Backup data prediction method, system and report generation method based on artificial intelligence

Abstract

The present invention relates to a backup data prediction method, system and report generation method based on artificial intelligence, which belongs to the field of disaster recovery backup. The backup data prediction method comprises: collecting data from a backup system, standardizing and fusing it into a data set, which includes task and system status information, taking the time period of the backup task as the reference period, and matching the most recent system status data; and then dividing the data set for training a linear regression model to predict future data. The system comprises: a data acquisition module, a data standardization module, a data set definition module, a data segmentation module, a model training module, and an output prediction data module. The present invention can accurately predict disk utilization, avoid insufficient or wasted disk resources, and at the same time improve the management effect of backup operations by realizing the prediction of backup task time and data volume.

Description

Technical Field

[0001] The present invention belongs to the field of disaster recovery and backup, and relates to a backup data prediction method, system and report generation method based on artificial intelligence. Background Art

[0002] With the rapid development of information technology, the amount of data in enterprises and institutions has increased dramatically, and data backup is an important means to help protect data security and integrity.

[0003] At present, in order to better complete the data backup task, it is necessary to make certain predictions on the backup data. For example, the prior art discloses a data backup method, a data backup system, an electronic device and a readable storage medium (application number CN202110497593.7). This method obtains relevant information of the backup task (such as the amount and type of backup data) and backup status information (such as the network status and storage container status of the target storage container), inputs them into the prediction model, generates a backup strategy, and executes the backup task according to the strategy. It can dynamically adjust the backup strategy according to real-time data to reduce the server load and improve backup efficiency. However, this technology only collects information related to the backup task, and the training data is not comprehensive. The prediction accuracy of the subsequent system load, resource utilization and allocation is prone to deviations.

[0004] Therefore, how to more accurately predict backup data and improve resource utilization and allocation efficiency is a technical problem that urgently needs to be solved. Summary of the invention

[0005] In order to solve the technical problems in the above-mentioned background technology, the present invention provides a backup data prediction method, system and report generation method based on artificial intelligence.

[0006] The technical solution of the present invention to solve the above technical problems is as follows:

[0007] In a first aspect, a backup data prediction method based on artificial intelligence is provided, the method comprising the steps of:

[0008] Obtain backup task data and system status data from the backup system;

[0009] Standardize backup task data and system status data;

[0010] The standardized backup task data and system status data are defined and merged into a data set. The data set structure includes task name, task start time, task end time, task data volume, system status time, system available space, number of system running tasks, and system load. The fusion is defined to take the time period of the backup task as the benchmark segment. If there is a single system status data in the benchmark segment, it is directly merged. If there are multiple system status data in the benchmark segment, the arithmetic mean of the system status data in the benchmark segment is taken for fusion. If there is no system status data in the benchmark segment, the next system status data adjacent to the benchmark segment is first taken for fusion, otherwise the previous system status data adjacent to the benchmark segment is taken for fusion.

[0011] Segment the data set to obtain training data;

[0012] Input the training data into the training model, and obtain the prediction model through training. The training model is a linear regression model.

[0013] Predict continuous data through predictive model output.

[0014] In a second aspect, a backup data prediction system based on artificial intelligence is provided, the system comprising:

[0015] A data acquisition module is used to acquire backup task data and system status data from the backup system;

[0016] Data standardization module, used to standardize backup task data and system status data;

[0017] The data set definition module is used to define and merge the standardized backup task data and system status data into one data set. The data set structure includes the task name, task start time, task end time, task data volume, system status time, system available space, number of system running tasks, and system load. The definition of fusion is based on the time period of the backup task as the reference period. If there is a single system status data in the reference period, it is directly merged. If there are multiple system status data in the reference period, the arithmetic mean of the system status data in the reference period is taken for fusion. If there is no system status data in the reference period, the next system status data adjacent to the reference period is first taken for fusion, otherwise the previous system status data adjacent to the reference period is taken for fusion.

[0018] The data segmentation module is used to segment the data set to obtain training data;

[0019] The model training module is used to input the training data into the training model and obtain the prediction model through training. The training model is a linear regression model.

[0020] The output prediction data module is used to output predicted continuous data through the prediction model.

