Power intelligent body process isolation protection method and system based on sandbox

Abstract

The present invention discloses a sandbox-based power intelligent body process isolation protection method and system, which relates to the field of power intelligent body process isolation protection technology, including the following steps: obtaining power system power frequency time series data, extracting frequency fluctuation curve to construct a first query sequence, using DTW algorithm to screen the sequence to obtain a reference sequence, and constructing a frequency feature vector for Fourier transform to generate a granularity matrix; inputting the granularity matrix into a multi-objective optimization function, using NSGA-III algorithm to evaluate memory allocation granularity; dividing the process memory space into several isolated areas according to the evaluation results, optimizing the memory capacity of each area with a quantum annealing algorithm to obtain the optimal capacity; deploying the optimal capacity to a memory protection unit to form an isolation barrier, and integrating it into a sandbox to achieve secure isolation. This method solves the problem of mismatch between memory allocation granularity and capacity in traditional process isolation, and avoids the problem of micro-granular memory attacks such as cross-area access and cache injection.

Description

Technical Field

[0001] The present invention relates to the technical field of power intelligent body process isolation and protection, and more specifically, to a sandbox-based power intelligent body process isolation and protection method and system. Background Art

[0002] In the current power system, with the widespread application of artificial intelligence technology, power intelligent agents have gradually become an important part of system operation. At the same time, power intelligent agents are highly dependent on computing resources and complex data interactions during operation, which leads to them facing increasingly severe security threats, especially at the memory level. Traditional process isolation mechanisms mostly use static memory allocation or coarse-grained strategies based on permission control, which are difficult to deal with high-frequency and multi-variant attack methods such as memory traversal, buffer overflow, and malicious process hijacking. In addition, although current sandbox technology is widely used in information system security, its isolation capabilities and resource adaptability are insufficient in the power system environment, and it cannot meet the requirements of power intelligent agents for real-time, high reliability, and refined protection.

[0003] For example, the invention patent announcement with announcement number CN118153035A discloses a sandbox-based smart contract protection method, device, and electronic device. The method includes: determining that a data sandbox accesses a smart contract, obtaining a security value for the smart contract, and obtaining a confidentiality value and an integrity value. The confidentiality value indicates the level of read permission, and the integrity value indicates the level of write permission. When the security value of the smart contract and the security value of the data sandbox meet a preset condition, the security value of the smart contract is updated. The present disclosure ensures that the smart contract process can be securely executed in a controlled and protected environment by updating the security value of the smart contract when the security value of the smart contract and the security value of the data sandbox meet a preset condition.

[0004] The above disclosed technical solutions have at least the following technical problems:

[0005] Traditional process isolation generally adopts a static allocation mechanism with fixed granularity in process memory protection, that is, dividing the memory into partitions of predefined sizes, without considering the time-varying and different nature of the memory access patterns of the process during operation, resulting in a memory allocation granularity that does not match actual needs. For example, some frequently accessed process areas may be allocated too few resources, resulting in frequent page changes and even information leakage risks; while low-activity areas occupy more memory resources, resulting in resource waste. In addition, this type of method lacks the ability to model dynamic behavior and cannot make optimal adjustments based on the memory usage characteristics of the power intelligent body at different operating times and different task stages, resulting in rough division of isolation areas and fuzzy boundaries, making it difficult to effectively prevent fine-grained memory attacks such as cross-region reading and writing, cache injection, etc. In response to the above problems, the present invention proposes a solution. Summary of the Invention

[0006] In order to overcome the above-mentioned defects of the prior art, an embodiment of the present invention provides a sandbox-based power intelligent body process isolation protection method and system, which realizes dynamic granularity division and optimal capacity allocation of the power intelligent body memory space by combining granularity matrix modeling with multi-objective optimization, so as to solve the problem that traditional process isolation does not consider the mismatch between memory allocation granularity and actual memory resource capacity required, resulting in cross-region reading and writing, cache injection and other fine-grained memory attacks.

