Industrial control system data backup and recovery method

Abstract

The invention provides an industrial control system data backup and recovery method, which comprises the following steps: acquiring to-be-backed-up data, and extracting a state feature vector of the to-be-backed-up data; inputting the state feature vector of the to-be-backed-up data into an intelligent agent model, and outputting a data backup strategy and a data recovery scheme of the to-be-backed-up data; and according to the data backup strategy, performing data backup on the to-be-backed-up data, and when data recovery needs to be triggered after the to-be-backed-up data is backed up, recovering the backup data based on the data recovery scheme. Through the intelligent agent model, the optimal data backup strategy and the optimal data recovery scheme can be dynamically generated according to the historical backup data, the current data state and the system resource condition of the industrial control system database, and an efficient, intelligent and high-adaptability solution is provided for data backup and recovery of the industrial control system.

Description

Technical Field

[0001] The present invention relates to the technical field of data management, and more specifically, to a method for backing up and restoring data of an industrial control system. Background Art

[0002] In the prior art, the data backup methods of industrial control systems mostly rely on a pre-set backup method decision model, for example, the KPCA-SVM (kernel principal component analysis-support vector machine) decision model based on machine learning technology. Although such models can flexibly adjust the backup strategy according to the influencing factors of data backup to a certain extent, the complexity of model training is relatively high, and it is incapable of coping with complex and changeable industrial environments. On the other hand, although the data recovery method in the CNC process provides a comprehensive and detailed data protection strategy, it has obvious deficiencies in intelligent backup strategy recommendations, which limits the potential for improving backup efficiency and data recovery efficiency. Summary of the invention

[0003] In view of the technical problems existing in the prior art, the present invention provides a method for backing up and restoring data of an industrial control system. The method can comprehensively consider the historical backup data of the industrial control system database, the current data status and the resource status of the system, so as to dynamically generate the optimal backup strategy recommendation and provide an efficient data recovery solution when necessary.

[0004] The present invention provides an industrial control system data backup and recovery method, comprising:

[0005] Acquire data to be backed up, and extract state feature vectors of the data to be backed up;

[0006] Inputting the state feature vector of the data to be backed up into the agent model, and outputting the data backup strategy and data recovery solution of the data to be backed up;

[0007] Backing up the data to be backed up according to the data backup strategy, and restoring the backup data based on the data recovery solution when data recovery needs to be triggered after the data to be backed up is backed up;

[0008] Among them, the intelligent agent model includes a first intelligent agent sub-model and a second intelligent agent sub-model, the first intelligent agent sub-model is used to generate a corresponding data backup strategy according to the state feature vector of the data to be backed up, and the second intelligent agent sub-model is used to generate a data recovery plan for the data to be backed up according to the data backup strategy of the data to be backed up and the current system resource status.

[0009] The present invention provides a method for data backup and recovery of an industrial control system, which extracts a state feature vector of the data to be backed up; inputs the state feature vector of the data to be backed up into an intelligent agent model, and outputs a data backup strategy and a data recovery plan for the data to be backed up; backs up the data to be backed up according to the data backup strategy, and when data recovery needs to be triggered after backing up the data to be backed up, recovers the backup data based on the data recovery plan. The present invention can dynamically generate the optimal data backup strategy and data recovery plan based on the historical backup data, current data status, and system resource status of the industrial control system database through the intelligent agent model, providing an efficient, intelligent, and highly adaptable solution for data backup and recovery of industrial control systems. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] Figure 1 The present invention provides a flowchart of a method for backing up and restoring data of an industrial control system. DETAILED DESCRIPTION

[0011] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments in the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention. In addition, the technical features in the various embodiments or single embodiments provided by the present invention can be arbitrarily combined with each other to form a feasible technical solution. This combination is not subject to the constraints of the sequence of steps and / or the structural composition mode, but must be based on the ability of ordinary technicians in this field to achieve. When the combination of technical solutions is contradictory or cannot be achieved, it should be considered that such a combination of technical solutions does not exist and is not within the scope of protection required by the present invention.

