelligent indexing and fast query. At the same time, AI can also monitor the running state of the storage system in real time, predict faults, and automatically perform fault tolerance and migration, significantly enhancing the stability and sustainability of the data system.

[0003] Data compression plays a core optimization role in the data storage process. Its main purpose is to perform encoding conversion on the original data through algorithms, reduce the storage space occupied by redundant information, and thus significantly reduce the consumption of storage resources and transmission costs. Compression can not only improve the utilization rate of storage media, but also reduce the bandwidth burden during data transmission and improve transmission efficiency.

[0004] The prior art has the following deficiencies: In the prior art, in order to simplify data storage management, a unified compression ratio is usually used for data compression. The main purpose of this approach is to maintain consistency and simplicity in management in a large-scale storage environment, and ensure stability and predictability during data storage, transmission, and access through a unified compression ratio, thereby simplifying the design and implementation of compression strategies. However, due to the fact that the access frequency of data usually changes dynamically, using a unified compression ratio for data compression may lead to the following problems: For data with a low access frequency, the unified low compression ratio cannot fully compress the data, resulting in suboptimal utilization of storage space. Although low-frequency data is not frequently accessed, its storage requirements can still achieve higher space utilization through a higher compression ratio. However, the unified compression ratio fails to flexibly adapt to this requirement, resulting in waste of storage space.

[0005] For data with a high access frequency, the unified high compression ratio may cause performance bottlenecks. During each access, a higher compression ratio requires more decompression operations, thereby significantly increasing the computational and time overheads. Although this compression method can save storage space, it reduces the data access efficiency, especially in application scenarios with high requirements for real-time performance and response speed, which may affect the overall performance and response ability of the system.

[0006] Therefore, the practice of using a unified compression ratio in the prior art fails to effectively address the demand differences brought about by changes in data access frequency, resulting in suboptimal storage efficiency and access performance.

[0007] The above information disclosed in the background section is only used to enhance the understanding of the background of the present disclosure. Therefore, it may include information that does not constitute the prior art known to those of ordinary skill in the art. Summary of the Invention

[0008] The object of the present invention is to provide an artificial intelligence-based data information storage method. Through intelligent dynamic regulation of the compression ratio, it can adjust the compression strategy in real time according to changes in data access frequency, thereby reducing the decompression burden during high-frequency access and improving the system response speed; increasing the compression ratio during low-frequency access to maximize the utilization of storage resources. This compression mechanism that flexibly adapts to data access patterns not only solves the problems of wasted storage space and performance bottlenecks caused by a unified compression ratio but also realizes the efficient management of storage resources, ensuring the efficient storage and rapid access of data, further enhancing the overall performance of the storage system and the user experience, so as to solve the problems in the above background technology.

[0009] To achieve the above object, the present invention provides the following technical solution: An artificial intelligence-based data information storage method, comprising the following steps: According to the types and expected access patterns of different data, first assign a basic compression ratio to each type of data to ensure that the data is effectively compressed before storage; During the storage process after data compression, use a data access monitoring tool to track and collect the access situation of each type of data in real time, and establish a dynamic analysis set based on the real-time monitored data access situation to reflect changes in data access frequency; Extract high-frequency access features from the analysis set. Within the monitoring window, after deeply analyzing the extracted high-frequency access features, identify the change pattern of data access frequency and provide accurate input for the subsequent prediction model; Input the features after in-depth analysis as feature vectors into a pre-trained machine learning model to predict the trend and changes in data access frequency for a period of time in the future; Through a fuzzy logic algorithm, dynamically adjust the compression ratio of the data according to the access frequency prediction result, so that the compression ratio can intelligently adapt to different data access patterns, optimize the utilization rate of storage space, and improve data access performance.

[0010] Preferably, high-frequency access features are extracted from the analysis set. The extracted features include the degree of intensive access to data and the degree of convergence of the time intervals between two consecutive accesses to data. Within the monitoring window, after in-depth analysis of the extracted high-frequency access features, an access concentration index and an access interval compression index are respectively generated. The variation law of data access frequency is identified through the access concentration index and the access interval compression index, and precise inputs are provided for the subsequent prediction model.

[0011] Preferably, the access concentration index and the access interval compression index after in-depth analysis are used as feature vectors and input into a pre-trained machine learning model. An access evaluation coefficient is generated through the machine learning model, and based on the access evaluation coefficient, the data access frequency trend and changes in the future period are predicted.

