A method for constructing semantic information flow in power systems in response to complex information attacks

Abstract

This invention provides a method for constructing semantic information flow in power systems in response to complex information attacks, relating to the field of power system network security and information communication modeling technology. The method involves: acquiring initial state information, semantic communication paths, and network attack characteristics; determining the type of semantic communication element based on the network attack characteristics; calculating the information age increment and semantic error increment based on the initial state information and semantic communication element type; performing path accumulation processing based on the semantic computability criterion to obtain the cumulative information age and cumulative semantic error at the sink node; and constructing the power system semantic information flow based on the cumulative information age and cumulative semantic error at the sink node. This invention solves the problems of existing power system network attack modeling methods failing to characterize the coupling relationship and non-superposition distortion effect between sub-attacks, and lacking a unified computational approach.

Description

Technical Field

[0001] This invention relates to the field of power system network security and information communication modeling technology, and in particular to a method for constructing semantic information flow in power systems in response to complex information attacks. Background Technology

[0002] The widespread application of advanced information and communication technologies has enabled modern power systems to efficiently manage large-scale and geographically dispersed demand-side resources, such as solar power generation, flexible loads, and electric vehicles. However, the deep integration of cyber-physical systems inevitably introduces various network vulnerabilities, posing a serious threat to the safe and stable operation of power systems. Typical cybersecurity incidents demonstrate that, compared to single-form cyberattacks, complex information attacks, such as those employing spear-phishing, malware, denial-of-service attacks, and manipulation commands simultaneously, pose extremely high operational risks to the power grid, and may even lead to widespread and prolonged power outages. Therefore, it is urgent to conduct risk analysis for such threats and develop corresponding mitigation strategies.

[0003] Currently, existing research on addressing cyberattacks on power systems primarily follows a one-to-one modeling and defense paradigm. Specifically, for each specific cyberattack, such as spoofing, denial-of-service (DoS) attacks, or replay attacks, researchers typically develop separate, targeted mathematical models to generate hypothetical attack scenarios and assess their individual impacts. For example, DoS attacks are characterized by segmented state signals, or delay attacks are characterized by incorporating time-delay functions from power plant dynamic models. While this control method, based on independent parameters or event triggers, has shown good defensive performance against single-type attacks and effectively characterizes the unique features of individual attacks, it has revealed significant limitations in complex environments.

[0004] With the continuous large-scale integration of demand-side resources, the amount of data transmitted in the power system has increased significantly, and communication endpoints have become more diversified and exhibit strong geographical distribution characteristics. Power communication networks are gradually evolving from dedicated private infrastructure to a new hybrid architecture combining private and public networks. This transformation significantly expands the system's risk exposure, and multi-vector composite network attacks are likely to become the main form of threat in the future. Under composite attacks, the number of possible attack combinations grows exponentially, and their propagation paths in cyber-physical networks often exhibit strong coupling. In this context, the traditional paradigm of a single attack or a single model is completely inapplicable because this paradigm assumes that the attack effects are approximately a simple superposition, failing to effectively characterize the interactions, masking effects, amplification and suppression relationships between different sub-attacks, as well as the nonlinear superposition effects on the power system, leading to risk assessment results that deviate significantly from actual operating conditions. Summary of the Invention

[0005] To address the aforementioned shortcomings in existing technologies, this invention provides a method for constructing semantic information flow in power systems in response to complex information attacks. This method solves the problem that existing power system network attack modeling methods often adopt a paradigm of independently modeling single attacks. When facing complex information attacks under a hybrid communication architecture, these methods cannot characterize the coupling relationship and non-superposition distortion effects between sub-attacks and lack a unified computational approach.

[0006] To achieve the aforementioned objectives, the technical solution adopted by this invention is: a method for constructing semantic information flow in a power system oriented towards complex information attacks, comprising:

[0007] S1: Obtain the initial state information sent by the source node of the power system, the semantic communication path between the source node and the sink node, and the network attack characteristics of the composite information attack against the power system. The semantic communication path is composed of multiple semantic communication links.

[0008] S2: Based on the network attack characteristics of composite information attacks, determine the semantic communication element type corresponding to each semantic communication link;

[0009] S3: Based on the initial state information and the semantic communication element type corresponding to each semantic communication link, calculate the information age increment and information semantic error increment corresponding to each semantic communication link respectively;

[0010] S4: Based on the semantic computability criterion, the information age increment and information semantic error increment corresponding to all semantic communication links on the semantic communication path are respectively processed by path accumulation to obtain the cumulative information age and cumulative information semantic error at the sink node.

[0011] S5: Based on the accumulated information age and accumulated information semantic error at the sink node, the semantic information flow of the power system is constructed, thus completing the construction of the semantic information flow of the power system.

[0012] Further, S2 includes:

[0013] By analyzing the characteristics of network attacks, we can identify the types of impacts that complex information attacks have on communication links.

[0014] When the impact type is either increased data transmission time or interrupted status update, the semantic communication element type is determined to be information age modulator;

[0015] When the impact type is tampering with the semantic content of information, the semantic communication element type is determined to be an information semantic error modulator;

[0016] When the impact type includes both time delay and semantic content tampering, the semantic communication element type is determined to be a hybrid modulator.

[0017] Further, S3 includes:

[0018] The initial state information is used as the link input state information corresponding to each semantic communication link;

[0019] When the semantic communication element type is determined to be an information age modulator, the time drift caused by the increased data transmission time in the network attack characteristics is calculated based on the sampling interval of the link input state information. The change in the state probability distribution is determined based on the reference time offset caused by the time drift. Based on the time drift and the change in the state probability distribution, the information age increment and the information semantic error increment corresponding to the information age modulator are calculated.

[0020] When the semantic communication element type is determined to be an information semantic error modulator, the semantic distortion function in the network attack characteristics is used to tamper with the link input state information, obtain the probability density function of the false data at the receiving end, and calculate the information semantic error increment corresponding to the information semantic error modulator based on the initial probability distribution of the link input state information and the probability density function of the false data at the receiving end, and determine that the information age increment corresponding to the information semantic error modulator is zero.

[0021] When the semantic communication element type is determined to be a hybrid modulator, the order of action of time drift and semantic distortion in the network attack characteristics is determined. When time drift occurs first and semantic distortion occurs later, a distortion function is applied at the reference time where the offset occurs to process the link input state information to obtain the first processed state information. When semantic distortion occurs first and time drift occurs later, a distortion function is applied at the original reference time and then time drift is superimposed to process the link input state information to obtain the second processed state information. Based on either the first processed state information or the second processed state information, the information age increment corresponding to the hybrid modulator and the information semantic error increment corresponding to the hybrid modulator are calculated.