[0021] According to a third aspect, a report generation method is provided for generating a report based on the predicted data obtained by the above-mentioned artificial intelligence-based backup data prediction method and existing data, wherein the report includes recommendations on disk utilization and task intervals.

[0022] The beneficial effects of the present invention are:

[0023] The backup task data and system status data are defined and integrated in a preset manner, which greatly eliminates the impact of uncertain time period matching and time period mismatch in the collection of backup task data and system status data, and ensures the consistency and integrity of the training data set. The prediction model trained in this way can more accurately predict disk utilization, backup task time and data volume, thereby laying the foundation for providing users with more accurate backup task reports and suggestions. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0025] Figure 1 The figure is a flowchart of a backup data prediction method based on artificial intelligence in an embodiment of the present invention.

[0026] Figure 2 This is an example diagram of data set fusion definition in an embodiment of the present invention.

[0027] Figure 3 This is a graph showing the final accuracy results of the segmentation data comparison experiment in an embodiment of the present invention.

[0028] Figure 4 This is a line graph comparing the accuracy rates at different data segmentation ratios in an embodiment of the present invention.

[0029] Figure 5 This is a working process diagram of the gradient descent of model training in an embodiment of the present invention.

[0030] Figure 6 This is a schematic diagram of the iterative optimization principle of the gradient descent algorithm for model training in an embodiment of the present invention.

[0031] Figure 7 Schematic diagram of the structure of a backup data prediction system based on artificial intelligence in an embodiment of the present invention.

[0032] Figure 8 Schematic diagram of the structure of the data acquisition module in an embodiment of the present invention.

[0033] Fig. 9 Schematic diagram of the structure of the model training module in an embodiment of the present invention.

[0034] Fig.10 It is a schematic diagram of the execution of tasks with prediction trends in an embodiment of the present invention.

[0035] In the accompanying drawings, the components represented by the reference numerals are listed as follows:

[0036] 3000, data acquisition module; 3001, data standardization module; 3002, data set definition module; 3003, data segmentation module; 3004, model training module; 3005, testing module; 3006, evaluation module; 3007, output prediction data module; 30001, backup task data acquisition unit; 30002, system status data acquisition unit; 30041, initialization unit; 30042, training unit; 30043, difference comparison unit; 30044, update unit; 30045, repeated tuning unit. DETAILED DESCRIPTION

[0037] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0038] A backup system is an information system used for data backup and recovery, consisting of a backup server, a media server, and backup media. The data copies created and stored by the backup system can enable the production system to continue to operate normally when the original data is lost, damaged, or deleted.

[0039] With the rapid development of information technology, the amount of data in enterprises and institutions has increased dramatically, and data backup has become an important means to help protect data security and integrity. However, the current data backup system generally has the following problems:

[0040] (1) Inaccurate prediction of disk utilization: Existing backup systems often have difficulty accurately predicting disk utilization, which leads to insufficient disk space or waste of resources. Inaccurate predictions may also result in the inability to store necessary data at critical moments or waste of storage resources during off-peak hours;

[0041] (2) The running time of the backup task cannot be accurately grasped: The running time of the backup task is crucial to ensure timely data backup. However, existing backup systems often cannot accurately predict the running time of the backup task, which makes the management and execution of the backup plan vulnerable to interference, affecting the timely backup of data;

[0042] (3) The amount of backup data is difficult to estimate: The estimation of the amount of backup data is crucial for the reasonable allocation of bandwidth and storage resources. However, existing backup systems often have difficulty accurately estimating the amount of backup data, which may lead to unreasonable allocation of bandwidth and storage resources during the backup process, affecting backup efficiency and data recovery speed.

[0043] In view of the above problems, an embodiment of the present invention provides a backup data prediction method based on artificial intelligence. Figure 1 A flowchart of a backup data prediction method based on artificial intelligence provided by an embodiment of the present invention is shown in FIG. Figure 1 As shown, the method includes:

[0044] Step S101, acquiring backup task data and system status data from a backup system.

[0045] It is understandable that the data source of the backup system is the database. The database is a system for storing and managing data. It usually contains multiple tables, each of which has a specific structure and data type.