[0012] Preferably, according to the access frequency prediction result, the compression ratio of data is dynamically adjusted through a fuzzy logic algorithm. The specific steps are as follows: Based on the access evaluation coefficient output by the access prediction model, combined with the set access frequency reference threshold, the heat level of the current data access frequency is judged through a fuzzy logic function, and the compression adjustment factor is calculated accordingly. The calculation expression of the fuzzy logic function is as follows: In the formula: is the access evaluation coefficient, output by the machine learning model, representing the prediction intensity of the future access frequency of the current data. The larger the value, the higher the access frequency; is the access evaluation reference threshold, used to distinguish the boundary between "high access frequency" and "low access frequency", usually determined according to historical statistical data; is the adjustment sensitivity parameter, controlling the slope of the output of the fuzzy function, making the compression adjustment more gentle or sensitive to access fluctuations; The hyperbolic tangent function is used to realize the non-linear change of the compression adjustment factor within the interval of -1 and 1, making the compression ratio adjustment show a continuous and smooth adjustment trend.

[0013] Preferably, after obtaining the compression adjustment factor, the actual applied compression ratio is updated in combination with the originally set basic compression ratio. The update expression is as follows: In the formula: is the basic compression ratio, determined according to the data type, initial setting or historical static analysis results; is the compression ratio adjustment amplitude coefficient, used to balance the actual adjustment amplitude of the adjustment factor and avoid excessive changes in the compression ratio; 、 are the allowable minimum and maximum compression ratio boundary values, used to constrain the reasonable range of the actual compression ratio; For the compression ratio finally applied to the data, it dynamically responds to the change in access frequency to balance performance and storage.

[0014] Preferably, within the monitoring window, the specific steps for generating the access concentration index after deeply analyzing the degree of intensive access to the data are as follows: First, divide the monitoring window into several time periods, and statistically process the access frequency of the target stored data for each time period. To quantify the concentration contribution of this time period to the overall access structure, an exponential weighted function is introduced, and the weighted function is defined as: , where: represents the access frequency of the data in the time period ; i is the index variable of the time period; represents the weighted value of this time period in the composition of the access concentration; is the total number of accesses within the monitoring window.

[0015] Preferably, on the basis of obtaining the access weight distribution values of all time periods, construct an access concentration index to quantify the access concentration trend of the target data within the monitoring window. The access concentration index adopts the following calculation formula: , where: is the access concentration index; represents the maximum access weighted value in each time period; is a constant offset term, usually taking a value of , to prevent the denominator from being zero; n represents the total number of time periods within the monitoring window.

[0016] Preferably, within the monitoring window, the specific steps for generating the access interval compression index after deeply analyzing the convergence degree of the time interval between two consecutive accesses to the data are as follows: Within the set monitoring window, record the time point sequence when the target data is accessed in real time, N represents the total number of the time point sequence, and accordingly construct an interval sequence of adjacent access times to form an access time interval vector , where, ; Based on the access time interval vector, construct a time interval compression gradient vector for measuring the change trend of the access time interval, and the definition of each component is as follows: , where, represents the time interval compression gradient of the kth time interval, represents the relative change rate between the kth time interval and the next time interval, and is a quantization index for whether the access frequency tends to be stable.

[0017] Preferably, after obtaining the compressed gradient vector, the access interval compression index is calculated in combination with the overall fluctuation range of the access time interval to measure whether the access behavior within the monitoring window has a high-frequency tendency. The calculation expression is as follows: , where is the access interval compression index; is a regulation factor used to adjust the sensitivity of access behavior fluctuations; represents the maximum value in the access time interval compression gradient, reflecting the interval change with the strongest access instability; is the global span of the access time interval, used to measure the concentration degree of access behavior in the time distribution, where , and respectively represent the maximum value and the minimum value of all access time intervals.

[0018] In the above technical solution, the technical effects and advantages provided by the present invention are as follows: Through intelligent dynamic regulation of the compression ratio, the present invention can adjust the compression strategy in real time according to the change of data access frequency, thereby reducing the decompression burden during high-frequency access and improving the system response speed; increasing the compression ratio during low-frequency access to maximize the utilization of storage resources. This compression mechanism that flexibly adapts to the data access pattern not only solves the problems of storage space waste and performance bottleneck caused by a unified compression ratio, but also realizes the efficient management of storage resources, ensures the efficient storage and fast access of data, and further improves the overall performance and user experience of the storage system. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings required in the embodiments. Obviously, the drawings described below are only some embodiments recorded in the present invention. For those of ordinary skill in the art, other drawings can also be obtained based on these drawings.

[0020] Figure 1 is a mind map of a data information storage method based on artificial intelligence according to the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0021] Now, the exemplary embodiments will be described more comprehensively with reference to the accompanying drawings. However, the exemplary embodiments can be implemented in various forms and should not be construed as limited to the examples set forth herein; rather, these exemplary embodiments are provided so that the present disclosure will be more complete and comprehensive, and will fully convey the concept of the exemplary embodiments to those skilled in the art.