[0007] To achieve the above object, the present invention provides the following technical solutions:

[0008] A sandbox-based power intelligent agent process isolation and protection method includes the following steps: obtaining power system power frequency time series data, extracting the current frequency fluctuation curve to construct a first query sequence; using the DTW algorithm to perform sequence screening on the first query sequence to obtain a reference sequence; constructing a frequency feature vector based on the reference sequence and performing Fourier transform to obtain a granularity matrix; inputting the granularity matrix into a preset multi-objective optimization function, and using the NSGA-III evolutionary algorithm to evaluate the memory allocation granularity; dividing the intelligent agent's process memory space into several isolation areas based on the evaluation results, and performing quantum annealing optimization on the memory capacity of each isolation area to obtain the optimal memory capacity; deploying the optimal memory capacity to a memory protection unit to generate an isolation barrier, and integrating the isolation barrier into the sandbox to achieve secure isolation.

[0009] In a preferred embodiment, the extraction of the current frequency fluctuation curve to construct the first query sequence is specifically as follows: based on the power system power frequency time series data, the frequency fluctuation curve is extracted through an adaptive sliding window; the frequency fluctuation curve is decomposed at multiple scales to obtain several long correlation components; and the first query sequence is constructed based on the several long correlation components.

[0010] In a preferred embodiment, the use of the DTW algorithm to perform sequence screening on the first query sequence to obtain a reference sequence specifically comprises: obtaining a set of candidate sequences from a historical frequency fluctuation database; using the DTW algorithm to calculate the DTW distance between each candidate sequence and the first query sequence, and performing sequence screening on the first query sequence based on the DTW distance to obtain a reference sequence.

[0011] In a preferred embodiment, the frequency feature vector is constructed based on the reference sequence and Fourier transform is performed to obtain a granularity matrix, specifically: the reference sequence is divided into several sequence segments, and the DTW distance between each sequence segment is calculated using the DTW algorithm; a graph structure is constructed with the sequence segments as nodes and the DTW distance as the edge weight; the features of the graph structure are aggregated through a preset graph convolution layer to generate a first feature vector; Fourier transform is performed on the first feature vector to output a spectrum matrix; the frequency matrix is probabilistically fused, and a granularity matrix is constructed based on a preset convolutional neural network.

[0012] In a preferred embodiment, the frequency matrix is probabilistically fused and a granularity matrix is constructed based on a preset convolutional neural network, specifically: the historical access log of the power intelligent body is obtained, the page access frequency of the same process in different time slices is extracted, and an access probability vector is constructed; the spectrum matrix and the access probability vector are aligned according to a preset time window, and the frequency domain energy characteristics of each time window are calculated; a second eigenvector is constructed based on the frequency domain energy characteristics and the access probability vector; and the second eigenvector is input into the preset convolutional neural network to obtain a granularity matrix.

[0013] In a preferred embodiment, the granularity matrix is input into a preset multi-objective optimization function, and the memory allocation granularity is evaluated using the NSGA-III evolutionary algorithm, specifically: singular value decomposition is performed on the granularity matrix to extract singular value vectors; the granularity matrix is sliced according to the singular value vectors to generate several memory block mapping structure matrices; a multi-objective optimization function is constructed for each memory block mapping structure matrix and encoded as an initial population individual of the NSGA-III algorithm; crossover and mutation operations are iteratively performed to calculate the multi-objective optimization function value of each individual; and the Pareto front solution set is output based on the non-dominated sorting and multi-objective optimization function value mechanism to obtain the optimal memory allocation granularity.

[0014] In a preferred embodiment, the process memory space of the intelligent agent is divided into several isolated areas according to the evaluation results, specifically: based on the optimal memory allocation granularity, the page access frequency sequence of the intelligent agent process in a preset historical time slice is obtained, and an access density vector is constructed; the access density vector is segmented using a sliding window mechanism to generate the first access density sub-vectors of several windows; the first access density vector is clustered and the mean and standard deviation of the page access frequency of each cluster are calculated; the process memory space of the intelligent agent is divided into several isolated areas based on the ratio of the mean and standard deviation of the access frequency.