[0012] Figure 1 A flow chart of a data backup and recovery method for an industrial control system provided by the present invention is as follows: Figure 1 As shown, the method includes:

[0013] Step 1: Acquire data to be backed up and extract state feature vectors of the data to be backed up.

[0014] It is understandable that for the data to be backed up, the state feature vector of the data to be backed up is extracted. Before extracting the state feature vector with the backup data, the backup data is preprocessed. The data preprocessing includes data cleaning and normalization processing to eliminate outliers and dimensional differences, and provide a high-quality data set for subsequent feature extraction and model training. The formula for normalization processing is:

[0015]

[0016] Among them, x is the data to be backed up, x norm is the normalized data, x min and x max are the minimum and maximum values ​​in the data set, respectively.

[0017] For the preprocessed data to be backed up, the state feature vector of the data to be backed up is extracted as the input of the deep reinforcement learning agent model, wherein the feature extraction is usually based on domain knowledge and data analysis.

[0018] Among them, feature extraction uses principal component analysis to reduce the dimension of the feature vector and improve computational efficiency; the principal components are arranged from large to small according to the variance, and the first principal component has the largest variance, that is, it contains the most information. The steps of state feature vector extraction include:

[0019] Covariance matrix: The formula for calculating the covariance matrix C of the data is expressed as:

[0020]

[0021] Where X is the original data matrix, n is the number of original data samples;

[0022] Eigenvalues ​​and eigenvectors: Solve the eigenvalues ​​and eigenvectors of the covariance matrix C. The formula is expressed as:

[0023] CV i =λ i λV i

[0024] Among them, λ i is the i-th eigenvalue, V i is the corresponding eigenvector;

[0025] Select the principal component: According to the size of the eigenvalue, select the eigenvectors corresponding to the first k largest eigenvalues ​​to form the projection matrix P = [V 1 , V 2 , ..., V k ], which is the state feature vector of the original data;

[0026] Dimensionality reduction: Project the original data matrix onto the projection matrix, and the formula for the reduced-dimensional data is: reduced =XP.

[0027] Step 2: input the state feature vector of the data to be backed up into the agent model, and output the data backup strategy and data recovery solution of the data to be backed up.

[0028] It is understandable that the state feature vector of the data to be backed up extracted in step 1 is input into the deep learning agent model to obtain the data backup strategy and data recovery solution of the data to be backed up.

[0029] Among them, the intelligent agent model includes a first intelligent agent sub-model and a second intelligent agent sub-model. The first intelligent agent sub-model is used to generate a corresponding data backup strategy according to the state feature vector of the data to be backed up, and the second intelligent agent sub-model is used to generate a data recovery plan for the data to be backed up according to the data backup strategy of the data to be backed up and the current system resource status.

[0030] The entire agent model includes a first agent sub-model and a second agent sub-model, which are used to generate a data backup strategy and a data recovery solution respectively. The training process of the entire agent model includes training the first agent sub-model and the second agent sub-model respectively, and then training the agent model composed of the first agent sub-model and the second agent model as a whole.

[0031] The training process of the first agent sub-model includes:

[0032] Step 21 : acquiring historical backup data information, extracting a data backup strategy of historical original data from the historical backup data information, and extracting a state feature vector of the historical original data.

[0033] It is understandable that the historical data backup information is collected, and the historical backup data information includes backup time, recovery time, data increment size, data difference change rate, total data, system resource utilization, backup frequency, backup storage location, and system log information. The data backup strategy of the historical original data can be extracted from the acquired historical data backup information. Specifically, the data backup strategy of the historical original data includes the backup method of the historical original data, the backup time window, the backup storage location, and the backup frequency.

[0034] The state feature vector of the historical original data is extracted. The steps of extracting the state feature vector of the historical original data can refer to the steps of extracting the state feature vector of the backup data mentioned above, and will not be repeated here.

[0035] Step 22: training the first agent sub-model based on the state feature vector of the historical original data and the data backup strategy.

[0036] Among them, the first intelligent agent sub-model includes a policy network and a value network. The policy network is responsible for generating backup strategy recommendations based on the current state, and the value network is used to evaluate the expected benefits of adopting a backup strategy under a given state, assisting in the optimization of the policy network.