[0022] Furthermore, the expressions for the information age increment corresponding to the information age modulator and the information semantic error increment corresponding to the information age modulator are respectively:

[0023] ;

[0024] ;

[0025] in, This represents the information age increment corresponding to the information age modulator. This represents the increment of the semantic error corresponding to the information age modulator. This indicates the time drift caused by the increased data transmission time. The sampling interval represents the link input status information. Represents a node The generation time of the most recently received message obtained based on the timestamp. This represents a positive-zero integer parameter obtained based on the time drift span. Represents a node Information age at the location, Represents a node Information age at the location, This represents the link input state information obtained based on the system physical model at time [time]. The probability density function, Represents the logarithmic function. Indicates the current moment. Represents a node Corresponding to the initial state probability distribution before the time shift occurs, Represents a node This corresponds to the state probability distribution after the time shift occurs.

[0026] Furthermore, the expression for the information semantic error increment corresponding to the information semantic error modulator is:

[0027] ;

[0028] in, This represents the increment of the information semantic error corresponding to the information semantic error modulator. The probability density function representing the link input state information. Represents a node Information age at the location, This represents the initial probability distribution before the time shift occurred. The probability density function representing spoofed data at the receiving end. Indicates the current moment;

[0029] The expression for the false data received is:

[0030] ;

[0031] in, Represents a node Based on false data status obtained through communication reception methods, The semantic distortion function represents the characteristics of a network attack. Represents a node The original, unaltered state information Represents a node Information age.

[0032] Furthermore, the expression for the information semantic error increment corresponding to the hybrid modulator is:

[0033] When time drift occurs first, followed by semantic distortion:

[0034] ;

[0035] When semantic distortion occurs first, followed by time drift:

[0036] ;

[0037] in, This represents the semantic error increment corresponding to the hybrid modulator. The probability density function representing the link input state information. Represents the initial probability distribution. The probability density function represents the first processing state information. The probability density function representing the second processing state information;

[0038] The expressions for the first processing status information and the second processing status information are as follows:

[0039] ;

[0040] ;

[0041] in, This indicates the first processing status information. This indicates the second processing status information. This represents a semantic distortion function. This indicates the state information corresponding to the reference time at which the offset occurred. This indicates the status information corresponding to the original reference time. Represents a node Information age at the location, Represents a node Information age at the location, Indicates the current moment.

[0042] Further, S4 includes:

[0043] Based on the semantic computability criterion, the information age increment and information semantic error increment of each semantic communication link are linearly accumulated sequentially along the directed communication path to obtain the path accumulation result.

[0044] Analyzing the path accumulation results yields the accumulated information age, which reflects the overall aging of the transmission path from the source node to the sink node, and the accumulated information semantic error, which reflects the overall semantic deviation.

[0045] Furthermore, the expressions for the accumulated information age and the accumulated information semantic error are respectively:

[0046] ;

[0047] ;

[0048] in, This indicates the accumulated information age. This indicates the cumulative semantic error of the information. This represents the limit on the total number of nodes on the semantic communication link obtained based on path topology information. Indicates the link sequence number. This represents the information age increment of a single semantic communication link. This represents the increment of semantic error in a single semantic communication link. This represents the information age of the sink node at time t. This represents the information age of the source node at time t. This represents the cumulative semantic error of the sink node at time t. This represents the initial semantic error of the source node at time t.

[0049] Further, S5 includes:

[0050] Construct a semantic transfer matrix based on the communication network topology of the power system;

[0051] By using the semantic transfer matrix to perform network-based aggregation of the cumulative information age and cumulative information semantic error of multiple semantic communication paths, the network-wide coupling distortion state parameters are obtained.

[0052] Based on the network-wide coupling distortion state parameters, a power system semantic information flow characterizing the network-wide coupling distortion state is constructed, thus completing the construction of the power system semantic information flow.

[0053] Furthermore, the expression for the network-wide coupling distortion state parameter is:

[0054] ;

[0055] ;

[0056] in, This represents the first compact vector composed of the age increments of all information on the internet. The second compact vector represents the semantic error increment of the entire network information. The first and second compact vectors together constitute the network-wide coupling distortion state parameters. Represents the semantic transfer matrix. This represents a vector form representing the cumulative information age of multiple semantic communication paths. It represents the vector form of the cumulative semantic errors of multiple semantic communication paths.

[0057] The beneficial effects of this invention are as follows: This invention provides a method for constructing semantic information flow in power systems to counter complex information attacks. By calculating the information age increment and semantic error increment for each link based on state information and the semantic communication element type corresponding to each semantic communication link, the method then introduces and performs path accumulation processing based on semantic computability criteria, clarifying the evolutionary laws of information obsolescence and content tampering along the communication path. This breaks the existing situation where the calculation logic of each model is independent, ensuring the consistency of the two types of quantitative indicators in the full-link calculation, and exhibiting strong scalability.

[0058] After acquiring the characteristics of network attacks, the semantic communication element types corresponding to each semantic communication link are determined based on these characteristics. This changes the inefficient approach of traditional models that require rebuilding a massive equation system for each attack combination. Through this mechanism, flexible adaptive calculations can be performed for network attacks with arbitrary multi-vectors, significantly improving compatibility with new and derivative network attacks.

[0059] Finally, based on the accumulated information age and accumulated information semantic error at the sink node, a semantic information flow of the power system oriented towards composite information attacks is constructed. Since incremental calculations were performed on each component type in the preliminary steps and the accumulation followed the semantic computability criterion, this construction method no longer treats the attack impact as a simple superposition, but can accurately capture and reproduce the interaction and masking effect between the sub-attacks within the composite attack, making the analysis results more consistent with the actual distortion conditions of the real power grid when subjected to composite attacks. Attached Figure Description

[0060] This specification will be further described by way of exemplary embodiments, which will be described in detail with reference to the accompanying drawings. These embodiments are not limiting; in these embodiments, the same reference numerals denote the same structures, wherein:

[0061] Figure 1 This is an exemplary flowchart illustrating a method for constructing semantic information flow in a power system against complex information attacks, according to some embodiments of this specification.

