Nuclear power plant external anomaly identification method and device, electronic equipment and storage medium

Abstract

The application discloses a nuclear power plant external anomaly identification method and device, electronic equipment and storage medium. The method comprises the following steps: collecting a plurality of external field detection signals through a plurality of types of sensors installed outside the nuclear power plant; determining a plurality of target objects and a plurality of target trajectory segments corresponding to the plurality of target objects according to the plurality of external field detection signals; determining a plurality of abnormal nodes according to a plurality of target objects corresponding to a plurality of target trajectory segments according to a preset rule; determining an abnormal association graph according to a plurality of abnormal nodes; determining the probability of harm occurrence of each abnormal node to the nuclear power plant according to the association relationship between the abnormal nodes in the abnormal association graph; and displaying the probability of harm occurrence of each abnormal node. The application can realize accurate acquisition of target trajectories and abnormal association analysis of the external field of the nuclear power plant by fusing multiple sources of sensors, quantify the harm probability of each abnormal node, and significantly improve the identification, evaluation and visualization prevention and control capability of safety risks.

Description

Technical Field

[0001] This invention relates to the field of information processing technology, and in particular to a method, apparatus, electronic device, and storage medium for identifying external anomalies in nuclear power plants. Background Technology

[0002] The importance of nuclear power safety has become increasingly apparent. Nuclear power plants need to rely on multi-sensor field perception to identify abnormal behavior and assess hazards and risks, thereby enhancing their proactive defense capabilities.

[0003] In related technologies, perimeter security systems for nuclear power plants often employ video surveillance combined with infrared or microwave beam sensors, using electronic fences and motion thresholds to trigger intrusion alarms. Some systems introduce multi-sensor fusion technology, utilizing Kalman filtering for target trajectory tracking and identifying anomalies based on single target behavior rules (such as boundary crossing or timeout). However, these methods only handle isolated target behaviors, ignoring the potential spatiotemporal correlations of multiple abnormal events, making it difficult to identify complex threats such as coordinated infiltration and phased reconnaissance. Summary of the Invention

[0004] This invention provides a method, device, electronic device, and storage medium for identifying external anomalies in nuclear power plants, in order to solve the problems of insufficient identification, weak correlation, and difficulty in quantifying the probability of harm in the field of nuclear power plants under multi-source sensing.

[0005] According to one aspect of the present invention, a method for identifying external anomalies in a nuclear power plant is provided, the method comprising:

[0006] Multiple types of field detection signals are collected by various sensors installed outside the nuclear power plant, and multiple target objects and multiple target trajectory segments corresponding to the multiple field detection signals are determined based on the multiple field detection signals;

[0007] Multiple abnormal nodes are determined based on preset rules and multiple target objects corresponding to multiple target trajectory segments. An abnormal association graph is determined based on the multiple abnormal nodes. The probability of each abnormal node causing harm to the nuclear power plant is determined based on the association relationship between the abnormal nodes in the abnormal association graph, and the probability of each abnormal node causing harm is displayed.

[0008] According to another aspect of the present invention, a nuclear power plant external anomaly identification device is provided, comprising:

[0009] The trajectory segment determination module is used to collect various field detection signals from various types of sensors installed outside the nuclear power plant, and to determine multiple target objects and multiple target trajectory segments corresponding to the multiple target objects based on the various field detection signals.

[0010] The hazard occurrence probability determination module is used to determine multiple abnormal nodes based on preset rules and multiple target objects corresponding to multiple target trajectory segments, determine an abnormal association graph based on the multiple abnormal nodes, determine the hazard occurrence probability of each abnormal node to the nuclear power plant based on the association relationship between the abnormal nodes in the abnormal association graph, and display the hazard occurrence probability of each abnormal node.

[0011] According to another aspect of the present invention, an electronic device is provided, the electronic device comprising:

[0012] At least one processor; and

[0013] A memory communicatively connected to the at least one processor; wherein,

[0014] The memory stores a computer program that can be executed by the at least one processor, which enables the at least one processor to perform the nuclear power plant external anomaly identification method according to any embodiment of the present invention.

[0015] According to another aspect of the present invention, a computer-readable storage medium is provided, the computer-readable storage medium storing computer instructions for causing a processor to execute and implement the nuclear power plant external anomaly identification method according to any embodiment of the present invention.

[0016] The technical solution of this invention collects various field detection signals from multiple types of sensors installed outside the nuclear power plant. Based on these signals, multiple target objects and corresponding trajectory segments are determined. This multi-source sensor fusion enables accurate identification and complete trajectory reconstruction of field targets, improving the comprehensiveness and continuity of the nuclear power plant's perimeter situational awareness. Furthermore, based on preset rules and the target objects corresponding to the trajectory segments, multiple abnormal nodes are identified. An abnormal correlation graph is then established, and the probability of each abnormal node posing a hazard to the nuclear power plant is determined based on the correlation relationships between these nodes. This process displays the probability of each abnormal node's hazard, achieving structured correlation and quantitative display of abnormal target behavior. This solves the problems of insufficient identification, weak correlation, and difficulty in quantifying hazard probabilities in the field of nuclear power plants under multi-source sensing. It integrates multi-source sensing to achieve accurate acquisition of target trajectories and abnormal correlation analysis, quantifying the hazard probability of each abnormal node, and significantly improving the identification, assessment, and visualized prevention and control capabilities of safety risks.

[0017] It should be understood that the description in this section is not intended to identify key or essential features of the embodiments of the present invention, nor is it intended to limit the scope of the invention. Other features of the invention will become readily apparent from the following description. Attached Figure Description

[0018] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0019] Figure 1 This is a flowchart of a method for identifying external anomalies in a nuclear power plant according to Embodiment 1 of the present invention;

[0020] Figure 2 This is a flowchart of a method for identifying external anomalies in a nuclear power plant according to Embodiment 2 of the present invention;

[0021] Figure 3 This is a schematic diagram of the structure of an external anomaly identification device for a nuclear power plant according to Embodiment 3 of the present invention;

[0022] Figure 4 This is a schematic diagram of an anomaly association diagram provided according to an embodiment of the present invention;

[0023] Figure 5 This is a schematic diagram of the interface of an identification system provided according to an embodiment of the present invention;

[0024] Figure 6 This is a schematic diagram of the structure of an electronic device that implements the nuclear power plant external anomaly identification method according to an embodiment of the present invention. Detailed Implementation

[0025] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0026] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of the invention described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0027] It should be noted that the terms "a" and "a plurality of" used in this disclosure are illustrative rather than restrictive, and those skilled in the art should understand that, unless otherwise expressly indicated in the context, they should be understood as "one or more".

[0028] The names of messages or information exchanged between multiple devices in the embodiments of this disclosure are for illustrative purposes only and are not intended to limit the scope of such messages or information.