[0015] In a preferred embodiment, the memory capacity of each isolation region is subjected to quantum annealing optimization to obtain the optimal memory capacity, specifically by: defining a Hamiltonian model for each isolation region, using a quantum annealing algorithm to solve the Hamiltonian ground state in the solution space of the Hamiltonian model to obtain a first candidate capacity solution set; constructing a Voroni graph based on the first candidate capacity solution set, and dividing the solution space into several spaces based on the Voroni graph; calculating the solution density of each space, and randomly perturbing the first candidate capacity solution set based on the solution density to obtain a second candidate capacity solution set; applying the second candidate capacity solution set to the next round of quantum annealing optimization, and jumping out of the local optimal solution through the tunneling effect to output the optimal memory capacity.

[0016] In a preferred embodiment, the optimal memory capacity is deployed to the memory protection unit, an isolation barrier is generated, and the isolation barrier is integrated into the sandbox to achieve secure isolation, specifically: the optimal memory capacity vector is loaded into the memory protection unit, and the base address register and boundary register are configured; a hardware isolation barrier is generated based on the base address register and boundary register mapping; the hardware isolation barrier is embedded in the sandbox kernel to monitor the memory access requests of the intelligent process in real time; when illegal cross-region access is detected, a security interrupt is triggered and redirected to the sandbox isolation log pool.

[0017] The technical effects and advantages of the sandbox-based power intelligent body process isolation protection method and system of the present invention are as follows:

[0018] 1. This invention utilizes the DTW algorithm and multi-scale analysis to screen and model frequency sequences. It further combines Fourier transforms and convolutional neural networks to generate a granularity matrix, fully exploiting the time-frequency characteristics of agent behavior. Furthermore, the NSGA-III evolutionary algorithm performs multi-objective optimization of memory allocation granularity to ensure the scientific and rational nature of memory partitioning. Furthermore, the process is isolated based on access density vectors, and the memory capacity of each isolation zone is optimized using a quantum annealing algorithm to achieve the optimal memory capacity. This optimal capacity is deployed to the memory protection unit through hardware configuration, forming a dynamic isolation barrier that is monitored and protected in real time within the sandbox kernel. When unauthorized access occurs, the system rapidly responds and isolates the risk, safeguarding the integrity of the core process operating environment. Driven by optimal memory capacity, this method establishes a precise protection mechanism for agent memory access, transcending the limitations of coarse-grained isolation in traditional sandboxes and providing an efficient and adaptable technical approach for the safe operation of agents in power systems. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] Figure 1 The figure is a flow chart of the sandbox-based power intelligent body process isolation and protection method of the present invention.

[0020] Figure 2 This is a structural diagram of the sandbox-based power intelligent process isolation and protection system of the present invention. DETAILED DESCRIPTION

[0021] The following will provide a clear and complete description of the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. All other embodiments obtained by ordinary technicians in this field based on the embodiments of the present invention without making any creative efforts shall fall within the scope of protection of the present invention.

[0022] Example 1, Figure 1 The present invention provides a sandbox-based power intelligent agent process isolation and protection method, which includes the following steps:

[0023] S1, obtain the power system power frequency time series data, extract the current frequency fluctuation curve to construct the first query sequence;

[0024] In this example, the power system power frequency time series data is obtained, specifically:

[0025] To achieve dynamic awareness of the power agent's operating environment and memory isolation, power frequency time series data is first acquired from the power system's data acquisition equipment (such as PMUs (synchronized phasor measurement units) and SCADA systems). This power frequency time series data is a continuously recorded grid frequency value, with a sampling frequency that can be set to 50Hz, 100Hz, or even higher to meet high-resolution requirements. This data is typically stored as timestamp-frequency pairs.

[0026] Power system power frequency time series data primarily refers to the continuous sampling and recording of power frequency components of physical quantities such as voltage and current in the power grid over time, typically centered around the system frequency. This data, characterized by time series characteristics, provides essential information for describing power system operational stability, load variations, and abnormal disturbances. Specifically, it includes system frequency, timestamp, voltage phase angle, voltage and current amplitudes, and sampling point or sequence numbers.

[0027] In this example, the current frequency fluctuation curve is extracted to construct the first query sequence, specifically:

[0028] Based on the power system power frequency time series data, the frequency fluctuation curve is extracted through an adaptive sliding window;

[0029] Perform multi-scale decomposition on the frequency fluctuation curve to obtain several long-range correlation components;

[0030] A first query sequence is constructed based on a plurality of long correlation components.