[0062] Figure 2 This is an exemplary schematic diagram illustrating an analogy between a power transmission element and a semantic communication element according to some embodiments of this specification;

[0063] Figure 3 These are exemplary schematic diagrams of four semantic communication elements according to some embodiments of this specification; wherein, (a) an information age modifier; (b) an information semantic modifier; and (c)-(d) a hybrid modifier;

[0064] Figure 4 These are exemplary schematic diagrams of the power and communication topologies of node systems according to some embodiments of this specification; wherein, (a) is a power topology diagram of an IEEE 14-node system; (b) is a communication topology diagram of an IEEE 14-node system; (c) is a power topology diagram of an IEEE 39-node system; and (d) is a communication topology diagram of an IEEE 39-node system. Detailed Implementation

[0065] The specific embodiments of the present invention are described below to enable those skilled in the art to understand the present invention. However, it should be understood that the present invention is not limited to the scope of the specific embodiments. For those skilled in the art, various changes are obvious as long as they are within the spirit and scope of the present invention as defined and determined by the appended claims. All inventions utilizing the concept of the present invention are protected.

[0066] Example

[0067] Figure 1 This is an exemplary flowchart illustrating a method for constructing semantic information flow in a power system to counter complex information attacks, based on some embodiments of this specification. Figure 1 As shown, the process includes the following steps. In some embodiments, the process may be executed by a processor.

[0068] S1: Obtain the initial state information sent by the source node of the power system, the semantic communication path between the source node and the sink node, and the network attack characteristics of the composite information attack against the power system. The semantic communication path is composed of multiple semantic communication links.

[0069] Initial state information is raw data about the underlying physical operation of a power system's source nodes, generated before they are subjected to any network communication interference or attacks. For example, initial state information may include uncontaminated basic electrical quantities such as the actual active power output of generator sets, the voltage amplitude and phase angle of nodes, and real-time current measurements of transformers.

[0070] In some embodiments, the processor can acquire physical operating parameters of power components in real time through sensors and measuring devices deployed at the bottom of the power system, and convert these acquired analog or digital signals into initial data packets in a standard format to obtain initial state information.

[0071] A semantic communication path is a connected route of directed communication nodes used to transmit state updates about the same physical state from a source node to a sink node. For example, a semantic communication path may include a complete network data transmission line starting from a source generator measurement node, passing through intermediate routers and switch nodes in multiple substations, and finally reaching the dispatch control center server node.

[0072] In some embodiments, the processor can identify all intermediate network nodes and their sequential connection order that a data packet generated by a specific source node must pass through to reach a specific sink node by reading the power system's communication network configuration table or routing protocol forwarding table, thereby obtaining the semantic communication path. For example... Figure 2 As shown, Figure 2 This paper illustrates an analogy between power transmission elements and semantic communication elements in describing port behavior. In traditional power systems, the port behavior of power transmission elements is typically described by physical quantities at their input and output terminals, such as voltage and power. Similarly, in the semantic information flow model of this invention, semantic communication elements can be characterized by the corresponding transformations of information age and information semantic error between the input and output terminals. Specifically, this element acts as an operator, transforming the information age and information semantic error of the message at the input terminal of the communication link into a corresponding binary tuple at the output terminal.

[0073] Network attack characteristics describe the specific interference attributes exhibited by malicious intrusion behaviors against power system communication networks in the information transmission dimension. For example, network attack characteristics may include the waiting time value of maliciously delaying data packet transmission, the duration of blockage caused by interrupting data updates, and the drift coefficient and injected spurious noise patterns used to tamper with measurement data.

[0074] In some embodiments, the processor can continuously monitor network traffic through an intrusion detection system deployed on communication network nodes, analyze packet arrival time deviations, abnormal surges in packet loss rates, and abnormal mutations in message payload content, and compare them with normal communication baseline characteristics to obtain network attack characteristics.

[0075] A semantic communication link is the basic transmission unit that constitutes a complete communication path, responsible for connecting two adjacent communication nodes in a network. For example, a semantic communication link may include a physical fiber optic channel and related logical transmission protocol connection channels connecting a local aggregation switch of a wind farm with its upstream regional dispatch center gateway router.

[0076] In some embodiments, the processor can extract any two nodes that have a direct data exchange relationship by parsing the topology diagram of the communication network, and identify their port connection status, IP address allocation and link bandwidth configuration information, thereby obtaining the semantic communication link.

[0077] S2: Based on the network attack characteristics of composite information attacks, determine the semantic communication element type corresponding to each semantic communication link.

[0078] Semantic communication element types are abstract classifications of damaged communication links based on the specific distortion effects caused by network attacks on information flows. For example, semantic communication element types may include information age modulator types that only increase the staleness of data, information semantic error modulator types that only destroy the authenticity of data, and hybrid modulator types that contain both types of destructive effects.

[0079] In some embodiments, the processor can determine whether the attack suffered by the current link has caused time delay, data tampering, or a combination of both by matching a preset attack impact rule base. Based on the determination result, the processor classifies the corresponding damaged link into a specific element category, thereby obtaining the semantic communication element type. For example... Figure 3 As shown, Figure 3 Four semantic communication element structures for characterizing heterogeneous network attacks are presented. (a) shows an information age regulator, which characterizes factors affecting information timeliness (such as delay attacks and denial-of-service attacks). While it does not directly modify semantic content, it indirectly causes semantic distortion by increasing data transmission time. (b) shows an information semantic regulator, which represents factors causing direct semantic distortion of information (such as spoofing attacks and replay attacks). It alters the probability distribution by applying a semantic distortion function to the information but does not affect timeliness. (c) and (d) show hybrid regulators, representing the combined effect of the above two regulators acting on the same communication link. (c) is the first type of hybrid regulator, where the information age regulator acts first (time drift), followed by the information semantic regulator (semantic distortion). (d) is the second type of hybrid regulator, where the information semantic regulator first generates distortion, and then the information age regulator increases the information's obsolescence.

[0080] In some embodiments, the processor can analyze network attack characteristics to obtain the type of impact of composite information attacks on the communication link; when the impact type is either increased data transmission time or interruption status update, the semantic communication element type is determined to be an information age modulator; when the impact type is tampering with information semantic content, the semantic communication element type is determined to be an information semantic error modulator; when the impact type includes both time delay and semantic content tampering, the semantic communication element type is determined to be a hybrid modulator.

[0081] Impact type refers to the specific physical and logical manifestations of interference caused by a network attack directly affecting the information flow transmission process of a power system. For example, impact types can include increased transmission time due to data packet queuing, interruption of state updates due to network connection disconnection, and semantic information tampering caused by directly rewriting message content through spoofed data injection.

[0082] In some embodiments, the processor can use deep packet inspection technology and timestamp comparison technology to detect the delivery status of data packets and the integrity verification results of the payload content, and determine whether the data stream has encountered network-level blocking, dropping or application-level malicious overwriting, thereby obtaining the type of impact.

[0083] An information age modulator is a semantic communication component used to characterize, and only characterize, the single dimension of increased information staleness caused by a network attack. For example, an information age modulator may include operator components for characterizing the indefinite blocking of data packets caused by a denial-of-service attack, or operator components for characterizing the significantly delayed arrival time of data packets caused by a delay attack.