[0029] It is understood that before using the technical solutions disclosed in the various embodiments of this disclosure, users should be informed of the types, scope of use, and usage scenarios of the personal information involved in this disclosure in an appropriate manner in accordance with relevant laws and regulations, and user authorization should be obtained.

[0030] For example, upon receiving a user's active request, a prompt message is sent to the user to explicitly inform them that the requested operation will require the acquisition and use of the user's personal information. This allows the user to independently choose whether to provide personal information to the software or hardware, such as the electronic device, application, server, or storage medium performing the operations of this disclosed technical solution, based on the prompt message.

[0031] As an optional but non-limiting implementation, in response to a user's active request, sending a prompt message to the user can be done via a pop-up window, where the prompt message can be presented in text format. Furthermore, the pop-up window can also include a selection control allowing the user to choose "agree" or "disagree" to provide personal information to the electronic device.

[0032] It is understood that the above notification and user authorization process are merely illustrative and do not constitute a limitation on the implementation of this disclosure. Other methods that comply with relevant laws and regulations may also be applied to the implementation of this disclosure.

[0033] It is understood that the data involved in this technical solution (including but not limited to the data itself, the acquisition or use of the data) shall comply with the requirements of relevant laws, regulations and related provisions.

[0034] Example 1

[0035] Figure 1 The flowchart of the method for identifying external anomalies in nuclear power plants provided in Embodiment 1 of the present invention is applicable to the safety control situation of multi-source sensing, abnormal behavior analysis and hazard risk quantification and early warning in the field of nuclear power plants. The method can be executed by a nuclear power plant external anomaly identification device, which can be implemented in hardware and / or software. Optionally, it can be implemented through electronic devices, such as mobile terminals, PCs or servers.

[0036] like Figure 1 As shown, the method may specifically include:

[0037] S110. Collect various field detection signals through various types of sensors installed outside the nuclear power plant, and determine multiple target objects and multiple target trajectory segments corresponding to the multiple target objects based on the various field detection signals.

[0038] The nuclear power plant, as described above, can be understood as an industrial facility that uses nuclear energy (such as nuclear fission) to generate heat energy, which is then converted into electrical energy through steam turbines and generators. As the subject of monitoring and protection, its safe and stable operation depends on the perception of anomalies in the external environment—identifying potential threats (such as intruders, hazardous materials transport vehicles, etc.) by collecting external field signals, and providing early warnings to reduce accident risks. The various types of sensors can be understood as a collection of devices used to collect different external field signals, including but not limited to video detection sensors (collecting image / video signals), radar sensors (collecting target distance, speed, and azimuth signals), infrared sensors, etc. (collecting thermal radiation signals). "Multimodal perception" compensates for the limitations of single sensors (such as cameras being susceptible to lighting conditions and radar being unable to identify target types), achieving comprehensive and redundant detection of complex external environments and ensuring the accuracy and completeness of signal acquisition. The external field detection signals can be understood as the raw environmental data collected by sensors outside the nuclear power plant, serving as the "data source" for subsequent analysis. As the "perception layer output," it is the bridge connecting the physical world and digital analysis—all subsequent target identification, trajectory analysis, and anomaly judgment are based on these signals, and their quality directly determines the reliability of the system. Examples include video frames from cameras, range-velocity matrices from radar, and thermal images from infrared sensors. The target object can be understood as the entity of interest identified in the field detection signal, i.e., the object the system needs to track and analyze. By identifying the target object, it is separated from the complex background environment, focusing on "specific entities that may threaten the safety of the nuclear power plant," avoiding interference from invalid information. The target trajectory segment can be understood as a segment of the target object's movement path within a certain time period or spatial region. Decomposing the continuous movement of the target object into "quantifiable path units" facilitates the analysis of its spatiotemporal characteristics.

[0039] S120. Based on preset rules and multiple target objects corresponding to multiple target trajectory segments, determine multiple abnormal nodes, determine an abnormal association graph based on the multiple abnormal nodes, determine the probability of each abnormal node causing harm to the nuclear power plant based on the association relationship between the abnormal nodes in the abnormal association graph, and display the probability of each abnormal node causing harm.

[0040] The preset rules can be understood as predefined anomaly judgment logic defined by the system, used to distinguish between "normal trajectories" and "abnormal trajectories," serving as "judgment criteria for anomaly identification"—transforming the abstract concept of "abnormality" into computable rules, allowing the system to automatically and consistently filter abnormal trajectory segments, avoiding biases from subjective human judgment. Abnormal nodes can be understood as abnormal points / segments within the target trajectory segment, representing key locations or intervals in the trajectory that "deviate from the normal pattern." Abnormal nodes can also be called threat nodes. The anomaly correlation graph can be understood as a graph structure with abnormal nodes as vertices and the relationships between nodes as edges, breaking the isolation of individual abnormal nodes, revealing the inherent connections between abnormal events, helping the system understand the "holistic" rather than "fragmented" nature of anomalies, and providing a more comprehensive context for hazard assessment. The correlation relationships can be understood as the logical or spatiotemporal connections between abnormal nodes, serving as the "edge weights" for constructing the anomaly correlation graph, acting as the "link" of the anomaly correlation graph—through these relationships, the system can connect scattered abnormal nodes into a "meaningful chain of abnormal events," avoiding the omission of "cooperative threats" and improving the systematic nature of the analysis. The probability of hazard occurrence can be understood as a quantitative value of the likelihood that a certain abnormal node will trigger a safety accident at a nuclear power plant. This transforms the "abstract degree of hazard" into a comparable and rankable numerical indicator, helping operation and maintenance personnel to quickly identify "high-priority threats".

[0041] Based on the above scheme, optionally, the step of determining multiple abnormal nodes according to preset rules and multiple target objects corresponding to multiple target trajectory segments includes: resampling multiple target trajectory segments according to a preset duration, determining the similarity between multiple target objects according to the resampling results, aggregating multiple target objects into multiple object clusters according to the similarity and a preset similarity threshold, performing detection based on the multiple object clusters, and determining multiple abnormal nodes according to the detection results.

[0042] The preset duration can be understood as a pre-defined time interval parameter used to divide continuous target trajectory segments into time units of equal length, solving the problem of "inconsistent length and uneven time granularity" among different target trajectory segments. By unifying the time scale (e.g., fixing it to 1 minute / segment), a standardized data structure is provided for subsequent cross-target trajectory similarity comparisons, avoiding similarity calculation deviations caused by differences in time granularity. The resampling can be understood as the operation of re-segmenting / interpolating the original target trajectory segments into time units of equal length according to the preset duration. This achieves standardized processing of target trajectory segments, eliminating the "temporal-spatial discrepancy" caused by different sensor sampling frequencies or changes in target motion speed, allowing trajectory segments of different targets to be compared in the same time dimension. The similarity can be understood as the degree of matching of the features of trajectory segments of two or more target objects. It is a quantitative indicator for measuring the consistency of target behavior patterns and a core basis for identifying target groups with consistent behavior patterns. The similarity threshold can be understood as a pre-set similarity cutoff value used to determine whether the trajectories of two or more target objects are "sufficiently similar" to be classified into the same category. This serves as a "judgment criterion for clustering operations," transforming fuzzy similarity into executable classification rules and avoiding overly coarse clustering. For example, setting the threshold to 0.7 means that targets with a similarity ≥ 0.7 are considered "highly similar" and can be clustered together; targets with a similarity < 0.7 are considered independent objects. The object cluster can be understood as multiple similar target groups formed after aggregation of target objects. Each cluster represents a "set of targets with consistent behavioral patterns," serving as a "unit for anomaly detection"—compared to a single target, the behavior of object clusters is more regular. By analyzing the overall characteristics of the cluster, "large-scale or persistent anomalies" can be identified more efficiently.