[0046] The backup task table (tb_task) is a table in the database that is used to store backup task related information, such as task name, task name, next task run time, and task status. The backup task table is usually used to record the configuration and scheduling information of the backup task so that the backup system can perform the backup operation according to the predetermined plan. It is also usually used to trigger and record the backup operation.

[0047] The backup data table (tb_task_data) is also a table in the database, used to store actual backup data records, which may include copies of files, database records or other important information, such as the data name, task name, task start time, task end time and task data volume required in this embodiment.

[0048] The system status table (tb_system_status) is another table in the database, which is used to record the system's operating status information and can provide context information during system operation, such as the record name, status time, used disk space, number of running tasks, and system load required in this embodiment.

[0049] It should be noted that the backup task table contains relevant information of the backup task. After the task is completed, the backup result data will be output to the backup data table. Therefore, this embodiment directly obtains the backup task data from the backup data table.

[0050] In addition, the system status table (tb_system_status) stores the status information of the operating environment, which is generally recorded every 30 minutes. However, in order to collect more ideal data, the collection frequency of system status data can be further adjusted.

[0051] Step S102, standardizing the backup task data and the system status data.

[0052] Step S103, define and merge the standardized backup task data and system status data into a data set, the data set structure includes task name, task start time, task end time, task data volume, system status time, system available space, number of system running tasks, and system load; define the fusion to be based on the time period of the backup task as the reference segment, if there is a single system status data within the reference segment, then directly perform the fusion; if there are multiple system status data within the reference segment, then take the arithmetic mean of the system status data within the reference segment for fusion; if there is no system status data within the reference segment, first take the next system status data adjacent to the reference segment for fusion, otherwise take the previous system status data adjacent to the reference segment for fusion.

[0053] It should be noted that by combining backup task information with system operation data as the source of data collection, the data dimension is enriched and the model's ability to understand backup needs is enhanced. Using these multi-dimensional data to train the prediction model can more accurately predict the impact of backup operations on system performance, greatly improve the selection of optimized backup time windows, improve resource utilization, and conduct risk management in advance, and significantly improve the level of intelligence.

[0054] However, due to the inconsistency between the execution time interval of the backup task and the collection time interval of the system status, the backup data table and the system status table data are prone to mismatch. The completion strategy adopted in this embodiment takes into account the continuity and relevance of the time series data, and completes the missing information by taking the latest system status data to ensure the coherence, integrity and consistency of the data set.

[0055] Secondly, if the backup task is performed in a specific time period, the relevant system status data should also be collected in this time period or at a close time point as much as possible. This can ensure that the match between the backup data and the system status data is accurate and has good time sensitivity. When the backup data is closely related to the system status, the analysis results are easier to understand and interpret, because each backup operation has a clear system status background, which provides a reliable basis for subsequent predictions.

[0056] In addition, this definition fusion method does not rely on a specific data structure or format, has good adaptability, and can be applied to a wider range of data sets and scenarios.

[0057] To facilitate understanding, we provide an example of defining fusion, such as Figure 2 As shown, the horizontal axis represents time. Within a certain time range, the following are collected in sequence: system status data 1, backup data 1, system status data 2, backup data 2, system status data 3, system status data 4, system status data 5, and backup data 3. System status data 3 and system status data 4 are both within the time period from the beginning to the end of backup data 2. When defining fusion, (1) backup data 1 and system status data 2 are defined to be fused; (2) the average value of system status data 3 and system status data 4 is calculated, and then the average value and backup data 2 are defined to be fused; (3) backup data 3 and system status data 5 are defined to be fused. In case (1), the next system status data adjacent to the benchmark segment is taken for fusion; in case (2), the average value of the system status data within the benchmark segment is taken for fusion; in case (3), the previous system status data adjacent to the benchmark segment is taken for fusion.

[0058] It is worth noting that a backup task has a start time and an end time, so it is a time period; the system status data is only data at a point in time, so the system status data will only be before, during, and after the backup task time period.