[0031] It should be noted that the system first obtains real-time grid frequency time series data from power monitoring equipment (such as PMUs or dispatching master stations) and uses an adaptive sliding window algorithm to extract the current frequency fluctuation curve. The length of this sliding window is dynamically adjusted based on the intensity of historical frequency fluctuations. For example, during periods of severe fluctuations, the window length can be shortened to enhance response sensitivity, while during stable operation, the window length can be extended to improve steady-state modeling accuracy. The extracted frequency fluctuation curve is then subjected to a multi-scale decomposition, such as using wavelet transforms or empirical mode decomposition (EMD) to decompose the original fluctuation curve into several long-range correlation components representing different time scales. These components can better reveal different periodic or trend characteristics in system operation, avoiding misjudgments due to local disturbances.

[0032] Next, a first query sequence is constructed based on multiple long-correlation components, and a large number of candidate sequence sets are extracted from the historical frequency fluctuation database. These candidate sequences are derived from historical records of different typical load scenarios, fault disturbances, frequency jumps, or regional power flow changes during the operation of the power grid. Afterwards, the DTW algorithm is used to calculate the DTW distance between the candidate sequence and the current first query sequence one by one. The DTW algorithm can accurately measure the similarity between two sequences in the presence of time offset and rate changes. Finally, according to the DTW distance, a group of sequences with the smallest DTW distance are selected as the reference sequence. This reference sequence will serve as the basic input for subsequent frequency behavior modeling and granularity matrix generation.

[0033] S2, using the DTW algorithm to perform sequence screening on the first query sequence to obtain a reference sequence;

[0034] In this example, the DTW algorithm is used to perform sequence screening on the first query sequence to obtain a reference sequence, specifically:

[0035] Obtain a candidate sequence set from a historical frequency fluctuation database;

[0036] The DTW algorithm is used to calculate the DTW distance between each candidate sequence and the first query sequence, and the first query sequence is screened according to the DTW distance to obtain a reference sequence.

[0037] S3, construct the frequency feature vector based on the reference sequence and perform Fourier transform to obtain the granularity matrix;

[0038] In this example, a frequency feature vector is constructed based on the reference sequence and Fourier transform is performed to obtain a granularity matrix, specifically:

[0039] The reference sequence is divided into several sequence segments, and the DTW algorithm is used to calculate the DTW distance between each sequence segment;

[0040] The sequence segments are used as nodes and the DTW distance between each sequence segment is used as the edge weight to construct the graph structure;

[0041] The features of the graph structure are aggregated through the preset graph convolution layer to obtain the first feature vector;

[0042] Perform Fourier transform on the first eigenvector to obtain a spectrum matrix;

[0043] The frequency matrix is probabilistically fused, and a granularity matrix is constructed based on a preset convolutional neural network.

[0044] It should be noted that the reference sequence is time series data with a length of 1000 and a sampling frequency of 50 Hz, representing frequency changes within 20 seconds. The reference sequence is divided into 10 segments, each containing 100 sampling points. The DTW distance between each of these 10 segments is calculated to generate a 10×10 distance matrix. For example, the DTW distance between segments 1 and 2 is 15.2, and that between segments 1 and 3 is 30.8. And so on, to obtain the distance matrix.

[0045] In this example, the frequency matrix is probabilistically fused, and a granularity matrix is constructed based on a preset convolutional neural network, specifically:

[0046] Obtain the historical access logs of the power agent, extract the page access frequencies of the same process in different time slices, and construct an access probability vector;

[0047] Align the spectrum matrix and access probability vector according to the preset time window, and calculate the frequency domain energy characteristics of each time window;

[0048] constructing a second feature vector based on the frequency domain energy feature and the access probability vector;

[0049] The second eigenvector is input into the preset convolutional neural network to obtain the granularity matrix.

[0050] For example, the page access frequency of a power intelligent agent process in the past 6 time slices (each time slice is 1 second) is shown in the following table: (Unit: number of accesses):

[0051]

[0052] Normalize the access frequencies of each page in each time slice to construct an access probability vector. For example, the access probability vector for time slice 1 is [0.5, 0.2, 0.1, 0.05, 0.15].

