Power plant security and protection intelligent monitoring method and system

Abstract

The present application belongs to the technical field of power safety, and particularly relates to a power station security and protection intelligent monitoring method and system. The present application discloses a power station security and protection intelligent monitoring method, which comprises the following steps: configuring a video data collection device, transmitting video data to an intelligent analysis model of the video data, processing the video data through the intelligent analysis model to obtain a category label and a target frame of the video data, judging the category label and the target frame of the video data by using a judgment model of abnormal conditions and potential threats of the video data to obtain a judgment result of the video data, executing the judgment result and alarm and response rules of the video data by using an alarm and response model of abnormal conditions and potential threats of the video data to obtain alarm and response actions of the video data, and displaying the judgment result, alarm and response action information of the video data to managers and operators through a data visualization platform. Active protection and rapid reaction of the power station are realized.

Description

Technical Field

[0001] This invention belongs to the field of power safety technology, specifically relating to a method and system for intelligent monitoring of power plant security. Background Technology

[0002] Power plants are crucial national infrastructure, and their safe operation is vital to national energy security and social stability. With societal development and change, power plants face increasing security risks and threats, such as fires, explosions, intrusions, and terrorist attacks. These events, once they occur, can cause severe casualties and property damage, and even jeopardize national security and interests. Therefore, the security work of power plants is of paramount importance, requiring effective measures to enhance their security capabilities and emergency response capabilities. Summary of the Invention

[0003] In view of the above-mentioned problems, the present invention is proposed.

[0004] Therefore, the technical problems solved by this invention are: manual inspections are inefficient, have a small coverage area, are prone to omissions and errors, and cannot detect and handle abnormal situations in a timely manner; video surveillance equipment is scattered, the monitoring center is not unified, the quality of the monitoring images is poor, the storage time is short, and it is impossible to achieve full coverage and long-term retention, and it is impossible to perform intelligent analysis and processing of video data; there is a lack of advanced technical security equipment such as electronic inspection systems, drone active defense systems, intrusion and emergency alarm systems, which makes it impossible to achieve active protection and rapid response for power plants;

[0005] To address the aforementioned technical problems, this invention provides the following technical solution: Configuring video data acquisition equipment to transmit video data to an intelligent video data analysis model; preprocessing, feature extraction, temporal modeling, classification, and detection of the video data using the intelligent analysis model to obtain category labels and bounding boxes for the video data; using a judgment model for abnormal situations and potential threats in the video data to judge the category labels and bounding boxes for the video data, obtaining the judgment results; using an alarm and response model for abnormal situations and potential threats in the video data to execute the judgment results and alarm and response rules for the video data, obtaining alarm and response actions for the video data; and displaying the judgment results, alarms, and response action information of the video data to managers and operators through a data visualization platform.

[0006] As a preferred embodiment of the intelligent monitoring method for power plant security described in this invention, the video data acquisition equipment includes a full-color intelligent dome network camera, a global intelligent dome network camera, a full-color tube-type AI network camera, a full-color intelligent dome surveillance camera, an AI intelligent infrared bullet camera, a full-color tube-type intelligent network camera, and a dual-spectrum thermal imaging 8-inch dome camera.

[0007] As a preferred embodiment of the intelligent monitoring method for power plant security described in this invention, the intelligent analysis model for the video data includes the following components:

[0008] The video data is divided into several frames, and each frame is subjected to image enhancement, cropping, and scaling. A convolutional neural network (CNN) is used to extract features from each frame to obtain the feature vector of each frame. A recurrent neural network (RNN) is used to perform temporal modeling on the feature vector to obtain the temporal feature vector of the video data.

[0009] Fully connected layers and activation functions are used to classify and detect temporal feature vectors, resulting in category labels and bounding boxes for video data.

[0010] The video data is V = v1, v2, ..., v T , where v t Let t represent the t-th frame, and T represent the total number of frames in the video data; the convolutional neural network is a CNN, and its parameters are θ. CNN Its output is CNN(v t ;θ CNN ), representing the feature vector of the t-th frame;

[0011] Recurrent neural networks, or long short-term memory networks, are RNNs, and their parameters are θ. RNN Its output is RNN(CNN(v) t ;θ CNN );θ RNN ), representing the temporal feature vector of frame t;

[0012] The fully connected layer is FC, and its parameter is θ. FC Its output is FC(RNN(CNN(v) t ;θ CNN );θ RNN );θ FC ), representing the category label and target bounding box of frame t;

[0013] The intelligent analysis model for video data is as follows:

[0014]

[0015] Where θ=θ CNN ,θ RNN ,θ FC This represents the total parameters of the model;