[0022] The present invention provides as Figure 1A data information storage method based on artificial intelligence is shown as follows, including the following steps: According to the types and expected access patterns of different data, first assign a basic compression ratio to each type of data to ensure that the data is effectively compressed before storage; This compression ratio is determined based on the characteristics of the data (such as file type, data structure, storage requirements, etc.) and the predicted access frequency. For example, data with low access frequency (such as archived data, backup data, etc.) can be set with a higher compression ratio to save storage space, while data with high access frequency (such as real-time data, popular files, etc.) should be set with a lower compression ratio to improve data reading performance. The setting of the basic compression ratio helps to avoid waste of storage space in the initial storage stage and provides a starting point for subsequent intelligent adjustment. The role of this step is to assign a reasonable basic compression ratio to the data according to the data type and the preliminary estimated access frequency, ensuring that the initial storage requirements of different data types can be met during compressed storage.

[0023] During the storage process after data compression, use a data access monitoring tool to track and collect the access situation of each type of data in real time, and establish a dynamic analysis set based on the real-time monitored data access situation to reflect the changes in data access frequency; Use a data access monitoring tool to track and collect the access situation of each type of data in real time. Usually, data logging and event-driven mechanisms are adopted. Specifically, the monitoring tool will be integrated into the storage system to record each data access event in real time, including information such as data reading, writing, modification, access frequency, timestamp, and visitor identification. To achieve efficient real-time tracking, the monitoring tool can use the cache mechanism in memory to quickly store access events and synchronize these data to the database or log management system regularly. The access tool can also generate alerts or reports automatically according to specific access patterns or thresholds through the event trigger mechanism for analyzing the data access behavior. In addition, the tool can display the access statistical information of the data through a visual interface and provide a real-time access data query function through the API interface for subsequent feature extraction and analysis. This real-time tracking and data collection method ensures that the access situation of each type of data can be comprehensively and accurately monitored.

[0024] Extract high-frequency access features from the analysis set. Within the monitoring window, conduct in-depth analysis on the extracted high-frequency access features to identify the change rules of data access frequency and provide accurate input for the subsequent prediction model; Extract high-frequency access features from the analysis set. Among them, the extracted features include the degree of intensive access to data and the degree of convergence of the time interval between two consecutive accesses to data. After deeply analyzing the extracted high-frequency access features within the monitoring window, an access concentration index and an access interval compression index are generated respectively. By using the access concentration index and the access interval compression index, the change law of data access frequency is identified, and precise input is provided for the subsequent prediction model.

[0025] A relatively high degree of intensive access to data usually indicates that the data is called or read multiple times per unit time, reflecting its high access frequency. Intensive access means that within a short time window, the data is repeatedly accessed by multiple users, processes, or system tasks, manifested as a high concentration of access requests on the time axis. This high-density access usually occurs in hot data, key configuration files, shared resources, or core data relied on by multiple tasks. By identifying intensive access behavior, the activity level of data in the current or upcoming time period can be accurately judged, effectively reflecting its usage frequency and importance in the business. Therefore, the degree of intensive access can be used as an important characteristic index to measure the high-frequency access state of data.

[0026] The specific steps for generating the access concentration index after deeply analyzing the degree of intensive access to data within the monitoring window are as follows: First, divide the monitoring window into several time periods, and statistically process the access frequency of the target stored data in each time period. In order to quantify the contribution of this time period to the overall access structure, an exponential weighted function is introduced, and the weighted function is defined as: , where: represents the access frequency of the data in the time period ; i is the index variable of the time period; represents the weighted value of this time period in the composition of the access concentration; is the total number of accesses within the monitoring window; Through the setting of this weighted function, the intensity of access behavior in the time dimension can be strengthened, so that the time periods with prominent access have a more significant weight expression in the overall distribution.

[0027] On the basis of obtaining the access weight distribution values of all time periods, construct an access concentration index to quantify the access concentration trend of the target data within the monitoring window. The access concentration index adopts the following calculation formula: , where: is the access concentration index; represents the maximum access weighted value in each time period; is a constant offset term, usually taking the value of , to prevent the denominator from being zero; n represents the total number of time periods within the monitoring window; By squaring the highest access weight, the response ability to local access burst behavior is significantly enhanced. At the same time, by performing a normalization ratio process on this peak value and the overall weighted total, it is possible to effectively depict whether there is a significant concentration phenomenon in data access.