[0043] By adopting this technical solution, the granularity of trajectory is unified through resampling, and similar targets are aggregated into object clusters based on similarity. This upgrades the analysis from individual to group, accurately identifies coordinated anomalies, avoids missing scattered threats, improves the accuracy and efficiency of anomaly node determination, and enhances the nuclear power plant's ability to perceive complex external threats.

[0044] Based on the above scheme, optionally, the step of determining the abnormal association graph based on the multiple abnormal nodes includes: determining the collaboration factor between every two abnormal nodes; constructing the abnormal association graph with the multiple abnormal nodes as nodes and the collaboration factor between the abnormal nodes as edges.

[0045] The synergy factor can be understood as an indicator used to quantify the "strength of synergy" between any two anomalous nodes. It reflects the potential "correlation, cooperation, or causality" between the two anomalous nodes in the process of threatening the safety of a nuclear power plant. It transforms the abstract "synergy" into a calculable numerical value, which takes the form of a number (such as 0~1 or a real number). The larger the value, the stronger the synergy between the two nodes. The node can be understood as a "vertex" in the anomalous association graph, corresponding to a specific anomalous node.

[0046] In one optional implementation, a target trajectory segment is received, and multiple trajectory segments are processed simultaneously. Using the target trajectory segment as input, physical targets are identified and materialized to form anomalous nodes. Weighted edges representing the collaboration / association between anomalous nodes are established, and finally, an anomalous association graph G(V,E) for higher-order reasoning is output.

[0047] In the input section, multiple target trajectory fragments are grouped into a set {Tracklet};

[0048] In the output section, in the anomaly correlation graph G(V,E), the first... One abnormal node Each edge .

[0049] In a specific implementation, the target object is first calculated to determine the identifiable abnormal nodes. For example, this calculation process includes the following steps:

[0050] The trajectory segments are further standardized and windowed. Examples include:

[0051] The trajectory segment timestamps are resampled to a fixed step size Δt according to a preset duration. Δt can optionally take the value of 0.1, 0.2, or 0.5 s. The standardized and windowed trajectory segment T is output. k Let k be the k-th processed trajectory segment. Each segment contains the position mean. Average speed Equivalent feature vectors, and the trajectory segment T k confidence level C m-k .

[0052] For example, the average location Average speed It refers to the statistical average of the position and velocity observed by the system for the same target object within a short time window corresponding to the target trajectory segment.

[0053] Data aggregation is performed on each target object. This is based on the similarity of the features of each target object. If the similarity of two or more target objects meets a certain threshold, they are merged into the same anomalous node. .

[0054] More specifically, in an exemplary implementation, the positional similarity of two target objects a and b is... The calculation method can be:

[0055] ;

[0056] In the above formula, , These are vectors representing the average positions of two target objects, a and b, respectively. The Euclidean norm represents the distance between two position coordinates, which is the average distance between the positions of two target objects a and b in space. It is a scale constant, and its value can be taken according to the actual situation. For example, it can be 50m or 100m.

[0057] For example, the speed similarity between two target objects a and b The calculation method can be:

[0058] ;

[0059] Other feature similarity can be calculated using corresponding mathematical methods, which will not be elaborated here.

[0060] Furthermore, a comprehensive similarity sim calculation is performed; the similarity between two target objects a and b is... One exemplary calculation method is as follows:

[0061] ;

[0062] In the above formula, For the first of two target objects Similarity of features; For the corresponding The calculated weight values ​​can be specifically set in practical applications based on the importance of each feature.

[0063] Two or more target objects that meet a similarity threshold are grouped into multiple object clusters. Detection is then performed based on these object clusters, and multiple anomalous nodes are identified based on the detection results. This is done when the similarity between two or more target objects meets a certain threshold. After that, that is, if Then, such target objects can be merged into an abnormal node.

[0064] The target objects mentioned in the above steps are not confirmed abnormal nodes that have been materialized, but rather hypothetical targets with potential threats that are inferred from observational features at the data level.

[0065] Furthermore, the collaboration factor γ between every two anomalous nodes is calculated.

[0066] For example, for two abnormal nodes and Then we have:

[0067] ;

[0068] In the above formula, Two abnormal nodes and The Similarity of features For corresponding The calculated weight values ​​can be specifically set in actual anomaly calculation based on the importance of each feature.

[0069] The weight of the edge is defined as the collaboration factor. , Represents the nodes after quantization calculation relative nodes The synergy or synergistic injury potential between the nodes in terms of spatiotemporal, motion, and behavioral characteristics; the calculation of the synergy factor takes into account the relative position, relative velocity, temporal overlap, and behavioral similarity between the abnormal nodes;

[0070] In the aforementioned anomaly correlation graph, the edge between two anomaly nodes represents the correlation between the two nodes and the degree of correlation; the weighted edge is the weight assigned to the edge. This is used to quantify the "strength of association" or "degree of synergistic behavior." Preferably, the synergistic factor... The value ranges from [0 to 1]. The synergistic factor It can be used to: (1) determine whether a graph algorithm considers two target nodes as related, for example, setting lower limit ,when (1) When the time is considered relevant; (2) In graph neural networks or other aggregation calculations, node information is aggregated by weight.

[0071] By adopting this technical solution, the linkage relationship of abnormal events can be intuitively revealed by quantifying the collaborative factors between abnormal nodes and constructing a correlation graph. This helps to identify collaborative threat chains, improves the systematic nature of anomaly analysis, and provides key support for accurately assessing the probability of harm.

[0072] Based on the above scheme, optionally, determining the probability of each abnormal node posing a hazard to the nuclear power plant according to the association relationship between the abnormal nodes in the abnormal association graph includes: determining the independent abnormal probability of each abnormal node, and determining the synergy factor between each abnormal node and its associated nodes in the abnormal association graph; determining the probability of each abnormal node posing a hazard to the nuclear power plant according to the independent abnormal probability and the synergy factor.

[0073] The independent anomaly probability can be understood as the probability that a single anomalous node poses a threat to the nuclear power plant without considering the influence of other nodes, without considering the impact of other anomalous events.