[0053] S4, input the granularity matrix into the preset multi-objective optimization function, and use the NSGA-III evolutionary algorithm to evaluate the memory allocation granularity;

[0054] In this example, the granularity matrix is input into the preset multi-objective optimization function, and the NSGA-III evolutionary algorithm is used to evaluate the memory allocation granularity, specifically:

[0055] Perform singular value decomposition on the granularity matrix to obtain singular value vectors, and slice the granularity matrix based on the singular value vectors to obtain several memory block mapping structure matrices;

[0056] Construct a multi-objective optimization function for each memory block mapping structure matrix and encode it as the initial population individual of the NSGA-III algorithm;

[0057] Iteratively perform crossover and mutation operations to calculate the multi-objective optimization function value of each individual in the initial population;

[0058] Based on the non-dominated sorting and multi-objective optimization function value mechanism, the Pareto front solution set is output to obtain the optimal memory allocation granularity.

[0059] It should be noted that the granularity matrix is first subjected to singular value decomposition (SVD) to extract singular value vectors that reflect the core characteristics and structural information of the matrix. Then, the granularity matrix is sliced based on the singular value vectors to generate multiple memory block mapping structure matrices, each of which corresponds to a different memory allocation unit. For each block matrix, a multi-objective optimization function is constructed and encoded as an initial population individual of the NSGA-III algorithm. By iteratively performing crossover and mutation operations, the multi-objective function value of each individual is calculated, and a non-dominated sorting mechanism is used to screen the Pareto front solution set, ultimately obtaining the optimal memory allocation granularity that meets both resource utilization efficiency and security isolation. This method achieves accurate modeling and dynamic adjustment of complex memory access behavior, improving the intelligence and refinement of process isolation protection for power intelligent entities.

[0060] S5, based on the evaluation results, the agent's process memory space is divided into several isolated areas, and the memory capacity of each isolated area is optimized by quantum annealing to obtain the optimal memory capacity;

[0061] In this example, the agent's process memory space is divided into several isolated areas based on the evaluation results, specifically:

[0062] Based on the optimal memory allocation granularity, obtain the page access frequency of the agent's process within several preset historical time slices and construct an access density vector;

[0063] The access density vector is segmented using a sliding window mechanism to obtain the first access density vectors of several windows;

[0064] Perform cluster analysis on the first access density vectors of several windows, and calculate the mean and standard deviation of page access frequency of each cluster;

[0065] The agent's process memory space is divided into several isolation areas based on the ratio of the page access frequency mean and standard deviation of each cluster.

[0066] The calculation formula for the mean page access frequency of the cluster is as follows:

[0067]

[0068] in, is the mean access density of the k-th cluster, is the memory page set of the k-th cluster, is the access frequency of the tth memory page per unit time.

[0069] The calculation formula for the standard deviation of page access frequency of clusters is as follows:

[0070]

[0071] in, is the standard deviation of the access density of the kth cluster.

[0072] It should be noted that, first, based on the optimal memory allocation granularity, page access frequency data for the agent process over multiple historical time slices is collected to construct an access density vector. The "first access density vector" here refers to the local access density features obtained by segmenting the continuous access frequency data using a sliding window mechanism, which reflects the dynamic changes in memory access behavior. Next, cluster analysis is performed on these segmented access density vectors to identify sets of memory pages with similar access patterns. The mean and standard deviation of the page access frequency for each cluster are calculated, and the ratio of the mean to the standard deviation is used to determine access stability and concentration. Finally, based on these statistical features, the agent process's memory space is partitioned into several "isolation zones": a group of independent memory blocks with similar access behavior. These isolation zones are physically or logically separated, effectively limiting the spread of abnormal accesses and achieving refined and dynamically adjustable security isolation. This approach enables the system to dynamically partition memory zones based on actual access behavior patterns, making memory protection more precise and efficient, significantly improving the security capabilities of the power agent process.