[0059] It is also worth noting that if the execution time of the backup task corresponds to the collection time of the system status, it is ideal. If there are multiple system status data in the time period, the arithmetic mean is taken (arithmetic mean = (V1 + V2 + ... +V n ) / n), where V1, V2 ... V n is the value of each system status data, and n is the total number of data). If the execution time of the backup task and the collection time of the system status do not correspond, the next closest system data point is closer to the execution time of the backup task, and it often contains the latest system status information, which is more beneficial for ensuring that the backup task can reflect the latest situation of the system; secondly, in some cases, the backup task may need to cover all data changes before the execution time, so using the next closest system data point can ensure that the backup task contains all system status changes before the execution time; in addition, if the system status has an obvious change trend over time, the next system data point can better reflect this trend, thereby providing more accurate context information for the backup task, which is conducive to training.

[0060] Step S104, segmenting the data set to obtain training data.

[0061] Step S105, input the training data into the training model, and train to obtain a prediction model, where the training model is a linear regression model.

[0062] The purpose of the artificial intelligence model is to achieve the function of data prediction by establishing a mathematical model, inputting corresponding values, and obtaining output values. The training model used in this embodiment is a linear regression model, which can predict continuous data, has high computational efficiency, good stability and is easy to optimize.

[0063] Step S106, outputting predicted continuous data through the prediction model.

[0064] Compared with the prior art, in this embodiment, the two types of data, backup task data and system status data, are defined and integrated in a preset manner, which greatly eliminates the influence of uncertain time period matching and time period mismatch in the collection of backup task data and system status data, and ensures the consistency and integrity of the training data set. The prediction model trained in this way can more accurately predict disk utilization, backup task time and data volume, and thus lay the foundation for providing users with more accurate backup task reports and suggestions.

[0065] Based on the above embodiment, in the method, step S102 further includes: using a logarithmic function to convert the task data volume and the system available space respectively, so that the data distribution of the two are both normally distributed.

[0066] The storage unit of the database is Bytes, which will cause the data interval to be large, the data will be skewed, and the generalization ability of the training model will be poor. Therefore, the logarithmic function should be used to transform the data, so that the data distribution can be close to the normal distribution.

[0067] Based on the above embodiment, in the method, step S102 further includes: performing feature engineering processing on the task start time, task end time and system status time to extract key information from the time series data.

[0068] During the model training process, by implementing feature engineering, we can more deeply explore the complex relationships between time data. For raw time data, such as "2023-12-11 11:11:11", we can usually convert it into a Unix timestamp (such as 1702264271) for calculation. However, this conversion method cannot effectively extract the potential relationship information in the time data. Through feature engineering, we can split the time data into finer time units, such as year, month, day, hour, minute, and second, and obtain a time series "2023, 12, 11, 11, 11, 11". This splitting method can better reveal the intrinsic relationship between various time features, thereby helping to improve the accuracy and performance of model training.

[0069] Based on the above embodiment, the backup data prediction method based on artificial intelligence provided in this embodiment includes:

[0070] Step S201, obtaining backup task data and system status data from a backup system;

[0071] Step S202, standardizing the backup task data and the system status data;

[0072] Step S203, the standardized backup task data and system status data are defined and merged into a data set, the data set structure includes task name, task start time, task end time, task data volume, system status time, system available space, number of system running tasks, and system load; the definition of fusion is based on the time period of the backup task as the reference segment, if there is a single system status data in the reference segment, then the fusion is directly performed; if there are multiple system status data in the reference segment, then the arithmetic mean of the system status data in the reference segment is taken for fusion; if there is no system status data in the reference segment, then the next system status data adjacent to the reference segment is first taken for fusion, otherwise the previous system status data adjacent to the reference segment is taken for fusion;

[0073] Step S204, segmenting the data set to obtain training data and test data, with the segmentation ratio of training data to test data being 8:2;

[0074] Step S205, inputting the training data into the training model, training to obtain a prediction model, the training model is a linear regression model;

[0075] Step S206, using the prediction model to predict the test data set to obtain a prediction result;

[0076] Step S207, using the true value of the test data and the prediction result, calculating the performance index, and then evaluating the generalization ability and performance of the prediction model according to the performance index;

[0077] Step S208, outputting predicted continuous data through the prediction model.

[0078] The difference of this embodiment is that the data is divided into training data and test data, wherein the training data is used to train the model so that the model can fit the patterns and rules of the data; the test data is used to evaluate the generalization ability of the model and measure the performance of the model. This data division can make better use of the data, not only taking into account the training of the model, but also fully evaluating the model.