[0074] This technical solution integrates the independent anomaly probability of nodes with synergistic factors to quantify the amplification effect of multi-node linkage on threats, avoids isolated assessment bias, accurately reflects the actual degree of harm of anomalous nodes, and provides a reliable basis for nuclear power plants to prioritize the handling of high-risk nodes and optimize emergency resources.

[0075] Based on the above scheme, optionally, the probability of each abnormal node posing a hazard to the nuclear power plant can be determined using the following formula:

[0076] ;

[0077] in, Indicates the first The independent anomaly probability of each anomalous node. Indicates the first The potential anomaly strength of each abnormal node; Indicates the first The probability of identifying an abnormal node as an abnormal target; Indicates the first Evaluation values ​​of the spatial geometric relationship between the abnormal nodes and the nuclear power plant; Indicates the first The probability of an abnormal node and its associated nodes being abnormal; This indicates the preset magnification factor; Indicates the cooperating factor; This represents the normalization parameter.

[0078] The potential anomaly intensity can be understood as a measure of the first... The "potential threat level" index is an indicator of the anomalous characteristics of each anomalous node. Its value is directly related to the node's anomalous attributes, such as the target's dangerousness (hazardous goods vehicles > ordinary vehicles), the destructiveness of the behavior (damaging fences > simply loitering), and the duration (long-term stay > short-term stay). The potential anomaly intensity serves as a correction factor for the independent anomaly probability, used to refine the threat differences within each node. The evaluation value can be understood as the [missing value]. The threat correlation score (quantified value, a positive real number, with a higher score for closer proximity and proximity to sensitive areas) between anomaly nodes and the spatial location of the nuclear power plant serves as a quantitative indicator of spatial threat, reflecting the impact of "location" on the probability of hazard. It is understandable that the safety risks of a nuclear power plant are highly dependent on the spatial location of anomaly nodes (e.g., anomalies closer to the core area are more dangerous than those in the suburbs). The amplification factor can be understood as a pre-set amplification factor of the synergistic effect, used to control the "amplification intensity" of the synergistic factor on the probability of hazard, avoiding over- or under-calculation of the synergistic effect. As an adjustment parameter for the synergistic effect, it allows the model to adjust the weight of synergistic threats according to the actual safety strategy of the nuclear power plant. The normalization parameter can be understood as an adjustment parameter used to limit the calculation results to a reasonable range (e.g., [0,1]), serving as a guarantee of probability compliance and ensuring that the output results conform to the basic definition of probability (value <1). Without normalization, contradictory results of "probability > 1" may occur, leading to misleading decisions (e.g., mistakenly regarding a node with a 120% probability as "inevitable").

[0079] In an alternative implementation, further, in damage quantification, an amplification factor can be introduced. To introduce synergy, with projects The amplification factor is incorporated into subsequent damage probability calculations to quantify the synergistic amplification effect of the two threat nodes. Preferably, the amplification coefficient... The values ​​can be obtained through empirical statistics from simulation experiments or by using artificial intelligence simulations, based on the characteristics of the identified threat nodes after training.

[0080] like Figure 4 The aforementioned includes three threat nodes 201, namely threat nodes A, B, and C. The cooperation factors between each pair of the three threat nodes are calculated. , , The synergy factor is calculated based on the orientation, velocity, mass, and other values ​​of the three entities, and is further used in the following steps to calculate the damage to the factory area. A spatial coordinate system can be established using the factory area as a reference. For example, see attached... Figure 4 The image only shows planar coordinates x and y. In practical applications, the coordinate system can be extended to spatial coordinates to obtain accurate calculation results.

[0081] Furthermore, the probability of harm or injury is calculated using the following formula. :

[0082] The probability of injury This refers to combining an abnormal node. Independent anomaly probability , and inter-node collaboration factors Calculate the probability of hazard occurrence in sensitive units within the factory area:

[0083] ;

[0084] In the above formula, where, Indicates the first The independent anomaly probability of each anomalous node. Indicates the first The potential anomaly strength of each abnormal node; Indicates the first The probability of identifying an abnormal node as an abnormal target; Indicates the first Evaluation values ​​of the spatial geometric relationship between the abnormal nodes and the nuclear power plant; Indicates the first The probability of an abnormal node and its associated nodes being abnormal; This indicates the preset magnification factor; Indicates the cooperating factor; This represents the normalization parameter. Where:

[0085] Potential anomaly strength of anomalous nodes Used to reflect the abnormal node itself The attack intensity or potential destructive capability can be obtained through specific analysis of the target trajectory fragments collected above. For example, the target's mass, such as the mass of a drone, the type of weapons it carries, its speed, and its energy, can be analyzed.

[0086] Indicates the first The probability of an anomalous node being identified as an anomalous target is calculated from the results of preceding multimodal fusion and classification. For example, if a drone is detected consistently by both vision and radar, then... Approaching 1; if only a single mode detects the target with low confidence, then It may be lower, for example, 0.3.

[0087] Indicates the first The evaluation value of the spatial geometric relationship between anomaly nodes and the nuclear power plant considers the distance, azimuth, and possible hit probability between the trajectory of the anomaly node and sensitive units in the plant area. This value can be calculated through real-time computer simulation. For example, if the drone's flight path is directly opposite the reactor building, then the corresponding... The value is relatively high; if the path runs parallel to the factory area and is not close to the sensitive area, the value is relatively low.

[0088] The second part of the calculation A synergistic effect is introduced: if multiple anomalies are related in time, space and behavior, they will amplify the overall probability of injury.

[0089] This represents the normalization parameter, which makes the final output probability within [0,1].

[0090] In an optional implementation, the recognition system can display the above information to the user through a graphical interface, as shown in the attached figure. Figure 5 As shown.

[0091] This technical solution integrates node independence probability, potential strength, identification reliability, spatial relationships, and synergistic effects to quantify hazard probability from multiple dimensions, avoiding bias from a single factor and accurately reflecting the actual threat of abnormal nodes. This provides scientific quantitative support for the graded prevention and control and efficient handling of nuclear power plants.

[0092] The technical solution of this invention collects various field detection signals from multiple types of sensors installed outside the nuclear power plant. Based on these signals, multiple target objects and corresponding trajectory segments are determined. This multi-source sensor fusion enables accurate identification and complete trajectory reconstruction of field targets, improving the comprehensiveness and continuity of the nuclear power plant's perimeter situational awareness. Furthermore, based on preset rules and the target objects corresponding to the trajectory segments, multiple abnormal nodes are identified. An abnormal correlation graph is then established, and the probability of each abnormal node posing a hazard to the nuclear power plant is determined based on the correlation relationships between these nodes. This process displays the probability of each abnormal node's hazard, achieving structured correlation and quantitative display of abnormal target behavior. This solves the problems of insufficient identification, weak correlation, and difficulty in quantifying hazard probabilities in the field of nuclear power plants under multi-source sensing. It integrates multi-source sensing to achieve accurate acquisition of target trajectories and abnormal correlation analysis, quantifying the hazard probability of each abnormal node, and significantly improving the identification, assessment, and visualized prevention and control capabilities of safety risks.