[0073] In this example, quantum annealing optimization is performed on the memory capacity of each isolated region to obtain the optimal memory capacity, specifically:

[0074] A Hamiltonian model is defined for each isolated region, and a quantum annealing algorithm is used to solve the Hamiltonian ground state in the solution space of the Hamiltonian model to obtain a first candidate capacity solution set;

[0075] Constructing a Voroni graph based on the first candidate capacity solution set, and dividing the solution space into several spaces based on the Voroni graph;

[0076] Calculate the solution density of each space, and randomly perturb the first candidate capacity solution set based on the solution density to obtain the second candidate capacity solution set;

[0077] The second candidate capacity solution set is applied to the next round of quantum annealing optimization, and the local optimal solution is jumped out through the tunneling effect to output the optimal memory capacity.

[0078] The specific calculation formula of the Hamiltonian model is as follows:

[0079]

[0080] in, is the Hamiltonian of the ith isolated region, is the current memory capacity configuration value of the i-th isolation area, The target capacity recommended for the current region, is the memory access variance within the region, is the communication weight between region i and region j, is the current memory capacity configuration value of the jth isolation area, is the set of other regions that share a boundary or adjacency with the i-th region.

[0081] It should be noted that a Hamiltonian model is first constructed for each isolation region. In quantum computing, the Hamiltonian describes the energy state of a system. The quantum annealing algorithm simulates quantum tunneling to search for the lowest-energy ground-state solution within the solution space defined by the Hamiltonian—in other words, the optimal memory capacity allocation. The resulting candidate capacity solutions are then used to construct a Voronoi diagram, a spatial partitioning method that divides the solution space into regions, each dominated by a corresponding candidate solution point, facilitating analysis of the distribution and density of solutions. The density of solutions within each region is further calculated based on the Voronoi diagram, and random perturbation techniques are used to adjust the candidate solutions to avoid local optima. Through continuous iterations of the quantum annealing process, a global search and optimization of memory capacity is achieved, ultimately outputting the optimal memory capacity configuration that both meets functional requirements and provides strong defensive capabilities. This optimization method, combining quantum annealing with the Voronoi diagram, effectively improves the intelligence and security of memory capacity allocation within isolation regions.

[0082] S5, deploys the optimal memory capacity to the memory protection unit to obtain an isolation barrier, and integrates the isolation barrier into the sandbox to achieve secure isolation.

[0083] In this example, the optimal memory capacity is deployed to the memory protection unit to obtain an isolation barrier, and the isolation barrier is integrated into the sandbox to achieve secure isolation. Specifically:

[0084] Load the optimal memory capacity vector into the memory protection unit and configure the base address register and boundary register;

[0085] Generate hardware isolation barriers based on base address register and boundary register mapping;

[0086] Embed a hardware isolation barrier in the sandbox kernel to monitor the memory access requests of the agent process in real time;

[0087] When illegal cross-zone access is detected, a security interrupt is triggered and redirected to the sandbox isolation log pool.

[0088] It should be noted that the calculated optimal memory capacity vector is loaded into the memory protection unit, and the starting address and size of each isolation area are specified by configuring the base address register and boundary register, thereby generating an isolation barrier for the physical memory at the hardware level. This isolation barrier is embedded in the sandbox kernel and can monitor the intelligent agent process's access requests to memory in real time. Once illegal cross-region access behavior is detected, the system immediately triggers a security interrupt and redirects the event to the sandbox isolation log pool for recording and processing. In this way, seamless integration from the optimal allocation of memory capacity to the hardware isolation barrier is achieved, and a dynamic, accurate, and efficient memory access protection mechanism is constructed, which greatly improves the security protection capabilities of the power intelligent agent process and the overall stability of the system.

[0089] Example 2, Figure 2 The present invention provides a sandbox-based power intelligent process isolation protection system, which includes a sequence generation module, a sequence screening module, a matrix generation module, an optimization decision module, a capacity optimization module, and a process isolation module:

[0090] A sequence generation module is used to obtain the power system power frequency time series data, extract the current frequency fluctuation curve and construct a first query sequence;

[0091] A sequence screening module, configured to perform sequence screening on the first query sequence using a DTW algorithm to obtain a reference sequence;

[0092] A matrix generation module is used to construct a frequency feature vector based on a reference sequence and perform Fourier transform to obtain a granularity matrix;

[0093] The optimization decision module is used to input the granularity matrix into the preset multi-objective optimization function and use the NSGA-III evolutionary algorithm to evaluate the memory allocation granularity;

[0094] The capacity optimization module is used to divide the agent's process memory space into several isolated areas based on the evaluation results, and perform quantum annealing optimization on the memory capacity of each isolated area to obtain the optimal memory capacity;

[0095] The process isolation module is used to deploy the optimal memory capacity to the memory protection unit, generate an isolation barrier, and integrate the isolation barrier into the sandbox to achieve secure isolation.