[0079] The data set segmentation ratio of this embodiment is preferably 8:2, that is, the training data accounts for 80% of the total data set, and the test data accounts for 20% of the total data set. In this case, the model training effect is better.

[0080] Below, we provide a comparative experiment to further illustrate this embodiment, as follows:

[0081] Table 1. Experimental environment configuration parameters

[0082] name Configuration CPU Intel Core i5 12400 operating system Windows 10 Memory 16 GB Programming Language Python 3.10.14 Programming language management tools Anconda 3

[0083] All experimental examples follow the method steps of the above embodiment, and use Python to write the same running code and run it on the computer. This experiment uses Python mainly because Python is well integrated and relatively simple to implement.

[0084] In the experiment, the test objects are: test case 0, test case 1, test case 2, test case 3, test case 4 and test case 5. Among them, the data set split ratio of test case 0 is 9:1; the data set split ratio of test case 1 is 8:2; the data set split ratio of test case 2 is 7:3; the data set split ratio of test case 3 is 6:4; the data set split ratio of test case 4 is 5:5; the data set split ratio of test case 5 is 4:6. Each test case is trained 100 times, and the training accuracy and test accuracy of each time are recorded. Finally, the final accuracy is calculated. The final accuracy result is as follows Figure 3 and Figure 4 shown.

[0085] from Figure 3 and Figure 4 It can be seen that the final accuracy of test case 0 is 0.788321; the final accuracy of test case 1 is 0.805753; the final accuracy of test case 2 is 0.783668; the final accuracy of test case 3 is 0.778322; the final accuracy of test case 4 is 0.779789; the final accuracy of test case 5 is 0.780278. Therefore, when the data set segmentation ratio is 8:2, the final accuracy is relatively high.

[0086] It is also worth noting that data set segmentation can also increase the validation set, which can help parameter tuning, that is, it is used to adjust the hyperparameters of the model, such as learning rate, regularization parameters, etc.; it can also help prevent overfitting, that is, by evaluating the model on the validation set, it can be discovered in time whether the model is overfitting on the training set. The evaluation of the model by the validation set is performed before the test set test. Preferably, the data set segmentation ratio is preferably 7:2:1, that is, the training data accounts for 70% of the total data set, the test data accounts for 20% of the total data set, and the validation set accounts for 10% of the total data set.

[0087] Based on the above embodiment, in the method, step S105 includes:

[0088] Step S1051, initialize the weights and biases of the linear regression model, where the linear regression function is y = w1*x1+w2*x2+...+wn *x n +b, w is the weight, b is the bias;

[0089] Step S1052, using the training data and the gradient descent algorithm to perform model training, wherein the gradient descent algorithm presets a learning rate in advance;

[0090] Step S1053, comparing the initial predicted value calculated by the training model with the true value to obtain a preliminary difference;

[0091] Step S1054, updating weights and biases according to the preliminary difference and learning rate;

[0092] Step S1055, repeat the above steps to obtain the final weight and final bias.

[0093] The core content of model training is gradient descent, that is, the predicted value is calculated through training data, weights and biases, the difference is obtained by calculation with the true value, and the weights and biases are updated according to the calculated difference and learning rate. By alternating calculations, the weights and biases are continuously updated to improve the model's fitting ability.

[0094] It is understandable that the working process of gradient descent, such as Figure 5 As shown, specifically:

[0095] 1. Training data: Use the training dataset to train the model.

[0096] 2. Gradient Descent:

[0097] (1) Calculate the loss (MSE mean square error) based on the weights and biases;

[0098] (2) Update weights and biases based on loss and learning rate.

[0099] 3. Test data: Enter the test data.

[0100] 4. Test data: Use the test data set to calculate the generalization ability of the model.

[0101] 5. End training: Complete the model training process.

[0102] It can also be understood that the gradient descent algorithm iterative optimization principle, such as Figure 6 As shown, specifically:

[0103] Figure 6It shows an optimization process in a three-dimensional space. In this space, the X-axis and the Y-axis form a horizontal plane, and the Z-axis is perpendicular to this horizontal plane and represents the function value. The numbers -4, -2, 0, 2, 4 on the X-axis and the Y-axis, and the numbers 3, 2, 1, 0, -1, -2 on the Z-axis are the scales on the coordinate axes, which can help define the position of points in the three-dimensional space. The goal of the gradient descent algorithm is to find the lowest point of this function, which is the minimum value on the Z-axis.