[0084] In some embodiments, the processor can instantiate and assign values ​​to a standardized time effect module template that contains only time drift parameters and not content tampering functions after identifying a time delay or interruption type attack suffered by the link, thereby obtaining an information age modulator.

[0085] An information semantic error modulator is a semantic communication component used to characterize, and only characterize, the single dimension of distortion or deviation in data content caused by a network attack. For example, an information semantic error modulator may include a mathematical distortion operation module for simulating a hacker intercepting transmission midway and superimposing Gaussian noise onto generator active power measurements and setting a linear drift coefficient.

[0086] In some embodiments, the processor can extract the distortion function expression from the attack characteristics after identifying the data tampering type attack suffered by the link, and encapsulate it into a standardized content destruction module template that does not affect the timestamp attribute and specifically changes the data probability distribution, thereby obtaining an information semantic error modulator.

[0087] A hybrid modulator is a composite semantic communication component used to characterize a communication link that suffers simultaneous or sequential damage in both time and content dimensions. For example, a hybrid modulator may include a data lag time operator caused by a delay attack and a content tampering distortion operator caused by spoofed data injection, combined in a specific order to form a comprehensive damage module.

[0088] In some embodiments, when multiple attack features are detected superimposed on a single link, the processor can extract the corresponding time drift parameters and distortion function parameters, confirm the time sequence of their effects based on network logs, and combine and encapsulate them to obtain a hybrid modulator.

[0089] S3: Based on the initial state information and the semantic communication element type corresponding to each semantic communication link, calculate the information age increment and information semantic error increment corresponding to each semantic communication link.

[0090] Information age increment is the amount of stale time that data packets experience when traversing a specific semantic communication link due to inherent network transmission time or disruption caused by network attacks. For example, information age increment can include the inherent routing delay of tens of milliseconds during normal transmission, or the additional several seconds or even longer abnormal delay caused by being forced to wait in the router buffer during a delay attack.

[0091] In some embodiments, the processor can obtain the information age increment by recording the input timestamp of the data packet entering a specific communication link and subtracting the output timestamp of the data packet leaving the link, or by directly performing algebraic assignment operations based on the delay parameters preset by the attack model.

[0092] The semantic error increment is the degree of deviation in the accuracy of data information representing the true state of a physical system before and after it flows through a specific semantic communication link, due to malicious tampering. For example, the semantic error increment can include the divergence change value calculated between the probability distribution of the normal state at the link input end and the probability distribution of the damaged state at the link output end after being injected with false data. The larger this value, the more severe the deviation in semantic content.

[0093] In some embodiments, the processor can obtain the information semantic error increment by acquiring the original probability density function of the data before it enters the link, and the spurious data probability density function of the data after it passes through the link and is subjected to the distortion function. The processor can then use the relative entropy calculation formula to perform integral calculation on these two distribution functions.

[0094] In some embodiments, the processor can use the initial state information as the link input state information corresponding to each semantic communication link; when the semantic communication element type is determined to be an information age modulator, based on the sampling interval of the link input state information, the time drift caused by the increased data transmission time in the network attack features is calculated, and the change in the state probability distribution is determined based on the reference time offset caused by the time drift. Based on the time drift and the change in the state probability distribution, the information age increment corresponding to the information age modulator and the information semantic error increment corresponding to the information age modulator are calculated; when the semantic communication element type is determined to be an information semantic error modulator, the semantic distortion function in the network attack features is used to tamper with the link input state information to obtain the probability density function of the false data at the receiving end, based on the initial probability distribution of the link input state information and the change in the state probability distribution. The probability density function of the received spurious data is used to calculate the information semantic error increment corresponding to the information semantic error modulator, and the information age increment corresponding to the information semantic error modulator is determined to be zero. When the semantic communication element type is determined to be a hybrid modulator, the order of action of time drift and semantic distortion in the network attack characteristics is determined. When time drift occurs first and semantic distortion occurs later, the distortion function is applied at the reference time where the offset occurs to process the link input state information to obtain the first processed state information. When semantic distortion occurs first and time drift occurs later, the distortion function is applied at the original reference time and the time drift is superimposed to process the link input state information to obtain the second processed state information. Based on either the first processed state information or the second processed state information, the information age increment corresponding to the hybrid modulator and the information semantic error increment corresponding to the hybrid modulator are calculated.

[0095] Link input state information refers to the data content of a data packet that represents a physical quantity of the power system at a certain reference moment, carried when it is about to enter a specific semantic communication link for transmission. For example, for the first link initiated by the source node, the link input state information may include the newly generated initial value of the undamaged generator power; for links in the middle of the path, it may be intermediate state data that has been delayed or tampered with by the previous link.

[0096] In some embodiments, the processor can obtain link input status information by extracting output port data cache information from the upstream node immediately adjacent to the specific link, or by directly reading the initial generated measurement data from the source node sensor when the link is the starting link.

[0097] Time drift is caused by cyberattacks, resulting in an abnormal lag between the actual physical state of the data received by the receiving end and the expected time. For example, time drift could include a one-second forced communication delay set by an attacker, causing the dispatch control center to receive data reflecting the generator's operating state one second prior, rather than the current real-time state.

[0098] In some embodiments, the processor can obtain the time drift by analyzing blocking or delay instructions in the network attack characteristics to set parameters, or by comparing the time difference between the baseline arrival time of the same information flow in a non-attack baseline environment and the actual arrival time in the current attacked environment.

[0099] Changes in the state probability distribution are probabilistic models reflecting the random fluctuations in the physical state quantities of a power system. These changes occur due to shifts in the distribution shape and location caused by misalignment of the observation reference time or alteration of its content. For example, changes in the state probability distribution may include an overall shift in the mean of the normal distribution of system data due to the injection of linear bias, or a broadening of the curve caused by an increase in the distribution variance due to the superposition of additional random noise.

[0100] In some embodiments, the processor can obtain the change in the state probability distribution by comparing the characteristics of the baseline state probability density function curve when it is not attacked with the characteristics of the target damaged probability density function curve obtained by recalculating and fitting after the time offset caused by delay or direct tampering with the distortion function.

[0101] The information age increment corresponding to the information age modulator is the additional time loss calculated based on the internal logic of the time modulator after determining that the link is affected by a simple delay or interruption. For example, the information age increment corresponding to the information age modulator can include the increment when no new data is received within the normal sampling interval. Specifically, this increment is manifested as a continuously widening and step-increasing time difference between the current time and the last successfully received data generation time.