[0096] The above formulas are all dimensionless and numerical calculations. The formulas are obtained by collecting a large amount of data and performing software simulation to obtain the most recent real situation. The preset parameters in the formulas are set by technicians in this field according to actual conditions.

[0097] The above embodiments may be implemented in whole or in part through software, hardware, firmware, or any other combination thereof. When implemented using software, the above embodiments may be implemented in whole or in part in the form of a computer program product.

[0098] Those skilled in the art will appreciate that the modules and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0099] In addition, each functional module in each embodiment of the present application may be integrated into one processing module, or each module may exist physically separately, or two or more modules may be integrated into one module.

[0100] The above description is merely a specific embodiment of the present application, but the scope of protection of the present application is not limited thereto. Any changes or substitutions that can be easily conceived by a person skilled in the art within the technical scope disclosed in this application should be included in the scope of protection of this application. Therefore, the scope of protection of this application should be based on the scope of protection of the claims.

[0101] Finally: The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A sandbox-based power intelligent body process isolation and protection method, characterized in that: The following steps are involved: Obtaining power system power frequency time series data, extracting the current frequency fluctuation curve to construct a first query sequence; The DTW algorithm is used to perform sequence screening on the first query sequence to obtain a reference sequence; Construct frequency feature vectors based on the reference sequence and perform Fourier transform to obtain the granularity matrix; The granularity matrix is input into the preset multi-objective optimization function, and the NSGA-III evolutionary algorithm is used to evaluate the memory allocation granularity. Based on the evaluation results, the agent's process memory space is divided into several isolated areas, and the memory capacity of each isolated area is optimized by quantum annealing to obtain the optimal memory capacity; Deploy the optimal memory capacity to the memory protection unit, generate an isolation barrier, and integrate the isolation barrier into the sandbox to achieve secure isolation; Based on the evaluation results, the agent's process memory space is divided into several isolated areas, specifically: Based on the optimal memory allocation granularity, obtain the page access frequency sequence of the agent process in the preset historical time slice and construct the access density vector; The access density vector is segmented using a sliding window mechanism to generate the first access density sub-vectors of several windows; Performing cluster analysis on the first access density sub-vector and calculating the mean and standard deviation of page access frequency of each cluster; The agent's process memory space is divided into several isolation areas based on the ratio of the mean and standard deviation of access frequency.

2. The sandbox-based power intelligent body process isolation and protection method according to claim 1 is characterized in that: The extracting of the current frequency fluctuation curve to construct the first query sequence is specifically as follows: Based on the power system power frequency time series data, the frequency fluctuation curve is extracted through an adaptive sliding window; Perform multi-scale decomposition on the frequency fluctuation curve to obtain several long-range correlation components; A first query sequence is constructed based on a plurality of long correlation components.

3. The sandbox-based power intelligent body process isolation and protection method according to claim 2 is characterized in that: The DTW algorithm is used to perform sequence screening on the first query sequence to obtain a reference sequence, specifically: Obtain a candidate sequence set from a historical frequency fluctuation database; The DTW algorithm is used to calculate the DTW distance between each candidate sequence and the first query sequence, and the first query sequence is screened according to the DTW distance to obtain a reference sequence.

4. The sandbox-based power intelligent body process isolation and protection method according to claim 3 is characterized in that: The frequency feature vector is constructed based on the reference sequence and Fourier transform is performed to obtain the granularity matrix, which is specifically: The reference sequence is divided into several sequence segments, and the DTW algorithm is used to calculate the DTW distance between each sequence segment; Build a graph structure with sequence segments as nodes and DTW distance as edge weight; The graph structure is aggregated through a preset graph convolution layer to generate the first feature vector; Perform Fourier transform on the first eigenvector and output the spectrum matrix; The frequency matrix is probabilistically fused, and a granularity matrix is constructed based on a preset convolutional neural network.