[0028] As can be seen from the access concentration index, within the monitoring window, the larger the performance value of the access concentration index generated after in-depth analysis of the degree of intensive access to the data, the higher the frequency of access to the data, and vice versa, which indicates a lower frequency of access to the data. The reason is that the access concentration index essentially measures the "aggregation" and "intensity" of access requests in the time distribution. The access concentration index not only considers the frequency of data access in a certain time period, but also strengthens the influence degree of the access concentration peak on the access concentration index (through square amplification processing). When the number of accesses in a certain time period is much higher than that in other time periods, the weight value of this time period will be significantly amplified, resulting in a significant increase in the access concentration index. This peak-concentrated access pattern is often a typical feature of data being frequently called or hotly accessed; on the contrary, if the access behavior is evenly distributed in each time period or the overall access frequency is low, the access concentration index is small, indicating that the data does not have the characteristics of intensive access. Therefore, the access concentration index can effectively reflect the change of data access density per unit time, and thus serve as an important basis for high-frequency access identification.

[0029] When the time interval between two consecutive accesses to a certain data shows a high degree of convergence, that is, the time interval tends to be stable and the interval is short, it usually indicates that the data is in a high-frequency access state. Specifically, if a certain data is accessed multiple times within a period of time, and the time intervals between these accesses gradually converge to a small fixed value, it indicates that the access behavior of the data is regular and continuous. This phenomenon is common in periodic tasks, background polling, or data resources accessed concurrently by multiple users. A high degree of convergence not only reflects the stability of the access time interval, but also indirectly indicates that the data is requested at a high frequency per unit time. Therefore, it can be used as one of the important features for judging whether the data is in a high-frequency access state. In the intelligent compression strategy, this feature can be used as an auxiliary judgment index for high-frequency access identification.

[0030] Within the monitoring window, the specific steps for generating the access interval compression index after in-depth analysis of the convergence degree of the time interval between two consecutive accesses to the data are as follows: Within the set monitoring window, the time point sequence of the target data being accessed is recorded in real time , where N represents the total number of the time point sequence, and based on this, an interval sequence of adjacent access times is constructed to form an access time interval vector , where, ; Based on the access time interval vector, construct a time interval compression gradient vector for measuring the changing trend of the access time interval. The definition of each component is as follows: , where represents the time interval compression gradient of the k-th time interval, represents the relative change rate between the k-th time interval and the next time interval, and is a quantitative index for whether the access frequency tends to be stable; The technical effect of this step is that by constructing a normalized compression gradient vector, the change amplitude between consecutive access time intervals can be accurately reflected, thereby determining whether the data access shows a convergent growth trend, providing a basis for dynamic behavior characteristics for subsequent access frequency recognition and compression regulation.

[0031] After obtaining the compression gradient vector, combine the overall fluctuation range of the access time interval to calculate the access interval compression index to measure whether the access behavior within the monitoring window has a high-frequency tendency. The calculation expression is as follows: , where is the access interval compression index; is the regulation factor used to adjust the sensitivity of the access behavior fluctuation; represents the maximum value in the access time interval compression gradient, reflecting the change in the interval with the strongest access instability; is the global span of the access time interval, used to measure the concentration degree of the access behavior in the time distribution, where , and represent the maximum value and the minimum value of all access time intervals respectively; This step can effectively identify the periodic high-frequency change trend in the access behavior by combining the evaluation of the maximum access time interval compression gradient and the interval range, and introducing an exponential function to perform non-linear amplification processing on it, improving the sensitive recognition ability for sudden hotspot data.

[0032] From the access interval compression metric, it can be seen that within the monitoring window, the larger the performance value of the access interval compression metric generated after deeply analyzing the convergence degree of the time interval between two consecutive accesses to the data, the higher the frequency of access to the data. Conversely, a smaller value indicates a lower frequency of access to the data. The reason is that the essence of the access interval compression metric is to quantify the frequency and stability of data access by analyzing the convergence degree of the time interval between two consecutive accesses to the data within the monitoring window. A larger performance value of the access interval compression metric indicates that the fluctuations in the access interval sequence are smaller, that is, the time intervals tend to be stable and concentrated, showing a continuous and regular access pattern. This is usually a typical feature of high-frequency accessed data. For example, timed reads, polling operations, or hot data requests will all form a trend of interval convergence. Conversely, if the value of the access interval compression metric is small, it means that the access intervals fluctuate violently and are unstable, and the time of data access is irregular, with a relatively low frequency. Therefore, the access interval compression metric can be used as an effective metric to judge whether the data is in a high-frequency access state. The larger its value, the higher the data access frequency it reflects.