[0093] Example 2

[0094] Figure 2This is a flowchart of a method for identifying external anomalies in a nuclear power plant according to Embodiment 2 of the present invention. This embodiment is a further refinement of the above embodiments, focusing on determining multiple target objects and multiple target trajectory segments corresponding to the multiple target objects based on various field detection signals. Optionally, determining multiple target objects and multiple target trajectory segments corresponding to the multiple target objects based on various field detection signals includes: preprocessing the multiple field detection signals to obtain target detection signals, wherein the preprocessing includes at least one of timestamp alignment, noise reduction, coordinate mapping, and filtering, and the coordinate mapping is to map the sensor coordinate system to the world coordinate system; determining multiple target objects based on the target detection signals, and determining the positions and corresponding target confidence levels of the multiple target objects; determining multiple target trajectory segments corresponding to the multiple target objects based on the multiple target objects, the positions of the target objects, and the target confidence levels. For detailed implementation, please refer to the description of this embodiment. Technical features that are the same as or similar to those in the foregoing embodiments will not be repeated here.

[0095] like Figure 2 As shown, the method may specifically include:

[0096] S210. Various field detection signals are collected by various types of sensors installed outside the nuclear power plant. The various field detection signals are preprocessed to obtain the target detection signal. The preprocessing includes at least one of timestamp alignment, noise reduction, coordinate mapping and filtering. The coordinate mapping is to map the sensor coordinate system to the world coordinate system.

[0097] The preprocessing can be understood as a preliminary processing step on the raw field detection signal, aiming to eliminate noise, unify the format, and align spatiotemporal information to provide reliable input for subsequent target detection. The preprocessing includes, but is not limited to, at least one of timestamp alignment, denoising, coordinate mapping, and filtering. Timestamp alignment can be understood as the operation of recalibrating field detection signals collected by different sensors according to a unified time reference, eliminating the "time asynchrony" problem of multi-sensor data. Coordinate mapping can be understood as the operation of converting observation data in the local coordinate system of the sensors into a unified world coordinate system, solving the "spatial heterogeneity" problem of multi-sensor data and enabling the observations of different sensors to be correlated within the same spatial framework.

[0098] S220. Based on the target detection signal, determine multiple target objects, and determine the positions and corresponding target confidence levels of the multiple target objects.

[0099] Based on the above scheme, optionally, the target confidence level corresponding to multiple target objects is determined, including: determining multiple types of evaluation indicators corresponding to multiple target objects, and determining the target confidence level corresponding to the target object based on the weighted summation result of the multiple types of evaluation indicators.

[0100] The evaluation metrics can be understood as a set of quantitative indicators used to assess the authenticity or reliability of a target object from different dimensions. This overcomes the limitations of a single indicator (e.g., the confidence score output by the detection algorithm may ignore sensor status or environmental influences), and through the complementarity of multi-dimensional indicators, it provides a more comprehensive and objective assessment of target confidence. The evaluation metrics include, but are not limited to, at least two of the following: detection confidence, location uncertainty, sensor health score, and environmental factors. Detection confidence can be understood as a probability or confidence score representing the error-free output of the sensor. Location uncertainty can be understood as the error range or reliability of the target object's location estimation; a smaller value indicates greater location uncertainty. Sensor health score can be understood as a score indicating the normality of the sensor's own operating state, serving as a "reliability correction term for target confidence" and reflecting the "source quality" of the sensor data. Environmental factors can be understood as a score indicating the impact of the external environment on the accuracy of target detection; a larger value indicates a more favorable environment for detection, serving as an "environmental correction term for target confidence" and reflecting the degree of interference from the external environment on detection.

[0101] This technical solution uses weighted fusion of multiple types of indicators to comprehensively evaluate the authenticity of targets, overcomes the limitations of single indicators, improves the comprehensiveness and accuracy of confidence determination, effectively filters false detections and retains highly credible targets, and lays a solid data foundation for subsequent trajectory analysis and anomaly detection.

[0102] S230. Determine multiple target trajectory segments corresponding to the multiple target objects based on the multiple target objects, the positions of the target objects, and the target confidence level.

[0103] S240. Based on preset rules and multiple target objects corresponding to multiple target trajectory segments, determine multiple abnormal nodes, determine an abnormal association graph based on the multiple abnormal nodes, determine the probability of each abnormal node causing harm to the nuclear power plant based on the association relationship between the abnormal nodes in the abnormal association graph, and display the probability of each abnormal node causing harm.

[0104] An optional implementation involves unifying the timestamps of raw data from various field detection signals from different sensors (visible / infrared cameras, short / medium / long-range radars, acoustic arrays, geomagnetic / vibration sensors, and radio frequency detectors); performing quality checks on the synchronized observation data; mapping each sensor from its own coordinate system to a unified plant area coordinate system; denoising and preprocessing filtering of the multimodal data; feature extraction from the preprocessed data; target detection based on data features, and local tracking of suspicious targets; calculating the data confidence level for each modal data point; and structuring the data processed in the above steps into trajectory segments and outputting them.

[0105] Specifically, the field detection signals are timestamped. For example, a timestamping priority strategy is applied to all sensor data: PTP (Precision Time Protocol) is prioritized; NTP is used when PTP is unavailable. Optionally, for sensors without a network time source, a local hardware timestamp is used at the acquisition end, and an offset estimate is recorded. Optionally, the goal of this step is to achieve an end-to-end time synchronization error of less than 10ms. Optionally, the following output fields are ultimately generated: original timestamp, synchronized timestamp, and clock offset estimate.

[0106] The system performs quality verification on the field detection signal, including verifying data format, frame integrity, CRC check, and sampling continuity; it also identifies frame loss, delays, or packet out-of-order delivery and generates anomaly flags. Preferably, this includes implementing rapid remedial strategies for abnormal situations, such as retrying reading, requesting retransmission, or interpolating with the most recent frame.

[0107] For each sensor, the coordinate system is mapped from the sensor body coordinate system to a unified world coordinate system. Preferably, it can be mapped to a unified plant area coordinate system (e.g., ENU) and combined with a digital elevation model (DEM). The final output of the acquired data includes the observed global position and position covariance.

[0108] The multimodal field detection signals are subjected to denoising and preprocessing filtering. For example, the processing of data for each modality includes:

[0109] Image data: Dehazing (dark spectrum estimation), rain / snow removal filtering (temporal / spatial median filtering), noise suppression (bilateral filtering or BM3D optional), inter-frame motion blur detection. For infrared images, perform bad pixel interpolation and temperature drift correction;

[0110] Radar data includes MTI / MTD operation to suppress ground clutter; CFAR detection; range gate / velocity gate pre-screening; bandpass filtering using Doppler filters, etc.