[0102] In some embodiments, the processor can determine whether the generation timestamp of the currently received message has changed by obtaining the preset time drift parameters and the sampling interval time of the system state information in the information age modulator, and perform algebraic assignment and judgment operations according to the time step function rules, thereby obtaining the information age increment corresponding to the information age modulator.

[0103] The semantic error increment corresponding to the information age modulator is a semantic cognitive bias indirectly caused by the fact that the received value cannot accurately reflect the current physical state, even though the content has not been directly tampered with. For example, the semantic error increment corresponding to the information age modulator may include the calculated divergence deviation between the two state distributions when the generator power probability distribution five seconds ago is used to estimate the current actual power distribution because the data is lagging by five seconds.

[0104] In some embodiments, the processor can obtain the stale time difference between the node input and output due to the action of the information age modulator, use this difference to determine two probability density functions before and after the state shift, and then obtain the information semantic error increment corresponding to the information age modulator by performing divergence integration on these two distribution functions.

[0105] In some embodiments, the expressions for the information age increment corresponding to the information age modulator and the information semantic error increment corresponding to the information age modulator are as follows:

[0106] ;

[0107] ;

[0108] in, This represents the information age increment corresponding to the information age modulator. This represents the increment of the semantic error corresponding to the information age modulator. This indicates the time drift caused by the increased data transmission time. The sampling interval represents the link input status information. Represents a node The generation time of the most recently received message obtained based on the timestamp. This represents a positive-zero integer parameter obtained based on the time drift span. Represents a node Information age at the location, Represents a node Information age at the location, This represents the link input state information obtained based on the system physical model at time [time]. The probability density function, Represents the logarithmic function. Indicates the current moment. Represents a node Corresponding to the initial state probability distribution before the time shift occurs, Represents a node This corresponds to the state probability distribution after the time shift occurs.

[0109] A probability density function is a mathematical distribution model used to quantitatively describe the probability of a power system state variable taking a specific value within its continuous value space. For example, a probability density function can include a continuous bell-shaped curve mathematical equation used to accurately depict the probability of generator active power measurements fluctuating around their true physical values, assuming that measurement communication noise follows a Gaussian distribution.

[0110] In some embodiments, the processor can collect a large amount of steady-state measurement data from historical normal operation phases, use statistical methods and power system physical dynamic constraint equations to perform data fitting analysis, determine the mean and variance parameters of state variable fluctuations, and then construct a continuous mathematical expression to obtain the probability density function.

[0111] The initial probability distribution is the baseline, unaltered statistical distribution characteristic of the physical state variables of a power system before information enters the communication link affected by an attack. For example, the initial probability distribution may include the pure Gaussian distribution baseline formed by the fluctuation of sensor measurements of the generator output power at the source node around its actual physical value, considering only the influence of natural equipment measurement white noise.

[0112] In some embodiments, the processor can obtain the initial probability distribution by reading the baseline model of the probability density function of the source node measurement end or the state information of a specific link input end when no network attack characteristics are applied, and extracting the original distribution curve determined by the baseline mean and baseline variance parameters.

[0113] The information semantic error increment corresponding to the information semantic error modulator is a severe cognitive bias in the system state caused by a distortion operation directly applied by an attack that specifically implements content tampering. For example, the information semantic error increment corresponding to the information semantic error modulator may include a huge divergence value generated in the calculation between the originally correct state probability distribution and the severely mutated probability distribution due to the attacker forcibly multiplying the generator output reading by two and mixing in additional noise.

[0114] In some embodiments, the processor can use the unaltered initial probability distribution function and the receiver false data probability density function obtained after being subjected to a specific semantic distortion function as input parameters, and substitute them into the information semantic error integral operation module to calculate the relative entropy, thereby obtaining the information semantic error increment corresponding to the information semantic error modulator.

[0115] In some embodiments, the expression for the information semantic error increment corresponding to the information semantic error modulator is:

[0116] ;

[0117] in, This represents the increment of the information semantic error corresponding to the information semantic error modulator. The probability density function representing the link input state information. Represents a node Information age at the location, This represents the initial probability distribution before the time shift occurred. The probability density function representing spoofed data at the receiving end. Indicates the current moment;

[0118] The expression for the false data received is:

[0119] ;

[0120] in, Represents a node Based on false data status obtained through communication reception methods, The semantic distortion function represents the characteristics of a network attack. Represents a node The original, unaltered state information Represents a node Information age.

[0121] The first processed state information is an intermediate, damaged data form created in a hybrid attack scenario where data packets first experience a transmission time lag, and then undergo content tampering based on this lag reference time. For example, the first processed state information may include a severely lagging generator power value that originally described the physical state one second ago, which has been further forcibly superimposed with spoofed high-frequency noise and drift coefficients to form a composite distorted signal payload.

[0122] In some embodiments, after determining the order of action of the composite attack, the processor first shifts the reference time point of the data backward based on the time drift parameter, and then substitutes the stale state information of the corresponding shifted time point into the attacker's semantic distortion function for calculation and transformation, thereby obtaining the first processing state information.

[0123] The second processing state information occurs in hybrid attack scenarios where data packets are maliciously rewritten at their initial generation. This corrupted content then encounters time delays in the transmission network, resulting in another form of damaged data. For example, the second processing state information could include forged generator output data that is immediately altered to double its average value upon generation at the source, and then further delayed for two seconds in the communication queue before being sent, resulting in distorted and outdated signal data.

[0124] In some embodiments, the processor can determine the order of action of the composite attack, first apply the attacker's semantic distortion function to mutate the content of the data at the original baseline reference time point, then retain the probability distribution after the mutation is damaged, and superimpose the time drift parameter set by the information age modulator to obtain the second processing state information.

[0125] The information age increment corresponding to a hybrid modulator is the additional stale time independently contributed by the sub-mechanism responsible for hindering transmission within a complex component that is susceptible to both delay and tampering attacks. For example, the information age increment corresponding to a hybrid modulator could include the cumulative waiting time of three sampling periods on a communication line that is both modified and congested, where the state update data packet is delayed solely due to the congestion.

[0126] In some embodiments, the processor can extract the influence of semantic distortion from the integrated processing logic of the hybrid modulator, and separately extract the additional time parameters or sampling interval lag multiple introduced by its internal time drift module for assignment calculation, thereby obtaining the information age increment corresponding to the hybrid modulator.

[0127] The semantic error increment corresponding to a hybrid modulator is a measure of the total deviation in the system's comprehensive understanding of the physical state under the combined effects of time and content attacks. For example, the semantic error increment corresponding to a hybrid modulator may include the overall divergence difference between the receiver distribution and the original distribution caused by the reference time shift due to extremely outdated data and the probability distribution being tampered with due to severe distortion.