[0033] Input the features after in-depth analysis as feature vectors into a pre-trained machine learning model to predict the data access frequency trend and changes in the future for a period of time; Input the access concentration metric and the access interval compression metric after in-depth analysis as feature vectors into a pre-trained machine learning model, generate an access evaluation coefficient through the machine learning model, and predict the data access frequency trend and changes in the future for a period of time based on the access evaluation coefficient.

[0034] A pre-trained machine learning model refers to an intelligent prediction model that has been trained, parameter-optimized, and model-validated based on historical data and specific task objectives before being formally applied to the target task. In this scenario, the core task of this model is to receive the feature vectors of the "access concentration metric" and the "access interval compression metric" as inputs and output an "access evaluation coefficient" that can quantify the future data access trend, which is then used to dynamically adjust the data compression strategy. The so-called "pre-trained" means that the model does not learn from scratch immediately in the current task, but through the learning process of a large amount of historical access data, it already has strong generalization ability and prediction ability, so as to quickly and accurately evaluate new data access patterns.

[0035] In the application context of data storage compression management, pre-trained machine learning models usually include algorithm frameworks such as Support Vector Regression (SVR), Random Forest (RF), Gradient Boosting Decision Tree (GBDT), Long Short-Term Memory Network (LSTM), etc. These models can model the complex non-linear mapping relationship between input features and target outputs, so as to achieve the prediction of future data access behaviors. For example, the access concentration index reflects the density of access distribution within a unit time window, while the access interval compression index reflects the compactness of the time interval between access events. The two jointly characterize the current access state of the data. The model learns the internal relationship between these indices and the actual access frequency during the training phase, and in subsequent actual applications, when new data is input, instead of re-learning, the input vector is directly substituted into the model to calculate the "access evaluation coefficient" representing the current access trend.

[0036] In the pre-training phase of the model, it is first necessary to construct a high-quality training dataset. This dataset generally includes access log records of multiple data objects, their access characteristics within a specific time period, and the corresponding actual access frequencies. Through supervised learning, the mapping relationship between these characteristics and access frequencies is modeled, and the model will optimize its internal weight parameters during continuous iteration, making the prediction results as close as possible to the actual observed values. Key steps in the training process also include cross-validation, overfitting control, feature importance evaluation, etc. These technical means can improve the generalization ability of the model under different data distributions, ensuring a high prediction accuracy when facing new access patterns.

[0037] After the model training is completed, a "pre-trained machine learning model" is formed. It is essentially a function structure with prediction ability, which can map any set of vectors that conform to the input feature format into an output value, which is the access evaluation coefficient here. This access evaluation coefficient can not only be regarded as the quantitative result of access activity, but also be used for dynamic decision support. For example, adjust the compression ratio according to the evaluation coefficient: if the evaluation coefficient is high, it is predicted that future accesses will be frequent, and the compression ratio needs to be reduced to improve access speed; if the evaluation coefficient is low, the compression ratio can be increased to save storage resources.

[0038] It is worth mentioning that the pre-training of the model not only improves the system's adaptability to large-scale data changes, but also significantly reduces the consumption of computing resources during system operation. Since the model has completed the learning process, only forward inference operations are required in the deployment stage to obtain prediction results, which has higher operation efficiency and response speed compared to performing full calculations or real-time statistics each time. From the perspective of system design, the pre-trained model can be deployed on the data management module or edge nodes to achieve lightweight prediction of access behavior. In addition, the model can also be updated periodically through an incremental learning strategy, that is, after a certain amount of new access data has been accumulated, the model parameters are re-trained or fine-tuned to maintain a high degree of consistency between the model and the latest data patterns. This mechanism further enhances the long-term effectiveness and adaptive ability of the model.

[0039] In summary, the pre-trained machine learning model plays a core role in intelligent compression control. It realizes the effective prediction of future access states by learning the mapping relationship between access features and access trends.

[0040] The machine learning model is not specifically limited here. Any machine learning model that can comprehensively analyze the access concentration index and the access interval compression index to generate an access evaluation coefficient is acceptable. To implement the technical solution of the present invention, the present invention provides a specific implementation method; the expression for generating the access evaluation coefficient is: , where , are the preset proportionality coefficients of the access concentration index and the access interval compression index respectively, and , are both greater than 0.