[0111] Acoustic data: bandpass filter, short-time Fourier transform (STFT) for spectral domain characterization, beamforming to estimate direction of travel.

[0112] Geomagnetic / vibration data: bandpass and peak detection, short-time energy calculation, etc.

[0113] Radio frequency data: energy detection, instantaneous spectrum analysis, mutual information / correlation for fingerprint comparison, etc.

[0114] Feature extraction is performed on the preprocessed data, which involves calculating or learning numerical values ​​or vectors from the original time-domain / frequency-domain / image / point cloud / waveform signals. These extracted features should reflect information about the target's morphology, kinematics, spectrum, modulation, or semantic category. The feature terms for each modality include:

[0115] Visual features include: target bounding box size, shape descriptor (Hu moment, orientation), keyframe image fingerprint (depth feature vector, such as MobileNet embedding); and human / vehicle / drone category confidence. For infrared image signals, thermal contours and maximum temperature points are extracted simultaneously.

[0116] Radar characteristics: such as RCS estimation, radial velocity, echo intensity spectrum, and micro-Doppler characteristics.

[0117] Acoustic characteristics: such as spectral peaks, bandwidth, propeller fundamental frequency, MFCC segment.

[0118] RF characteristics: such as average power, instantaneous spectrum, modulation characteristics, signal fingerprint hash.

[0119] Preferably, all features should be normalized / standardized to facilitate subsequent standardized algorithm analysis.

[0120] Target detection is performed based on the extracted data features, and potential target objects are locally tracked. For example, target object identification includes using radar data, image data, acoustic data, etc., and employing recognition algorithms to identify and lock onto the target object; subsequently, continuous detection frames or detection echoes are stitched together within a short time window, and the position and velocity of the target object are estimated using simple Kalman filtering or interactive multi-model estimation. Preferably, this also includes calculating the position / velocity covariance matrix for each target object.

[0121] Modal confidence is calculated for each modal data point at each acquisition point. The modal confidence Cm∈[0,1] is used to quantify the observation quality and reliability of a single sensor's modality (e.g., visible light / infrared / radar / acoustic / RF), serving as a weighting factor for subsequent cross-modal correlation and fusion. For example, the inputs for calculating the modal confidence Cm include:

[0122] (1) The detection confidence level d represents the probability or confidence score of the sensor output being error-free. d is normalized to [0,1].

[0123] (2) Location uncertainty The calculation method is as follows:

[0124] ;

[0125] In the above formula, It is the position covariance matrix of the target within a trajectory segment. Equals the sum of these variances; sensitivity parameter Indicates the sensitivity to control attenuation, generally The appropriate value should be one that makes the final result... The range is between 0.7 and 0.9.

[0126] Finally, in the above calculation formula, when When the numerical value is small, the location of the target is relatively certain. The value is close to 1;

[0127] when When the value is large, indicating that the target's location is highly uncertain, The value is close to 0.

[0128] (3) Sensor health score h∈[0,1] (self-test and link quality): The sensor health score is an indicator used to measure the current working status and data reliability of a single sensor. This parameter comprehensively characterizes the sensor's self-test results, such as hardware temperature, power supply voltage, internal calibration status, etc., as well as communication link quality, such as packet loss rate, latency, signal strength, etc. When the sensor is operating stably and data transmission is smooth, the score is close to 1; once hardware failure, environmental interference, or link abnormality occurs, the score will decrease, thereby reducing the impact of the sensor on the final result when multiple sensors are fused.

[0129] (4) Environmental factors e∈[0,1], for example, by collecting the comprehensive normalized factor of the influence of visibility, rainfall, wind speed and other factors on the mode at the moment, the larger the value, the more favorable the environment is for the observation of the mode.

[0130] Then calculate the modal confidence level Cm:

[0131] ;

[0132] In the above formula, the weights , , , ≥0, and Weights can be obtained by training labeled data using least squares or logistic regression. They can also be specified based on the specific design of the recognition system. For example, the following values ​​can be used: =0.4, =0.4, =0.1, =0.1. (Confidence level d, uncertainty) Sensor health score h, environmental factor e)

[0133] E800: The above information is encapsulated to form a standardized target trajectory segment, Tracklet. The Tracklet represents the sequence of the position, velocity, and related signal characteristics of the same target over a short time window. It is used to quickly describe the target object across multiple edge computing nodes, reduce signal bandwidth, and serve as the basic descriptive unit for subsequent multimodal signal association of the same target.

[0134] Preferably, the representation duration of each target trajectory segment can be 0.5 to 2 seconds; the update frequency is 30 to 200 ms, which can be set according to the modality and linkage level.

[0135] For example, the target trajectory segment Tracklet can be represented using a structured parameter description method:

[0136] Tracklet = {

[0137] id, / / Track segment number

[0138] modality, e.g., EO, IR, radar, acoustics, RF

[0139] t_start, t_end, / / Start and end times

[0140] position[t], / / Coordinate sequence (x, y, z) or latitude and longitude

[0141] velocity[t], / / velocity sequence

[0142] cov_position[t], / / Position covariance, representing uncertainty

[0143] features, / / Modal feature vectors (normalized)

[0144] detection_confidence, / / Modal confidence

[0145] sensor_meta / / Sensor ID, attitude, calibration version

[0146] }

[0147] The technical solution of this invention preprocesses multi-source field detection signals by performing time stamp alignment, denoising, coordinate mapping, and filtering to eliminate heterogeneity and noise interference, unify the spatiotemporal reference, and obtain reliable target detection signals. Combined with position and multi-dimensional confidence assessment, it accurately identifies real target objects. Then, based on the target object and its spatiotemporal information, it generates continuous trajectory segments, which not only ensures the accuracy and continuity of target detection, but also filters false detections and improves data quality, providing a solid and reliable input foundation for subsequent anomaly identification and hazard assessment, and significantly enhancing the nuclear power plant's ability to perceive and assess external threats.

[0148] Example 3

[0149] Figure 3 This is a schematic diagram of the structure of an external anomaly identification device for nuclear power plants provided in Embodiment 3 of the present invention. Figure 3 As shown, the device includes: a trajectory segment determination module 310 and a hazard occurrence probability determination module 320. Among them,

[0150] The trajectory segment determination module 310 is used to collect various field detection signals through various types of sensors installed outside the nuclear power plant, and determine multiple target objects and multiple target trajectory segments corresponding to the multiple target objects based on the various field detection signals.

[0151] The hazard occurrence probability determination module 320 is used to determine multiple abnormal nodes according to preset rules and multiple target objects corresponding to multiple target trajectory segments, determine an abnormal association graph according to the multiple abnormal nodes, determine the hazard occurrence probability of each abnormal node to the nuclear power plant according to the association relationship between the abnormal nodes in the abnormal association graph, and display the hazard occurrence probability of each abnormal node.