[0104] In this three-dimensional space, the process of gradient descent starts from the starting point (usually a higher value on the Z axis), and gradually moves downward along the gradient (slope) of the function to find the global minimum point of the function.

[0105] It can be seen that the gradient descent process is achieved through continuous iteration: starting from a set of randomly initialized parameters, the algorithm calculates the loss function value, determines the gradient of the current position, and then updates the parameters to reduce the loss. This iterative process continues until the termination condition is met, such as reaching the maximum number of iterations or the loss function value is small enough. Each iteration is like rolling a ball on the terrain, moving it in the steepest downhill direction, and the ultimate goal is to reach the lowest point of the "valley", that is, the minimum point of the loss function. This point represents the optimal setting of the model parameters, which enables the model to perform best on the training data, while having good generalization ability to cope with new, unseen data. In this way, the gradient descent algorithm helps this embodiment to efficiently search in a complex parameter space and find the best model configuration.

[0106] like Figure 7 As shown, an embodiment of the present invention provides a backup data prediction system based on artificial intelligence, the system comprising:

[0107] The data acquisition module 3000 is used to acquire backup task data and system status data from the backup system;

[0108] The data standardization module 3001 is used to standardize the backup task data and system status data;

[0109] The data set definition module 3002 is used to define and merge the standardized backup task data and system status data into a data set. The data set structure includes the task name, task start time, task end time, task data volume, system status time, system available space, number of system running tasks, and system load. The definition of fusion is based on the time period of the backup task as the reference period. If there is a single system status data in the reference period, it is directly merged. If there are multiple system status data in the reference period, the arithmetic mean of the system status data in the reference period is taken for fusion. If there is no system status data in the reference period, the next system status data adjacent to the reference period is first taken for fusion, otherwise the previous system status data adjacent to the reference period is taken for fusion.

[0110] The data segmentation module 3003 is used to segment the data set to obtain training data and test data, and the segmentation ratio of the training data to the test data is 8:2;

[0111] The model training module 3004 is used to input the training data into the training model to obtain a prediction model through training. The training model is a linear regression model.

[0112] The testing module 3005 is used to use the prediction model to predict the test data set and obtain the prediction result;

[0113] Evaluation module 3006, used to calculate performance indicators using the true value of the test data and the prediction results, and then evaluate the generalization ability and performance of the prediction model according to the performance indicators;

[0114] The output prediction data module 3007 is used to output the predicted continuous data through the prediction model.

[0115] Based on the above embodiment, in the system, the data acquisition module 3000, such as Figure 8 As shown, including:

[0116] The backup task data acquisition unit 30001 is used to acquire the backup task data from the backup data table of the backup system. The backup task data includes the task name, task start time, task end time and task backup data volume;

[0117] The system status data acquisition unit 30002 is used to acquire system status data from the system status table of the production system. The system status data includes system status time, system used disk space, number of tasks in system operation and system load.

[0118] Based on the above embodiment, in the system, the model training module 3004, such as Fig. 9 As shown, including:

[0119] Initialization unit 30041 is used to initialize the weights and biases of the linear regression model, where the linear regression function is y = w1*x1+w2*x2+...+w n *x n +b, w is the weight, b is the bias;

[0120] A training unit 30042 is used to perform model training using training data and a gradient descent algorithm, wherein the gradient descent algorithm presets a learning rate in advance;

[0121] The difference comparison unit 30043 compares the initial predicted value calculated by the training model with the true value to obtain a preliminary difference;

[0122] An updating unit 30044, for updating weights and biases according to preliminary differences and learning rates;

[0123] The repeated tuning unit 30045 is used to repeatedly execute the above units to obtain the final weight and the final bias.

[0124] This embodiment avoids insufficient or wasted disk resources by more accurately predicting disk utilization, and improves the management effect of backup operations by predicting backup task time and data volume.

[0125] An embodiment of the present invention provides a report generation method, which generates a report based on the prediction data obtained by the artificial intelligence-based backup data prediction method described in any of the above embodiments and the existing data, and the report includes suggestions on disk utilization and task execution.