[0128] In some embodiments, the processor can select, according to the determined order of action of the composite attack, to substitute the damaged probability density function corresponding to the first processing state information or the second processing state information together with the probability density function of the initial state before the time offset into the divergence integral formula for comprehensive calculation, thereby obtaining the information semantic error increment corresponding to the hybrid modulator.

[0129] In some embodiments, the expression for the semantic error increment corresponding to the hybrid modulator is:

[0130] When time drift occurs first, followed by semantic distortion:

[0131] ;

[0132] When semantic distortion occurs first, followed by time drift:

[0133] ;

[0134] in, This represents the semantic error increment corresponding to the hybrid modulator. The probability density function representing the link input state information. Represents the initial probability distribution. The probability density function represents the first processing state information. The probability density function representing the second processing state information;

[0135] The expressions for the first processing status information and the second processing status information are as follows:

[0136] ;

[0137] ;

[0138] in, This indicates the first processing status information. This indicates the second processing status information. This represents a semantic distortion function. This indicates the state information corresponding to the reference time at which the offset occurred. This indicates the status information corresponding to the original reference time. Represents a node Information age at the location, Represents a node Information age at the location, Indicates the current moment.

[0139] In some embodiments, node The expression for the semantic error at this location is:

[0140] ;

[0141] in, Represents a node semantic errors in the information at that location Indicates the source node At any moment The probability density function of the initial state information sent. Represents a node The probability density function of the received damaged state information. Represents the logarithmic function. An integral infinitesimal element representing a state variable.

[0142] In some embodiments, node The expression for the information age and its derivation are as follows:

[0143] ;

[0144] ;

[0145] in, Represents a node Information age at the location, Indicates the current moment. Represents a node The generation time of the most recently received message obtained based on the timestamp. Indicates the source node Generate the first The initial generation time of the message. Indicates the first The message actually arrived at the node Arrival time, This represents the function that takes the maximum value.

[0146] In some embodiments, the expression for defining the semantic increment of a single semantic communication link is:

[0147] ;

[0148] ;

[0149] in, This represents the increment of semantic error in a single semantic communication link. Indicates the output node of the link semantic errors in the information at that location Indicates the link input node Semantic errors in the information at the location; This represents the information age increment of a single semantic communication link. Indicates the output node of the link Information age at the location, Indicates the link input node Information age.

[0150] S4: Based on the semantic computability criterion, the information age increment and information semantic error increment corresponding to all semantic communication links on the semantic communication path are respectively processed by path accumulation to obtain the cumulative information age and cumulative information semantic error at the sink node.

[0151] Accumulated information age is the sum of all normal transmission times and additional delays caused by attacks on each node along the entire directed communication path from the source node to the sink node. For example, accumulated information age can include the total end-to-end extremely stale time accumulated when generator status data passes through five routers, where two hops have extremely short normal transmission times, and the other three hops are delayed for several seconds due to severe delay attacks, finally reaching the control center.

[0152] In some embodiments, the processor can obtain the cumulative information age by traversing all nodes and links on a specific semantic communication path from the source node to the sink node and performing sequential algebraic addition on the information age increment data calculated by the time element model for each semantic communication link.

[0153] Cumulative semantic error is the final total deviation formed when semantic distortions accumulate due to various network attacks or channel degradation encountered along the way during the entire network transmission process from the source node to the sink node. For example, cumulative semantic error may include the maximum deviation of the overall probability distribution divergence formed when generator data experiences two spoofing attacks with different coefficients along the path, and these tampering effects are superimposed at the control center receiver.

[0154] In some embodiments, the processor can obtain the cumulative semantic error by sequentially summing the incremental semantic errors of each damaged semantic communication link connected in series on the directed communication path, following the path attribute conservation relationship in the semantic computability criterion.

[0155] In some embodiments, the processor can perform a linear accumulation operation on the information age increment and information semantic error increment of each semantic communication link along the directed communication path based on the semantic computability criterion to obtain the path accumulation result; analyze the path accumulation result to obtain the cumulative information age reflecting the overall aging of the transmission path from the source node to the sink node and the cumulative information semantic error reflecting the overall semantic deviation.

[0156] The path accumulation result is an intermediate computational dataset that aggregates and summarizes the time obsolescence and content degradation dimensions generated by all intermediate communication links along a specific end-to-end data transmission path. For example, the path accumulation result may include a composite data packet output structure that summarizes and calculates the total delay time scalar value and the total semantic divergence scalar value for a specific dedicated line from Generator 1 to the control center.

[0157] In some embodiments, the processor can activate a directed path tracing algorithm to collect the incremental data of information age and the incremental data of information semantic error output by all semantic communication link modules involved in the communication path, and perform a step-by-step summation process to obtain the path accumulation result.

[0158] In some embodiments, the expressions for cumulative information age and cumulative information semantic error are respectively:

[0159] ;

[0160] ;

[0161] in, This indicates the accumulated information age. This indicates the cumulative semantic error of the information. This represents the limit on the total number of nodes on the semantic communication link obtained based on path topology information. Indicates the link sequence number. This represents the information age increment of a single semantic communication link. This represents the increment of semantic error in a single semantic communication link. This represents the information age of the sink node at time t. This represents the information age of the source node at time t. This represents the cumulative semantic error of the sink node at time t. This represents the initial semantic error of the source node at time t.

[0162] In some embodiments, the path additive relationship derivation expressions for cumulative information semantic error and cumulative information age are as follows:

[0163] ;

[0164] ;

[0165] in, This indicates an accumulation operation. This indicates the number of intermediate relay nodes. and Equivalent representation of the source node, and Equivalent representation of sink node.

[0166] In some embodiments, in the semantic information flow model of a single information flow, the expressions for the information age increment vector and the information semantic error increment vector are:

[0167] ;

[0168] ;

[0169] in, Indicates the source node The corresponding information age increment vector along the path of the information flow. Indicates the source node The corresponding semantic error increment vector along the information flow path, This represents a single-stream semantic transfer matrix constructed based on the adjacency matrix. Indicates the source node Related information: age vector Indicates the source node Related semantic error vector.

[0170] S5: Based on the accumulated information age and accumulated information semantic error at the sink node, the semantic information flow of the power system is constructed, thus completing the construction of the semantic information flow of the power system.

[0171] The semantic information flow of a power system is a global dynamic network model parameter that describes how the accuracy and timeliness of multi-source heterogeneous state update data evolve and degrade as it propagates through complex power communication topologies. For example, the semantic information flow of a power system can include a set of evolutionary equations for the state reporting paths of all generators in the entire network. This set comprehensively reflects the distributed parameter situation where the entire dispatch and control center's understanding of the underlying physical reality of the power grid is completely distorted under multi-point composite attacks.