[0152] The technical solution of this invention uses a trajectory segment determination module to collect various field detection signals from multiple types of sensors installed outside the nuclear power plant. Based on these signals, multiple target objects and corresponding trajectory segments are determined. This multi-source sensor fusion enables accurate identification and complete trajectory reconstruction of field targets, improving the comprehensiveness and continuity of the nuclear power plant's perimeter situational awareness. Furthermore, a hazard probability determination module identifies multiple abnormal nodes based on preset rules and the target objects corresponding to the trajectory segments. An abnormality correlation graph is then established based on these nodes. The probability of each abnormal node posing a hazard to the nuclear power plant is determined based on the correlation relationships between these nodes, and the probability of each node is displayed. This achieves structured correlation and quantified display of hazard probabilities for abnormal target behaviors, solving the problems of insufficient identification, weak correlation, and difficulty in quantifying hazard probabilities in the field of nuclear power plants under multi-source sensing. It integrates multi-source sensing to achieve accurate acquisition of target trajectories and abnormal correlation analysis in the field of nuclear power plants, quantifying the hazard probabilities of each abnormal node, and significantly improving the identification, assessment, and visualized prevention and control capabilities of safety risks.

[0153] Optionally, the trajectory segment determination module includes: a preprocessing submodule, a confidence level determination submodule, and a trajectory segment determination submodule. The preprocessing submodule is used to preprocess various field detection signals to obtain target detection signals. The preprocessing includes at least one of timestamp alignment, noise reduction, coordinate mapping, and filtering. The coordinate mapping maps the sensor coordinate system to a world coordinate system. The confidence level determination submodule is used to determine multiple target objects based on the target detection signals and to determine the positions and corresponding target confidence levels of the multiple target objects. The trajectory segment determination submodule is used to determine multiple target trajectory segments corresponding to the multiple target objects based on the multiple target objects, their positions, and their target confidence levels.

[0154] Optionally, the confidence determination submodule is specifically used to resample multiple target trajectory segments according to a preset time, determine the similarity between multiple target objects based on the resampling results, aggregate multiple target objects into multiple object clusters based on the similarity and a preset similarity threshold, perform detection based on the multiple object clusters, and determine multiple abnormal nodes based on the detection results.

[0155] Optionally, the hazard occurrence probability determination module includes: a synergy factor determination submodule and an anomaly association graph determination submodule. The synergy factor determination submodule is used to determine the synergy factor between every two of the anomaly nodes; the anomaly association graph determination submodule is used to construct an anomaly association graph using multiple anomaly nodes as nodes and the synergy factors between the anomaly nodes as edges.

[0156] Optionally, the hazard occurrence probability determination module includes: a synergy factor determination submodule and a hazard occurrence probability determination submodule. The synergy factor determination submodule is used to determine the independent anomaly probability of each anomalous node and the synergy factor between each anomalous node and its associated nodes in the anomaly association graph; the hazard occurrence probability determination submodule is used to determine the hazard occurrence probability of each anomalous node to the nuclear power plant based on the independent anomaly probability and the synergy factor.

[0157] Optionally, the hazard occurrence probability determination submodule determines the hazard occurrence probability of each of the abnormal nodes to the nuclear power plant based on the following formula:

[0158] ;

[0159] in, Indicates the first The independent anomaly probability of each anomalous node. Indicates the first The potential anomaly strength of each abnormal node; Indicates the first The probability of identifying an abnormal node as an abnormal target; Indicates the first Evaluation values ​​of the spatial geometric relationship between the abnormal nodes and the nuclear power plant; Indicates the first The probability of an abnormal node and its associated nodes being abnormal; This indicates the preset magnification factor; Indicates the cooperating factor; This represents the normalization parameter.

[0160] The nuclear power plant external anomaly identification device provided in the embodiments of the present invention can execute the nuclear power plant external anomaly identification method provided in any embodiment of the present invention, and has the corresponding functional modules and beneficial effects of the method.

[0161] Example 4

[0162] Figure 6 A schematic diagram of an electronic device 10, which can be used to implement embodiments of the present invention, is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital processors, cellular phones, smartphones, wearable devices (e.g., helmets, glasses, watches, etc.), and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely illustrative and are not intended to limit the implementation of the invention described and / or claimed herein.

[0163] like Figure 6 As shown, the electronic device 10 includes at least one processor 11 and a memory, such as a read-only memory (ROM) 12 or a random access memory (RAM) 13, communicatively connected to the at least one processor 11. The memory stores computer programs executable by the at least one processor. The processor 11 can perform various appropriate actions and processes based on the computer program stored in the ROM 12 or loaded from storage unit 18 into the RAM 13. The RAM 13 can also store various programs and data required for the operation of the electronic device 10. The processor 11, ROM 12, and RAM 13 are interconnected via a bus 14. An input / output (I / O) interface 15 is also connected to the bus 14.

[0164] Multiple components in electronic device 10 are connected to I / O interface 15, including: input unit 16, such as keyboard, mouse, etc.; output unit 17, such as various types of displays, speakers, etc.; storage unit 18, such as disk, optical disk, etc.; and communication unit 19, such as network card, modem, wireless transceiver, etc. Communication unit 19 allows electronic device 10 to exchange information / data with other devices through computer networks such as the Internet and / or various telecommunications networks.

[0165] Processor 11 can be a variety of general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of processor 11 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special-purpose artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, a digital signal processor (DSP), and any suitable processor, controller, microcontroller, etc. Processor 11 performs the various methods and processes described above, such as a method for identifying external anomalies in a nuclear power plant.

[0166] In particular, according to embodiments of the present invention, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments of the present invention include a computer program product comprising a computer program carried on a non-transitory computer-readable medium, the computer program containing program code for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via communication unit 19, or installed from storage unit 18, or installed from ROM 12. When the computer program is executed by processor 11, it performs the functions defined in the methods of the embodiments of the present invention.

[0167] In some embodiments, a method for identifying external anomalies in a nuclear power plant can be implemented as a computer program tangibly contained in a computer-readable storage medium, such as storage unit 18. In some embodiments, part or all of the computer program can be loaded and / or installed on electronic device 10 via ROM 12 and / or communication unit 19. When the computer program is loaded into RAM 13 and executed by processor 11, one or more steps of the method for identifying external anomalies in a nuclear power plant described above can be performed. Alternatively, in other embodiments, processor 11 can be configured to perform a method for identifying external anomalies in a nuclear power plant by any other suitable means (e.g., by means of firmware).

[0168] Various embodiments of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field-programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), application-specific standard products (ASSPs), systems-on-a-chip (SoCs), complex programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments may include implementations in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which may be a dedicated or general-purpose programmable processor, capable of receiving data and instructions from a storage system, at least one input device, and at least one output device, and transmitting data and instructions to the storage system, the at least one input device, and the at least one output device.