[0037] Training the first agent sub-model includes:

[0038] Inputting the state feature vector of the historical original data into the strategy network to generate a data backup strategy;

[0039] Evaluate the expected benefits of adopting the data backup strategy for historical original data backup through the value network;

[0040] According to the expected benefits, the strategy network is optimized and updated. The optimization update formula of the strategy network is:

[0041]

[0042] Among them, μ(s i ) is the action output by the policy network, i.e., the data backup strategy, s i is the i-th state feature vector, Q(s i ,a i |θ Q ) is the Q value output by the value network.

[0043] Wherein, the parameters of the policy network and the value network are updated based on the deep deterministic policy gradient algorithm, and the training objective function of the first agent sub-model is:

[0044]

[0045] Among them, θ is the first agent sub-model parameter, γ is the discount factor, r(s t ,a t ) is the state eigenvector s t Take action a t The instant reward obtained, t is the time step, the instant reward r(s t ,a t ) is calculated based on the sum of backup time and recovery time, data consistency before and after recovery, and system resource utilization.

[0046] The parameters of the policy network and the value network are continuously updated based on the training objective function until the training objective function is maximized.

[0047] After the training of the first agent sub-model is completed, the first agent sub-model can generate the optimal backup strategy recommendation in real time according to the state feature vector of the current industrial control system database. The update formula of the strategy recommendation can be expressed as:

[0048] α * =μ(s|θ μ )

[0049] Among them, α *is the optimal backup strategy recommendation, μ is the policy network, s is the state feature vector of the current data, and θ μ are the parameters of the policy network.

[0050] For the second intelligent agent sub-model, when the original data is backed up and the data recovery demand is triggered, the second intelligent agent sub-model generates the optimal data recovery plan based on the historical recovery data, the current system resource status and the data backup strategy.

[0051] Among them, the second agent sub-model uses the Monte Carlo tree search algorithm in reinforcement learning to generate the optimal data recovery solution. The Monte Carlo tree search algorithm uses the UCT formula to select the optimal path for exploration. The UCT formula can be expressed as:

[0052]

[0053] Among them, Q(v i ) is the node v i The cumulative reward, N(v i ) is the node v i The number of visits, C' is the trade-off parameter between exploration and utilization, v parent Yes i The parent node of .

[0054] In the process of training the second agent sub-model, an evaluation function of the data recovery solution output by the second agent sub-model is calculated according to the data recovery time, recovery success rate, data consistency and system resource occupancy during the recovery process: the parameters of the second agent sub-model are optimized based on the evaluation function. Through training, a trained second agent sub-model is obtained.

[0055] After the first agent sub-model and the second agent sub-model are trained, the agent model composed of the first agent sub-model and the second agent sub-model is trained. During the training process, the agent model is trained based on the state feature vector of the historical original data, the data backup strategy and the data recovery plan with the goal of minimizing the sum of the data backup time and the data recovery time.

[0056] When the intelligent agent model is trained, the performance evaluation index of the intelligent agent model obtained after training is calculated, and based on the performance evaluation index, the network structure of the intelligent agent model is optimized and adjusted, wherein the performance evaluation index includes the sum of backup time and recovery time, data consistency, system resource utilization and user satisfaction.

[0057] It should be noted that the first intelligent agent sub-model and the second intelligent agent sub-model are both neural network models. Optimizing and adjusting the network structure of the intelligent agent model according to the performance evaluation indicators mainly involves optimizing and adjusting the number of network layers, the number of neurons and the activation function of the first intelligent agent sub-model and the second intelligent agent sub-model to adapt to changes in the industrial control system environment and improve the accuracy and adaptability of generating data backup strategies and data recovery solutions.

[0058] Step 3: backing up the data to be backed up according to the data backup strategy, and when data recovery needs to be triggered after backing up the data to be backed up, restoring the backup data based on the data recovery solution.

[0059] It is understandable that when the current data needs to be backed up, the state feature vector of the data to be backed up is extracted, the extracted state feature vector is input into the trained agent model, and the agent model outputs the corresponding data backup strategy. The data to be backed up is backed up according to the data backup strategy output by the agent model.