[0172] In some embodiments, the processor can obtain the semantic information flow of the power system by combining the accumulated calculation model parameters of all active end-to-end single communication paths in the power network with the topology matrix representing the network structure and concatenating and integrating the state vectors.

[0173] In some embodiments, the processor can construct a semantic transfer matrix based on the communication network topology of the power system; use the semantic transfer matrix to perform networked aggregation processing on the cumulative information age and cumulative information semantic error of multiple semantic communication paths to obtain the whole network coupling distortion state parameters; based on the whole network coupling distortion state parameters, construct the power system semantic information flow characterizing the whole network coupling distortion state, and complete the construction of the power system semantic information flow.

[0174] Communication network topology is a graph-theoretic structure describing the connection relationships and communication architecture layout of various device nodes in the underlying information infrastructure of a power system that supports data interaction. For example, communication network topology may include a directed graph node adjacency matrix data table that records how switches, regional aggregation routers, and dispatch control center gateways at various levels of substations are interconnected via optical fiber or wireless channels.

[0175] In some embodiments, the processor can obtain the communication network topology by sending topology discovery protocol probe packets to the power communication network management and monitoring system, or by directly importing a static database file of network device connections pre-configured and maintained by the system administrator and performing protocol parsing. For example... Figure 4 As shown, Figure 4 The power and communication topologies of the IEEE 14-node and IEEE 39-node systems used to verify the effectiveness of the model of this invention are shown. (a) and (c) respectively present the power topologies of the two systems. Combined with the communication topologies corresponding to (b) and (d), it can be seen that each generator can be considered a data source, reporting status information to the control center along the communication path. The red components in (b) and (d) indicate communication links subjected to network attacks; in the actual model, these compromised links are modeled as... Figure 3 The information age regulator, information semantic regulator, or hybrid regulator shown are used to reflect the propagation path and coupling effect of multi-vector network attacks in cyber-physical networks.

[0176] The semantic transitivity matrix is ​​established based on the underlying communication network topology. It is a core operator array used to mathematically describe and map the semantic impairment propagation relationships and multipath coupling effects between nodes. For example, the semantic transitivity matrix can include a square matrix parameter with the total number of nodes in the network as its dimension. Its internal elements are generated by subtracting the transpose matrix related to the adjacency relationship of a single information flow route from the identity matrix, which determines how a specific impairment is matrix-propagated throughout the network.

[0177] In some embodiments, the processor can generate a corresponding adjacency matrix by extracting the adjacency relationships of the routing nodes through which a single information flow passes, then perform matrix subtraction between the system identity matrix and the transpose of the adjacency matrix to generate a single-flow transfer matrix, and finally use the block diagonal operator to aggregate all single-flow transfer matrices to obtain a semantic transfer matrix.

[0178] The network-wide coupling distortion state parameter is a comprehensive set of metrics used to macroscopically and compactly quantify the damage to all data streams in the entire power system under multi-vector, multi-node concurrent composite network attacks. For example, the network-wide coupling distortion state parameter may include a system-level comprehensive matrix parameter composed of a first compact column vector summed from the information age increments accumulated from all communication information streams in the network, and a second compact column vector summed from all semantic error increments.

[0179] In some embodiments, the processor can obtain the network coupling distortion state parameters by performing linear algebra operations such as matrix multiplication on the set of vectors representing the cumulative information age and cumulative information semantic error of all paths in the network and the semantic transfer matrix representing the network connection structure of the entire network.

[0180] In some embodiments, the expression for the network-wide coupling distortion state parameter is:

[0181] ;

[0182] ;

[0183] in, This represents the first compact vector composed of the age increments of all information on the internet. The second compact vector represents the semantic error increment of the entire network information. The first and second compact vectors together constitute the network-wide coupling distortion state parameters. Represents the semantic transfer matrix. This represents a vector form representing the cumulative information age of multiple semantic communication paths. It represents the vector form of the cumulative semantic errors of multiple semantic communication paths.

[0184] In some embodiments, the expression for the power system physical measurement benchmark model used to assess the impact of cyberattacks is:

[0185] ;

[0186] in, This represents the normal measured value of the generator set's active power. This indicates the actual active power output of the generator set. This represents the inherent measurement noise of the system.

[0187] In some embodiments, the expression for the receive state model under a linear spoofing attack is:

[0188] ;

[0189] in, This indicates that the receiver obtained forged active power status data after being subjected to a fake data injection attack. This represents the drift coefficient set in the network attack characteristics. This represents the additive noise term injected by the attacker.

[0190] In some embodiments, the expression for the receive state model under a denial-of-service attack is:

[0191] ;

[0192] in, This indicates outdated active power status data that the control center was forced to retain after a denial-of-service attack disrupted status updates. This indicates the moment when data was last successfully updated, as obtained from the system logs.

[0193] In some embodiments, the increased data transmission time causes time drift. The range of values ​​is The specific value depends on the strength setting of the delayed attack, for example, in some attack scenarios. Possible values Sampling interval for link input status information The value of is greater than 0 and is determined by the hardware sampling frequency of the power system measurement equipment, which is usually in the range of milliseconds to seconds. The parameter k obtained based on the time drift span is an integer greater than or equal to 1 and is used to characterize the time delay spanning several complete sampling periods.

[0194] In some embodiments, the drift coefficient set in the network attack characteristics A real number not equal to 1, used to characterize the degree to which an attacker linearly amplifies or reduces the true value of a physical quantity. In specific simulation scenarios, The values ​​can be sequentially set to 1.5, 2, 2.5, 3, etc., to simulate progressively increasing spoofed data injection attack strength; inherent measurement noise in the system. and additive noise terms injected by the attacker All of them follow a standard normal distribution, that is, their range of values ​​conforms to... The probability distribution characteristics.

Claims

1. A method for constructing semantic information flow in a power system to counter complex information attacks, characterized in that, include: S1: Obtain the initial state information sent by the source node of the power system, the semantic communication path between the source node and the sink node, and the network attack characteristics of the composite information attack against the power system. The semantic communication path is composed of multiple semantic communication links. S2: Based on the network attack characteristics of composite information attacks, determine the semantic communication element type corresponding to each semantic communication link; S3: Based on the initial state information and the semantic communication element type corresponding to each semantic communication link, calculate the information age increment and information semantic error increment corresponding to each semantic communication link respectively; S4: Based on the semantic computability criterion, the information age increment and information semantic error increment corresponding to all semantic communication links on the semantic communication path are respectively processed by path accumulation to obtain the cumulative information age and cumulative information semantic error at the sink node. S5: Based on the accumulated information age and accumulated information semantic error at the sink node, the semantic information flow of the power system is constructed, thus completing the construction of the semantic information flow of the power system.