[0126] Based on the above embodiments, in this method, the generated report types include: task information report, system monitoring information report, task execution status report, disk usage report, resource allocation report and forecast analysis report; the report is presented in the form of one or more combinations of tables, line charts, pie charts, and bar charts.

[0127] Specifically, this embodiment can generate a task information list using the task name, next task run time, task start time, task end time, task backup data volume, and backup interval. Based on this list, the running status of the task can be viewed intuitively.

[0128] This embodiment can also use the system recording time, system available disk space, and system load to generate a system monitoring information list. By looking at this list, the status of the system at each moment can be understood.

[0129] like Fig.10As shown in the figure, a backup data bar chart is generated with the task name as the legend, the start time as the horizontal axis, and the task backup data volume as the vertical axis. Through this bar chart, you can intuitively observe the backup data and backup trends of each task, and you can also see the predicted backup data size.

[0130] The computer storage medium of the embodiment of the present invention can adopt any combination of one or more computer-readable media. The computer-readable medium can be a computer-readable signal medium or a computer-readable storage medium. The computer-readable storage medium can be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or device, or any combination of the above. More specific examples (non-exhaustive list) of computer-readable storage media include: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In this document, a computer-readable storage medium can be any tangible medium containing or storing a program, which can be used by an instruction execution system, device or device or used in combination with it.

[0131] Computer-readable signal media may include data signals propagated in baseband or as part of a carrier wave, which carry computer-readable program code. Such propagated data signals may take a variety of forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the above. Computer-readable signal media may also be any computer-readable medium other than a computer-readable storage medium, which may send, propagate, or transmit a program for use by or in conjunction with an instruction execution system, apparatus, or device.

[0132] Program code embodied on a computer readable medium may be transmitted using any appropriate medium, including but not limited to wireless, wireline, optical fiber cable, RF, etc., or any suitable combination of the foregoing.

[0133] Computer program code for performing the operations of the present invention may be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, C++, Ruby, Go, and conventional procedural programming languages ​​such as "C" or similar programming languages. The program code may be executed entirely on the user's computer, partially on the user's computer, as a separate software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In the case of a remote computer, the remote computer may be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0134] The above-mentioned embodiments only express several implementation methods of the present invention, and the description thereof is relatively specific and detailed, but it cannot be understood as limiting the scope of the patent of the present invention. It should be pointed out that, for ordinary technicians in this field, several variations and improvements can be made without departing from the concept of the present invention, which all belong to the protection scope of the present invention. Therefore, the protection scope of the patent of the present invention shall be subject to the attached claims.

Claims

1. A backup data prediction method based on artificial intelligence, characterized in that: The method comprises the steps of: Acquire backup task data and system status data from the backup system. The backup task data is collected based on the backup task completion time, and the collection time interval is not fixed; the system status data is collected periodically, and the collection time interval is fixed; Standardize backup task data and system status data; The standardized backup task data and system status data are defined and merged into a data set. The data set structure includes task name, task start time, task end time, task data volume, system status time, system available space, number of system running tasks, and system load. The fusion is defined to take the time period of the backup task as the benchmark period. If there is a single system status data in the benchmark period, it is directly merged. If there are multiple system status data in the benchmark period, the arithmetic mean of the system status data in the benchmark period is taken for fusion. If there is no system status data in the benchmark period, the next system status data adjacent to the time period is first taken for fusion, otherwise the previous system status data adjacent to the benchmark period is taken for fusion. After fusion, the time intervals of the data in the data set are not fixed, and the time intervals between the data are the same as the time intervals between the backup task data. Split the data set to obtain training data; Input the training data into the training model, and obtain the prediction model through training. The training model is a linear regression model. Predict continuous data through predictive model output.

2. The backup data prediction method based on artificial intelligence according to claim 1, characterized in that: The step of standardizing the backup task data and the system status data also includes: using a logarithmic function to convert the task data volume and the system available space respectively, so that the data distribution of the two is normally distributed.

3. The backup data prediction method based on artificial intelligence according to claim 1, characterized in that: The step of standardizing the backup task data and the system status data also includes: performing feature engineering processing on the task start time, the task end time and the system status time to extract key information from the time series data.