[0060] When the backup data needs to be restored after being backed up, the data backup strategy and the current system resource status are input into the intelligent agent model, and the intelligent agent model outputs a data recovery plan, and the backup data is restored according to the data recovery plan.

[0061] The present invention provides a method for backing up and restoring data in an industrial control system, which has the following beneficial effects:

[0062] (1) The present invention uses a deep reinforcement learning model to dynamically generate optimal backup strategy recommendations based on the historical backup data, current data status, and system resource conditions of the industrial control system database. Compared with the traditional backup method based on preset models, the present invention greatly enhances the intelligence level and dynamic adaptability, and can better adapt to complex and changing industrial environments.

[0063] (2) Through interactive learning between the deep reinforcement learning agent and the industrial control system database, the present invention can learn how to select the optimal backup strategy under different conditions to minimize the sum of backup time and recovery time, which not only improves the backup efficiency but also optimizes the data recovery solution, thereby achieving a significant improvement in overall efficiency;

[0064] (3) In the feature extraction stage, the present invention uses principal component analysis to perform dimensionality reduction processing, which effectively reduces the dimension of the feature vector, improves the computational efficiency, helps the intelligent agent make decisions more quickly, and reduces the complexity of the model;

[0065] (4) The present invention establishes a model update and adaptive adjustment mechanism, which can regularly collect new backup and recovery data and update the model through an online learning mechanism, which not only ensures the timeliness of the model, but also enables it to adapt to changes in the industrial control system environment and continuously provide high-quality backup and recovery suggestions;

[0066] (5) By evaluating the performance of the backup strategy recommendations and data recovery solutions generated by the agent and providing feedback training to the agent based on the evaluation results, the present invention can continuously improve the accuracy and adaptability of strategy generation, thereby helping to ensure the effectiveness and reliability of backup and recovery operations;

[0067] (6) The present invention effectively solves the problems in the prior art such as high complexity of the backup method decision model, difficulty in adapting to complex industrial environments, and limited room for improving backup and recovery efficiency by introducing technical means such as deep reinforcement learning, feature extraction and dimensionality reduction optimization, model updating and adaptive adjustment, and comprehensive performance evaluation and feedback mechanism. It provides an efficient, intelligent and adaptable solution for data backup and recovery of industrial control systems.

[0068] It should be noted that in the above embodiments, the description of each embodiment has its own emphasis, and for parts that are not described in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0069] Those skilled in the art will appreciate that embodiments of the present invention may be provided as methods, systems, or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0070] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded computer, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0071] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0072] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process in the computer or other programmable device. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0073] Although the preferred embodiments of the present invention have been described, those skilled in the art may make other changes and modifications to these embodiments once they have learned the basic creative concept. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications that fall within the scope of the present invention.

[0074] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include these modifications and variations.

Claims

1. A method for backing up and restoring data of an industrial control system, characterized in that: include: Acquire the data to be backed up in the industrial control system database, and extract the state feature vector of the data to be backed up; Inputting the state feature vector of the data to be backed up into the agent model, and outputting the data backup strategy and data recovery solution of the data to be backed up; Backing up the data to be backed up according to the data backup strategy, and restoring the backup data based on the data recovery solution when data recovery needs to be triggered after the data to be backed up is backed up; Among them, the intelligent agent model includes a first intelligent agent sub-model and a second intelligent agent sub-model, the first intelligent agent sub-model is used to generate a corresponding data backup strategy according to the state feature vector of the data to be backed up, and the second intelligent agent sub-model is used to generate a data recovery plan for the data to be backed up according to the data backup strategy of the data to be backed up and the current system resource status.

2. The industrial control system data backup and recovery method according to claim 1, characterized in that: The training of the agent model includes: Acquire historical backup data information in the industrial control system database, extract a data backup strategy for historical original data from the historical backup data information, and extract a state feature vector of the historical original data; Training the first agent sub-model based on the state feature vector of the historical original data and the data backup strategy; Acquire historical recovery data information and current system resource status, and extract a data recovery solution from the historical recovery data information; Training the second agent sub-model based on a data backup strategy of historical original data, current system resource status, and a data recovery solution; Based on the state feature vectors of historical original data, data backup strategies and data recovery solutions, the intelligent agent model is trained with the goal of minimizing the sum of data backup time and data recovery time to obtain a trained intelligent agent model.