2. The method for constructing semantic information flow in a power system against complex information attacks according to claim 1, characterized in that, S2 includes: By analyzing the characteristics of network attacks, we can identify the types of impacts that complex information attacks have on communication links. When the impact type is either increased data transmission time or interrupted status update, the semantic communication element type is determined to be information age modulator; When the impact type is tampering with the semantic content of information, the semantic communication element type is determined to be an information semantic error modulator; When the impact type includes both time delay and semantic content tampering, the semantic communication element type is determined to be a hybrid modulator.

3. The method for constructing semantic information flow in a power system against complex information attacks according to claim 2, characterized in that, S3 includes: The initial state information is used as the link input state information corresponding to each semantic communication link; When the semantic communication element type is determined to be an information age modulator, the time drift caused by the increased data transmission time in the network attack characteristics is calculated based on the sampling interval of the link input state information. The change in the state probability distribution is determined based on the reference time offset caused by the time drift. Based on the time drift and the change in the state probability distribution, the information age increment and the information semantic error increment corresponding to the information age modulator are calculated. When the semantic communication element type is determined to be an information semantic error modulator, the semantic distortion function in the network attack characteristics is used to tamper with the link input state information, obtain the probability density function of the false data at the receiving end, and calculate the information semantic error increment corresponding to the information semantic error modulator based on the initial probability distribution of the link input state information and the probability density function of the false data at the receiving end, and determine that the information age increment corresponding to the information semantic error modulator is zero. When the semantic communication element type is determined to be a hybrid modulator, the order of action of time drift and semantic distortion in the network attack characteristics is determined. When time drift occurs first and semantic distortion occurs later, a distortion function is applied at the reference time where the offset occurs to process the link input state information to obtain the first processed state information. When semantic distortion occurs first and time drift occurs later, a distortion function is applied at the original reference time and then time drift is superimposed to process the link input state information to obtain the second processed state information. Based on either the first processed state information or the second processed state information, the information age increment corresponding to the hybrid modulator and the information semantic error increment corresponding to the hybrid modulator are calculated.

4. The method for constructing semantic information flow in a power system against complex information attacks according to claim 3, characterized in that, The expressions for the information age increment and the information semantic error increment corresponding to the information age modulator are as follows: ； ； in, This represents the information age increment corresponding to the information age modulator. This represents the increment of the semantic error corresponding to the information age modulator. This indicates the time drift caused by the increased data transmission time. The sampling interval represents the link input status information. Represents a node The generation time of the most recently received message obtained based on the timestamp. This represents a positive-zero integer parameter obtained based on the time drift span. Represents a node Information age at the location, Represents a node Information age at the location, This represents the link input state information obtained based on the system physical model at time [time]. The probability density function, Represents the logarithmic function. Indicates the current moment. Represents a node Corresponding to the initial state probability distribution before the time shift occurs, Represents a node This corresponds to the state probability distribution after the time shift occurs.

5. The method for constructing semantic information flow in a power system against complex information attacks according to claim 3, characterized in that, The expression for the information semantic error increment corresponding to the information semantic error modulator is: ； in, This represents the increment of the information semantic error corresponding to the information semantic error modulator. The probability density function representing the link input state information. Represents a node Information age at the location, This represents the initial probability distribution before the time shift occurred. The probability density function representing spoofed data at the receiving end. Indicates the current moment; The expression for the false data received is: ； in, Represents a node Based on false data status obtained through communication reception methods, The semantic distortion function represents the characteristics of a network attack. Represents a node The original, unaltered state information Represents a node Information age.

6. The method for constructing semantic information flow in a power system against complex information attacks according to claim 3, characterized in that, The expression for the semantic error increment corresponding to the hybrid modulator is: When time drift occurs first, followed by semantic distortion: ； When semantic distortion occurs first, followed by time drift: ； in, This represents the semantic error increment corresponding to the hybrid modulator. The probability density function representing the link input state information. Represents the initial probability distribution. The probability density function represents the first processing state information. The probability density function representing the second processing state information; The expressions for the first processing status information and the second processing status information are as follows: ； ； in, This indicates the first processing status information. This indicates the second processing status information. This represents a semantic distortion function. This indicates the state information corresponding to the reference time at which the offset occurred. This indicates the status information corresponding to the original reference time. Represents a node Information age at the location, Represents a node Information age at the location, Indicates the current moment.

7. The method for constructing semantic information flow in a power system against complex information attacks according to claim 1, characterized in that, S4 includes: Based on the semantic computability criterion, the information age increment and information semantic error increment of each semantic communication link are linearly accumulated sequentially along the directed communication path to obtain the path accumulation result. Analyzing the path accumulation results yields the accumulated information age, which reflects the overall aging of the transmission path from the source node to the sink node, and the accumulated information semantic error, which reflects the overall semantic deviation.

8. The method for constructing semantic information flow in a power system against complex information attacks according to claim 7, characterized in that, The expressions for the cumulative information age and the cumulative information semantic error are as follows: ； ； in, This indicates the accumulated information age. This indicates the cumulative semantic error of the information. This represents the limit on the total number of nodes on the semantic communication link obtained based on path topology information. Indicates the link sequence number. This represents the information age increment of a single semantic communication link. This represents the increment of semantic error in a single semantic communication link. This represents the information age of the sink node at time t. This represents the information age of the source node at time t. This represents the cumulative semantic error of the sink node at time t. This represents the initial semantic error of the source node at time t.

9. The method for constructing semantic information flow in a power system against complex information attacks according to claim 1, characterized in that, S5 includes: Construct a semantic transfer matrix based on the communication network topology of the power system; By using the semantic transfer matrix to perform network-based aggregation of the cumulative information age and cumulative information semantic error of multiple semantic communication paths, the network-wide coupling distortion state parameters are obtained. Based on the network-wide coupling distortion state parameters, a power system semantic information flow characterizing the network-wide coupling distortion state is constructed, thus completing the construction of the power system semantic information flow.

10. The method for constructing semantic information flow in a power system against complex information attacks according to claim 9, characterized in that, The expression for the network-wide coupling distortion state parameter is: ； ； in, This represents the first compact vector composed of the age increments of all information on the internet. The second compact vector represents the semantic error increment of the entire network information. The first and second compact vectors together constitute the network-wide coupling distortion state parameters. Represents the semantic transfer matrix. This represents a vector form representing the cumulative information age of multiple semantic communication paths. It represents the vector form of the cumulative semantic errors of multiple semantic communication paths.

Images