[0041] The preset proportionality coefficient refers to the parameter used to balance the influence weights of different access feature indicators on the final evaluation result in the calculation formula of the access evaluation coefficient. Specifically, they are and . They respectively correspond to the relative importance or contribution degree of the access concentration index and the access interval compression index in the evaluation model. The main purpose of setting these two proportionality coefficients is to reasonably allocate the influence of different features in the synthetic evaluation value according to experience, actual business needs, or model tuning results. For example, if the system pays more attention to concentrated access behavior in a short period of time, can be set to enhance the weight of the access concentration index . On the contrary, if more emphasis is placed on the compression trend of access time distribution, The proportion. Their "preset" nature means that these coefficients are usually set through experiments, historical data fitting, or domain knowledge before model training or system deployment, rather than being dynamically learned during real-time calculations. Ultimately, the two preset proportional coefficients jointly determine the numerical range and variation trend of the access evaluation coefficient mapped after the access feature synthesis.

[0042] From the access evaluation coefficient, it can be seen that within the monitoring window, the larger the performance value of the access concentration index generated by deeply analyzing the degree of intensive access to the data, and the larger the performance value of the access interval compression index generated by deeply analyzing the convergence degree of the time interval between two consecutive accesses to the data. That is, the larger the performance value of the access evaluation coefficient generated when predicting the data access frequency within a future period through a pre-trained machine learning model, the higher the access frequency of the data, and vice versa, indicating a lower access frequency of the data.

[0043] According to the access frequency prediction result, the compression ratio of the data is dynamically adjusted through the fuzzy logic algorithm, so that the compression ratio can adapt to different data access patterns, optimize the utilization rate of the storage space, and improve the data access performance; According to the access frequency prediction result, the compression ratio of the data is dynamically adjusted through the fuzzy logic algorithm. The specific steps are as follows: Based on the access evaluation coefficient output by the access prediction model, combined with the set access frequency reference threshold, the heat level of the current data access frequency is judged through the fuzzy logic function, and the compression adjustment factor is calculated accordingly , and the calculation expression of the fuzzy logic function is as follows: , where: is the access evaluation coefficient, output by the machine learning model, representing the prediction intensity of the future access frequency of the current data. The larger the value, the higher the access frequency; is the access evaluation reference threshold, used to distinguish the boundary between "high access frequency" and "low access frequency", usually determined according to historical statistical data; is the adjustment sensitivity parameter, controlling the slope of the output of the fuzzy function, making the compression adjustment more gentle or sensitive to access fluctuations; The hyperbolic tangent function is used to realize the non-linear change of the compression adjustment factor within the range of -1 and 1, making the compression ratio adjustment show a continuous and smooth adjustment trend; This step models the difference between the access evaluation coefficient and the access evaluation reference threshold through the fuzzy logic function to generate the compression adjustment factor . When the access evaluation coefficient is significantly higher than the access evaluation reference threshold, the compression adjustment factor , indicating that the compression ratio should be reduced to improve access performance; when the access evaluation coefficient is much lower than the access evaluation reference threshold, the compression adjustment factor is positive, indicating that the compression ratio can be increased to optimize the utilization of storage resources.

[0044] After obtaining the compression adjustment factor, the compression ratio actually applied is updated in combination with the originally set basic compression ratio. The update expression is as follows: , where: is the basic compression ratio, which is determined according to the data type, initial setting or historical static analysis results; is the compression ratio adjustment amplitude coefficient, which is used to balance the actual adjustment amplitude of the adjustment factor and avoid excessive changes in the compression ratio; 、 are the allowed minimum and maximum compression ratio boundary values, which are used to constrain the reasonable range of the actual compression ratio; is the compression ratio finally applied to the data, which dynamically responds to changes in the access frequency to balance performance and storage; This step dynamically adjusts the compression ratio according to the access status, so that the system automatically reduces the compression intensity when the data access heat is high, improving the decompression efficiency and response speed; when the access frequency is low, it automatically increases the compression ratio to save storage space. The upper and lower limit control ensures the safety of the adjustment process and the stability of the system.

[0045] The compression ratio of the data is dynamically adjusted according to the access frequency prediction result through the fuzzy logic algorithm. The main function is to realize the adaptive optimization of the compression strategy to better balance data storage and access performance. When the data access frequency is high, the compression ratio will be appropriately reduced to avoid frequent decompression operations, thereby improving the access speed and system response ability; when the data access frequency is low, the compression ratio will be increased to save storage space and optimize resource utilization. This dynamic adjustment mechanism can intelligently respond to different data access patterns, not only improving the efficiency of the storage system, but also ensuring the fast access of data and good system performance.

[0046] Through the above artificial intelligence-based data information storage method, the utilization rate of storage space can be significantly optimized, and the data access performance can be improved. This method can dynamically adjust the compression strategy in real time according to the change of data access frequency through intelligent compression ratio control, so as to reduce the decompression burden during high-frequency access and improve the system response speed; increase the compression ratio during low-frequency access to maximize the utilization of storage resources. This compression mechanism that flexibly adapts to data access patterns not only solves the problems of storage space waste and performance bottleneck caused by a unified compression ratio, but also realizes the efficient management of storage resources, ensures the efficient storage and fast access of data, and further improves the overall performance and user experience of the storage system.