3. The industrial control system data backup and recovery method according to claim 2, characterized in that: The historical backup data information includes backup time, recovery time, data increment size, data difference change rate, total data volume, system resource utilization, backup frequency, backup storage location, and system log information. The data backup strategy for extracting historical original data from the historical backup data information includes: The backup mode, backup time window, backup storage location and backup frequency of the historical original data are extracted from the historical backup data information.

4. The industrial control system data backup and recovery method according to claim 2, characterized in that: The step of extracting the state feature vector of the historical raw data includes: Calculate the covariance matrix C of historical raw data: Among them, X is the historical original data matrix, n is the number of historical original data samples; Solve for the eigenvalues ​​and eigenvectors of the covariance matrix C: CV i =λ i λV i Among them, λ i is the i-th eigenvalue, V i is the eigenvector corresponding to the i-th eigenvalue; According to the size of the eigenvalue, the eigenvectors corresponding to the first k largest eigenvalues ​​are selected to form the projection matrix P = [V1, V2, ..., V k ], P is the state feature vector of the extracted historical original data.

5. The industrial control system data backup and recovery method according to claim 2, characterized in that: The first agent sub-model includes a strategy network and a value network. The state feature vector and data backup strategy based on the historical original data are used to train the first agent sub-model, including: Inputting the state feature vector of the historical original data into the strategy network to generate a data backup strategy; Evaluate the expected benefits of adopting the data backup strategy for historical original data backup through the value network; According to the expected benefits, the strategy network is optimized and updated. The optimization update formula of the strategy network is: Among them, a i is the i-th action output by the policy network, i.e., the i-th data backup policy, s i is the i-th state feature vector, Q(s i ,a i |θ Q ) is the Q value output by the value network; The parameters of the policy network and the value network are updated based on the deep deterministic policy gradient algorithm, and the training objective function of the first agent sub-model is: Among them, θ is the first agent sub-model parameter, γ is the discount factor, r(s t ,a t ) is the state eigenvector s t Take action a t The instant reward obtained, t is the time step, the instant reward r(s t ,a t ) is calculated based on the sum of backup time and recovery time, data consistency before and after recovery, and system resource utilization factors; The parameters of the policy network and the value network are continuously updated based on the training objective function until the training objective function is maximized.

6. The industrial control system data backup and recovery method according to claim 2, characterized in that: The second agent sub-model adopts the Monte Carlo tree search algorithm in reinforcement learning according to the data backup strategy of the historical original data and the current system resource status, and uses the UCT formula to select the optimal path for exploration. The UCT formula is expressed as: Among them, Q(v i ) is the node v i The cumulative reward, N(v i ) is the node v i The number of visits, C' is the trade-off parameter between exploration and utilization, v parent Yes i The parent node of The evaluation function of the data recovery solution output by the second agent sub-model is calculated according to the data recovery time, recovery success rate, data consistency and system resource usage during the recovery process: The second agent sub-model parameters are optimized based on the evaluation function.

7. The industrial control system data backup and recovery method according to claim 2, characterized in that: The state feature vector, data backup strategy and data recovery scheme based on the historical original data are used to train the agent model with the goal of minimizing the sum of data backup time and data recovery time to obtain the trained agent model, and then further include: The performance evaluation indicators of the intelligent agent model obtained after the training are calculated, and based on the performance evaluation indicators, the network structure of the intelligent agent model is optimized and adjusted, wherein the performance evaluation indicators include the sum of backup time and recovery time, data consistency, system resource utilization and user satisfaction.

8. The industrial control system data backup and recovery method according to claim 7, characterized in that: The optimizing and adjusting the network structure of the agent model based on the performance evaluation index includes: Based on the performance evaluation index, the number of network layers of the intelligent agent model is increased or decreased, the number of neurons is adjusted, or different activation functions are adopted.

Images