5. The sandbox-based power intelligent body process isolation and protection method according to claim 4 is characterized in that: The frequency matrix is probabilistically fused, and a granularity matrix is constructed based on a preset convolutional neural network, specifically: Obtain the historical access logs of the power agent, extract the page access frequency of the same process in different time slices, and construct the access probability vector; Align the spectrum matrix and access probability vector according to the preset time window, and calculate the frequency domain energy characteristics of each time window; constructing a second feature vector based on the frequency domain energy feature and the access probability vector; The second eigenvector is input into the preset convolutional neural network to obtain the granularity matrix.

6. The sandbox-based power intelligent body process isolation and protection method according to claim 5 is characterized in that: The granularity matrix is input into a preset multi-objective optimization function, and the memory allocation granularity is evaluated using the NSGA-III evolutionary algorithm, specifically: Perform singular value decomposition on the granularity matrix and extract the singular value vector; Slice the granularity matrix according to the singular value vector to generate several memory block mapping structure matrices; Construct a multi-objective optimization function for each memory block mapping structure matrix and encode it as the initial population individual of the NSGA-III algorithm; Iteratively perform crossover and mutation operations to calculate the multi-objective optimization function value of each individual; Based on the non-dominated sorting and multi-objective optimization function value mechanism, the Pareto front solution set is output to obtain the optimal memory allocation granularity.

7. The sandbox-based power intelligent body process isolation and protection method according to claim 6 is characterized in that: The memory capacity of each isolated region is optimized by quantum annealing to obtain the optimal memory capacity, specifically: A Hamiltonian model is defined for each isolated region, and a quantum annealing algorithm is used to solve the Hamiltonian ground state in the solution space of the Hamiltonian model to obtain the first candidate capacity solution set; Constructing a Voroni graph based on the first candidate capacity solution set, and dividing the solution space into several spaces based on the Voroni graph; Calculate the solution density of each space, and randomly perturb the first candidate capacity solution set based on the solution density to obtain the second candidate capacity solution set; The second candidate capacity solution set is applied to the next round of quantum annealing optimization, and the local optimal solution is jumped out through the tunneling effect to output the optimal memory capacity.

8. The sandbox-based power intelligent body process isolation and protection method according to claim 7 is characterized in that: The optimal memory capacity is deployed to the memory protection unit, an isolation barrier is generated, and the isolation barrier is integrated into the sandbox to achieve secure isolation, specifically: Load the optimal memory capacity vector into the memory protection unit and configure the base address register and boundary register; Generate hardware isolation barriers based on base address register and boundary register mapping; Embed a hardware isolation barrier in the sandbox kernel to monitor the memory access requests of the agent process in real time; When illegal cross-zone access is detected, a security interrupt is triggered and redirected to the sandbox isolation log pool.

9. A sandbox-based power intelligent body process isolation and protection system, applying the sandbox-based power intelligent body process isolation and protection method according to any one of claims 1 to 8, characterized in that: It includes sequence generation module, sequence screening module, matrix generation module, optimization decision module, capacity optimization module and process isolation module: A sequence generation module is used to obtain the power system power frequency time series data, extract the current frequency fluctuation curve and construct a first query sequence; A sequence screening module, configured to perform sequence screening on the first query sequence using a DTW algorithm to obtain a reference sequence; A matrix generation module is used to construct a frequency feature vector based on a reference sequence and perform Fourier transform to obtain a granularity matrix; The optimization decision module is used to input the granularity matrix into the preset multi-objective optimization function and use the NSGA-III evolutionary algorithm to evaluate the memory allocation granularity; The capacity optimization module is used to divide the agent's process memory space into several isolated areas based on the evaluation results, and perform quantum annealing optimization on the memory capacity of each isolated area to obtain the optimal memory capacity; The process isolation module is used to deploy the optimal memory capacity to the memory protection unit, generate an isolation barrier, and integrate the isolation barrier into the sandbox to achieve secure isolation.

Images