[0169] Computer programs used to implement the methods of the present invention may be written in any combination of one or more programming languages. These computer programs may be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, such that when executed by the processor, the computer programs cause the functions / operations specified in the flowcharts and / or block diagrams to be performed. The computer programs may be executed entirely on a machine, partially on a machine, or as a standalone software package, partially on a machine and partially on a remote machine, or entirely on a remote machine or server.

[0170] In the context of this invention, a computer-readable storage medium can be a tangible medium that may contain or store a computer program for use by or in conjunction with an instruction execution system, apparatus, or device. A computer-readable storage medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination thereof. Alternatively, a computer-readable storage medium may be a machine-readable signal medium. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fibers, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof.

[0171] To provide interaction with a user, the systems and techniques described herein can be implemented on an electronic device having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and pointing device (e.g., a mouse or trackball) through which the user provides input to the electronic device. Other types of devices can also be used to provide interaction with the user; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including sound input, voice input, or tactile input).

[0172] The systems and technologies described herein can be implemented in computing systems that include backend components (e.g., as data servers), or middleware components (e.g., application servers), or frontend components (e.g., user computers with graphical user interfaces or web browsers through which users can interact with implementations of the systems and technologies described herein), or any combination of such backend, middleware, or frontend components. The components of the system can be interconnected via digital data communication of any form or medium (e.g., communication networks). Examples of communication networks include local area networks (LANs), wide area networks (WANs), blockchain networks, and the Internet.

[0173] A computing system can include clients and servers. Clients and servers are generally located far apart and typically interact through communication networks. The client-server relationship is created by computer programs running on the respective computers and having a client-server relationship with each other. The server can be a cloud server, also known as a cloud computing server or cloud host, which is a hosting product within the cloud computing service system to address the shortcomings of traditional physical hosts and VPS services, such as high management difficulty and weak business scalability.

[0174] It should be understood that the various forms of processes shown above can be used, with steps reordered, added, or deleted. For example, the steps described in this invention can be executed in parallel, sequentially, or in different orders, as long as the desired result of the technical solution of this invention can be achieved, and this is not limited herein.

[0175] The specific embodiments described above do not constitute a limitation on the scope of protection of this invention. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this invention should be included within the scope of protection of this invention.

Claims

1. A method for identifying external anomalies in a nuclear power plant, characterized in that, include: Multiple types of field detection signals are collected by various sensors installed outside the nuclear power plant, and multiple target objects and multiple target trajectory segments corresponding to the multiple field detection signals are determined based on the multiple field detection signals; Multiple abnormal nodes are determined based on preset rules and multiple target objects corresponding to multiple target trajectory segments. An abnormal association graph is determined based on the multiple abnormal nodes. The probability of each abnormal node causing harm to the nuclear power plant is determined based on the association relationship between the abnormal nodes in the abnormal association graph, and the probability of each abnormal node causing harm is displayed.

2. The method according to claim 1, characterized in that, The step of determining multiple target objects and multiple target trajectory segments corresponding to the multiple target objects based on various field detection signals includes: The various field detection signals are preprocessed to obtain the target detection signal. The preprocessing includes at least one of timestamp alignment, noise reduction, coordinate mapping, and filtering. The coordinate mapping is to map the sensor coordinate system to the world coordinate system. Based on the target detection signal, multiple target objects are determined, and the positions and corresponding target confidence levels of the multiple target objects are determined. Based on the multiple target objects, the positions of the target objects, and the target confidence level, determine multiple target trajectory segments corresponding to the multiple target objects.

3. The method according to claim 2, characterized in that, Determining the target confidence levels corresponding to multiple target objects includes: Multiple types of evaluation indicators are determined for the target objects. The target confidence level for each target object is determined based on the weighted summation of the multiple types of evaluation indicators. The evaluation indicators include at least two of the following: detection confidence level, location uncertainty, sensor health score, and environmental factors.

4. The method according to claim 1, characterized in that, The step of determining multiple abnormal nodes based on preset rules and multiple target objects corresponding to multiple target trajectory segments includes: Multiple target trajectory segments are resampled according to a preset duration. The similarity between multiple target objects is determined based on the resampling results. Multiple target objects are aggregated into multiple object clusters based on the similarity and a preset similarity threshold. Detection is performed based on the multiple object clusters. Multiple abnormal nodes are determined based on the detection results.

5. The method according to claim 1, characterized in that, The step of determining the anomaly association graph based on multiple anomaly nodes includes: Determine the collaboration factor between every two of the aforementioned anomalous nodes; An anomaly association graph is constructed using multiple anomalous nodes as nodes and the collaboration factors between the anomalous nodes as edges.

6. The method according to claim 1, characterized in that, The step of determining the probability of each abnormal node posing a hazard to the nuclear power plant based on the correlation between abnormal nodes in the abnormal correlation graph includes: Determine the independent anomaly probability of each anomaly node, and determine the collaboration factor between each anomaly node and its associated nodes in the anomaly association graph; The probability of each anomalous node posing a hazard to the nuclear power plant is determined based on the independent anomaly probability and the synergistic factor.

7. The method according to claim 6, characterized in that, The probability of each anomalous node posing a hazard to the nuclear power plant is determined based on the following formula: ； in, Indicates the first The independent anomaly probability of each anomalous node. Indicates the first The potential anomaly strength of each abnormal node; Indicates the first The probability of identifying an abnormal node as an abnormal target; Indicates the first Evaluation values ​​of the spatial geometric relationship between the abnormal nodes and the nuclear power plant; Indicates the first The probability of an abnormal node and its associated nodes being abnormal; This indicates the preset magnification factor; Indicates the cooperating factor; This represents the normalization parameter.

8. A device for identifying external anomalies in a nuclear power plant, characterized in that, include: The trajectory segment determination module is used to collect various field detection signals from various types of sensors installed outside the nuclear power plant, and to determine multiple target objects and multiple target trajectory segments corresponding to the multiple target objects based on the various field detection signals. The hazard occurrence probability determination module is used to determine multiple abnormal nodes based on preset rules and multiple target objects corresponding to multiple target trajectory segments, determine an abnormal association graph based on the multiple abnormal nodes, determine the hazard occurrence probability of each abnormal node to the nuclear power plant based on the association relationship between the abnormal nodes in the abnormal association graph, and display the hazard occurrence probability of each abnormal node.

9. An electronic device, characterized in that, The electronic device includes: At least one processor; and A memory communicatively connected to the at least one processor; wherein, The memory stores a computer program that can be executed by the at least one processor, the computer program being executed by the at least one processor to enable the at least one processor to perform the nuclear power plant external anomaly identification method according to any one of claims 1-7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer instructions that are used to cause a processor to execute the nuclear power plant external anomaly identification method according to any one of claims 1-7.

Images