[0047] To verify the effectiveness of the proposed artificial intelligence-based data information storage method in dynamic compression regulation, a comparative experimental platform was constructed, and a typical mixed access load dataset was selected for experiments, including three types of data: log files (low-frequency access), hot directory documents (high-frequency access), and picture resources (medium-frequency access). The experimental environment was deployed on a distributed file system with a storage capacity of 32TB. The performance of the unified compression ratio strategy (control group) and the AI-based dynamic compression regulation strategy (experimental group) was compared, and the following key performance indicators were mainly investigated: storage space utilization, data access latency, system throughput, and CPU load.

[0048] The experimental results show that within a 30-day data operation cycle: In terms of storage space utilization, the experimental group saved approximately 22.8% of the total storage space on average compared to the control group. Among them, the compression ratio of log data with low-frequency access increased by approximately 41.3%. In terms of data access latency, the average read latency of high-frequency access data in the experimental group decreased by 27.6%. The response time during the peak access period decreased more significantly, with the optimal decrease reaching 33.2%. The system throughput increased by 18.4% in the experimental group. Especially during the peak period of concurrent access requests, good read-write stability was maintained. The CPU load balance showed good performance. Due to the use of the fuzzy logic algorithm to balance the compression intensity and decompression burden, the overall CPU occupancy volatility in the experimental group decreased by 12.5%.

[0049] In addition, the evaluation of the access prediction accuracy shows that the machine learning model constructed based on the access concentration index and the access interval compression index reached an average prediction accuracy of 93.1%, providing a stable basis for subsequent compression adjustment. The above experimental results fully prove that this solution has significant advantages in improving the intelligence, efficiency, and dynamic adaptability of the storage system.

[0050] The above formulas are all dimensionless and take their numerical values for calculation. The formulas are obtained by collecting a large amount of data for software simulation to obtain a formula closest to the actual situation. The preset parameters in the formulas are set by those skilled in the art according to the actual situation.

[0051] Only some exemplary embodiments of the present invention have been described above by way of illustration. Undoubtedly, for those of ordinary skill in the art, the described embodiments can be modified in various different ways without departing from the spirit and scope of the present invention. Therefore, the above drawings and descriptions are illustrative in nature and should not be construed as limiting the protection scope of the claims of the present invention.

[0052] Those of ordinary skill in the art will appreciate that the units and algorithm steps of each example described in connection with the embodiments disclosed herein can be implemented in electronic hardware, or in a combination of computer software and electronic hardware. Whether these functions are executed in hardware or software depends on the specific application and design constraints of the technical solution. A person skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered to exceed the scope of this application.

[0053] Those skilled in the art can clearly understand that for the sake of convenience and brevity of description, the specific working processes of the systems, devices, and units described above can refer to the corresponding processes in the foregoing method embodiments, and will not be elaborated herein.

[0054] In the several embodiments provided in this application, it should be understood that the disclosed systems and methods can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For example, the division of the units is only a logical function division. In actual implementation, there may be other division methods. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the displayed or discussed couplings or direct couplings or communication connections to each other can be through some interfaces. The indirect couplings or communication connections of the devices or units can be in electrical, mechanical, or other forms.

[0055] The units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or may be distributed to multiple network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0056] In addition, the functional units in the various embodiments of this application can be integrated in a processing unit, or each unit can exist physically alone, or two or more units can be integrated in one unit.

[0057] When the above-mentioned functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art or a part of this technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in various embodiments of this application. The aforementioned storage medium includes: various media such as USB flash drives, mobile hard disks, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical discs that can store program codes.

[0058] As described above, the above are only specific implementation manners of this application, but the protection scope of this application is not limited thereto. Any person skilled in the art within the technical scope disclosed by this application can easily think of changes or substitutions, which should all be covered by the protection scope of this application. Therefore, the protection scope of this application should be subject to the protection scope of the claims.

Claims

1. A data information storage method based on artificial intelligence, characterized in that: The following steps are involved: Assign a basic compression ratio to each type of data based on the data type and expected access pattern; During the storage process after data compression, the data access monitoring tool is used to track and collect the access status of each type of data in real time, and a dynamic analysis set is established based on the real-time monitoring of data access status to reflect the changes in data access frequency; Extract high-frequency access features from the analysis set, and perform in-depth analysis on the extracted high-frequency access features within the monitoring window to identify the changing patterns of data access frequency; The features after deep analysis are input into a pre-trained machine learning model as feature vectors to predict the trend and change of data access frequency in the future; The fuzzy logic algorithm is used to dynamically adjust the data compression ratio based on the access frequency prediction results, so that the compression ratio can intelligently adapt to different data access modes, optimize the utilization of storage space, and improve data access performance.