4. The backup data prediction method based on artificial intelligence according to claim 1, characterized in that: The step of inputting the training data into the training model to obtain the prediction model through training also includes: Initialize the weights and biases of the linear regression model, where the linear regression function is y = w1*x1+w2*x2+...+w n *x n +b, w is the weight, b is the bias; Use training data and gradient descent algorithm to train the model, where the gradient descent algorithm presets the learning rate in advance; Compare the initial predictions calculated by the training model with the true values ​​to obtain preliminary differences; Update weights and biases based on preliminary differences and learning rate; Repeat the above steps to obtain the final weights and final biases.

5. The backup data prediction method based on artificial intelligence according to any one of claims 1 to 4, characterized in that: The step of segmenting the data set further includes: The data set is split into training data and test data, and the split ratio of training data to test data is 8:2; Before the step of outputting the predicted continuous data through the prediction model, the step further includes: Use the prediction model to predict the test data set and obtain the prediction results; Using the true values ​​and predicted results of the test data, the performance indicators are calculated, and then the generalization ability and performance of the prediction model are evaluated based on the performance indicators.

6. A backup data prediction system based on artificial intelligence, characterized in that: The system comprises: The data acquisition module is used to acquire backup task data and system status data from the backup system; the backup task data is collected according to the backup task completion time, and its collection time interval is not fixed; the system status data is collected periodically, and its collection time interval is fixed; Data standardization module, used to standardize backup task data and system status data; The data set definition module is used to define and merge the standardized backup task data and system status data into one data set. The data set structure includes the task name, task start time, task end time, task data volume, system status time, system available space, number of system running tasks, and system load. The definition of fusion is based on the time period of the backup task as the reference period. If there is a single system status data in the reference period, it is directly merged. If there are multiple system status data in the reference period, the arithmetic mean of the system status data in the reference period is taken for fusion. If there is no system status data in the reference period, the next system status data adjacent to the time period is first taken for fusion, otherwise the previous system status data adjacent to the reference period is taken for fusion. After fusion, the time intervals of the data in the data set are not fixed, and the time intervals between the data are the same as the time intervals between the backup task data. The data segmentation module is used to segment the data set to obtain training data; The model training module is used to input the training data into the training model and obtain the prediction model through training. The training model is a linear regression model. The output prediction data module is used to output predicted continuous data through the prediction model.

7. The backup data prediction system based on artificial intelligence according to claim 6, characterized in that: The data acquisition module comprises: A backup task data acquisition unit, used to acquire backup task data from a backup data table of a backup system, the backup task data including task name, task start time, task end time and task backup data volume; A system status data acquisition unit, used to acquire system status data from a system status table of a production system, the system status data including a selected system status time, a system used disk space, a number of tasks in operation of the system, and a system load; The model training module includes: Initialization unit, used to initialize the weights and biases of the linear regression model, where the linear regression function is y = w1*x1+w2*x2+...+w n *x n +b, w is the weight, b is the bias; A training unit, used for model training using training data and a gradient descent algorithm, wherein the gradient descent algorithm presets a learning rate in advance; A difference comparison unit is used to compare the initial predicted value calculated by the training model with the true value to obtain a preliminary difference; An update unit that updates weights and biases based on the preliminary difference and the learning rate; The repeated tuning unit is used to repeatedly execute the above units to obtain the final weight and the final bias.

8. The backup data prediction system based on artificial intelligence according to claim 6, characterized in that: The data segmentation module is also used to segment the data set to obtain training data and test data, and the segmentation ratio of the training data to the test data is 8:2; Before the output prediction data module, it also includes: The test module is used to use the prediction model to predict the test data set and obtain the prediction results; The evaluation module is used to calculate the performance indicators using the true values ​​and prediction results of the test data, and then evaluate the generalization ability and performance of the prediction model based on the performance indicators.

9. A report generation method, characterized in that: A report is generated using the prediction data obtained by the artificial intelligence-based backup data prediction method according to any one of claims 1 to 4 and existing data, the report including recommendations on disk utilization and task intervals.

10. The report generation method according to claim 9, characterized in that: The generated report types include: task information report, system monitoring information report, task execution report, disk usage report, resource allocation report and forecast analysis report; the report is displayed in one or more combinations of tables, line charts, pie charts, and bar charts.

Images