2. The data information storage method based on artificial intelligence according to claim 1 is characterized in that: High-frequency access features are extracted from the analysis set, where the extracted features include the degree to which the data is intensively accessed and the degree of convergence of the time interval between two consecutive data accesses. Within the monitoring window, after in-depth analysis of the extracted high-frequency access features, access concentration indicators and access interval compression indicators are generated respectively. The access concentration indicators and access interval compression indicators are used to identify the changing patterns of data access frequency and provide accurate input for subsequent prediction models.

3. The data information storage method based on artificial intelligence according to claim 2 is characterized in that: The access concentration index and access interval compression index after deep analysis are input into a pre-trained machine learning model as feature vectors. The access evaluation coefficient is generated by the machine learning model, and the data access frequency trend and changes in the future are predicted based on the access evaluation coefficient.

4. The data information storage method based on artificial intelligence according to claim 3 is characterized in that: The fuzzy logic algorithm is used to dynamically adjust the data compression ratio based on the access frequency prediction results. The specific steps are as follows: Based on the access evaluation coefficient output by the access prediction model and the set access frequency reference threshold, the fuzzy logic function is used to determine the heat level of the current data access frequency, and the compression adjustment factor is calculated accordingly. , the calculation expression of the fuzzy logic function is as follows: , where: The access evaluation coefficient is output by the machine learning model and represents the predicted strength of the future access frequency of the current data; assessing reference thresholds for access; To adjust the sensitivity parameter, control the slope of the fuzzy function output.

5. The data information storage method based on artificial intelligence according to claim 4 is characterized in that: After obtaining the compression adjustment factor, the actual compression ratio of the application is updated in combination with the originally set basic compression ratio. The updated expression is as follows: , where: is the basic compression ratio; is the compression ratio adjustment amplitude coefficient, which is used to balance the actual adjustment amplitude of the adjustment factor; , The minimum and maximum compression ratio boundary values ​​allowed are used to constrain the range of the actual compression ratio; The compression ratio that is ultimately applied to the data dynamically responds to changes in access frequency.

6. The data information storage method based on artificial intelligence according to claim 2 is characterized in that: The specific steps to generate the access concentration index after in-depth analysis of the degree of intensive data access within the monitoring window are as follows: The monitoring window is divided into several time periods, and the access frequency of the target storage data in each time period is statistically processed. In order to quantify the concentration contribution of the time period to the overall access structure, an exponential weighting function is introduced. The weighting function is defined as: ,in: Indicates data in time period The access frequency in the time period; i is the index variable of the time period; Indicates the time period Weighted value in the composition of access concentration; is the total number of visits within the monitoring window.

7. The data information storage method based on artificial intelligence according to claim 6 is characterized in that: Based on the access weight distribution values ​​of all time periods, an access concentration index is constructed to quantify the access concentration trend of the target data within the monitoring window. The access concentration index is calculated using the following formula: ,in: To access the concentration index; Indicates the maximum access weighted value in each time period; is a constant offset term; n represents the total number of time periods in the monitoring window.

8. The data information storage method based on artificial intelligence according to claim 2 is characterized in that: The specific steps for generating the access interval compression index after in-depth analysis of the convergence degree of the time interval between two consecutive data accesses within the monitoring window are as follows: Within the set monitoring window, the time sequence of the target data being accessed is recorded in real time , N represents the total number of time point sequences, and based on this, the interval sequence of adjacent access times is constructed to form the access time interval vector ,in, ; Based on the access time interval vector, a time interval compression gradient vector is constructed to measure the access time interval change trend. Each component of the vector is defined as follows: ,in, The time interval compression gradient of the kth time interval represents the relative rate of change between the kth time interval and the next time interval.

9. The data information storage method based on artificial intelligence according to claim 8, characterized in that: After obtaining the compression gradient vector, the access interval compression index is calculated in combination with the overall fluctuation range of the access time interval. The calculation expression is as follows: ,in, Compression index for access interval; is a regulatory factor used to adjust the sensitivity of visit behavior fluctuations; represents the maximum value in the access time interval compression gradient; is the global span of the access time interval, which is used to measure the concentration of access behavior in time distribution. , and Respectively represent the maximum and minimum values ​​of all access time intervals.

Images