[0016] The training objective of the model is to minimize the loss function for classifying and detecting video data. This loss function consists of cross-entropy loss and mean squared error loss, as follows:

[0017]

[0018] Where Y = y1, y2, ..., y T y represents the true label of the video data. t =y t,1 ,y t,2 ,...,y t,C ,y t,1 ,y t,2 ,...,y t,B y represents the true label of frame t. t,c Indicates whether frame t belongs to class c, y t,b p represents the coordinates of the b-th bounding box in frame t; t =p t,1 ,p t,2 ,...,p t,C ,p t,1 ,p t,2 ,...,p t,B p represents the predicted label for frame t. t,c p represents the probability that frame t belongs to class c. t,b Let C represent the coordinates of the b-th bounding box in frame t; let C represent the number of categories in the video data; let B represent the number of bounding boxes in the video data; and let λ represent the balance coefficient of the loss function.

[0019] The model parameters are updated using stochastic gradient descent (SGD) to minimize the loss function, as follows:

[0020]

[0021] Where α represents the learning rate. This represents the gradient of the loss function with respect to the model parameters;

[0022] Using the trained model parameters, intelligent analysis is performed on the input video data, outputting the video data's category labels and bounding boxes, as follows:

[0023]

[0024] in Represents the predicted labels for video data. This represents the predicted label for frame t. Let represent the probability that frame t belongs to class c. This represents the coordinates of the b-th bounding box in frame t; This represents the parameters of the trained model.

[0025] As a preferred embodiment of the intelligent power plant security monitoring method of the present invention, the judgment model for abnormal situations and potential threats of the video data includes judging whether there are abnormal situations and potential threats based on the category tags of the video data;

[0026] When the category label of the video data includes "flame", "smoke" and "explosion", it indicates that there is an abnormal situation of fire or explosion in the video. If the probability of the category label exceeds the set threshold, the abnormal situation is determined to be true.

[0027] When the category label of video data includes "drone", "airplane" or "helicopter", it indicates that there is a potential threat of airspace intrusion in the video. If the probability of the category label exceeds the set threshold, the potential threat is determined to be valid.

[0028] When the category labels of video data include "people", "vehicles" and "animals", it indicates that there is a potential threat of ground intrusion in the video. If the probability of the category label exceeds the set threshold, the potential threat is determined to be valid.

[0029] Based on the bounding boxes in the video data, determine whether there are any anomalies or potential threats;

[0030] Based on the category labels and bounding boxes of the video data, a comprehensive assessment is made to determine whether there are any abnormal situations or potential threats, as well as their severity and urgency.

[0031] If there are no abnormalities or potential threats in the category labels and bounding boxes of the video data, it means that the situation in the video is normal and no action needs to be taken.

[0032] If there are abnormalities or potential threats in the category labels or target boxes of video data, it indicates that the situation in the video is abnormal. It is necessary to judge its severity and urgency based on its type, quantity, scope, and impact, and select the appropriate level from the four levels of minor, general, serious, and critical.

[0033] If the category label of the video data is "flame" or "explosion", and the size, shape, and location characteristics of the target box indicate that the fire or explosion range is large and the impact is wide, it means that there is a serious or critical fire or explosion anomaly in the video, and fire extinguishing or evacuation emergency measures need to be taken immediately.

[0034] If the category label of the video data is "drone" or "airplane", and the number, proportion, and relationship characteristics of the target boxes indicate that there are many airspace intruders and they are close by, it means that there is a serious or critical potential threat of airspace intrusion in the video, and it is necessary to take immediate emergency measures such as drone interception or alarm.

[0035] If the category label of the video data is "personnel" or "vehicle", and the coordinates, size, shape, and location characteristics of the target box indicate that a ground intruder has entered a restricted or sensitive area, it means that there is a general or serious potential threat of ground intrusion in the video, and it is necessary to take emergency measures such as locking the access control or notifying the on-site personnel in a timely manner.

[0036] If the category label of the video data is "animals", and the coordinates, size, shape, position, and other features of the target box indicate that the animals have not caused damage or impact on the power station equipment, it means that there is a minor potential threat of ground intrusion in the video, and regular repelling or cleaning preventive measures are needed.

[0037] As a preferred embodiment of the intelligent power plant security monitoring method described in this invention, the judgment model for abnormal conditions and potential threats in the video data further includes the target bounding box of the video data. in This represents the target bounding box in frame t. This represents the coordinates of the b-th bounding box in frame t;

[0038] The results of judging anomalies and potential threats in the video data are S = s1, s2, ..., s T , where s t =s t,1 ,s t,2 ,...,s t,N s represents the judgment result of frame t. t,n This indicates the type, severity, and urgency of the nth anomaly or potential threat in frame t.

[0039] The set of anomalies and potential threats in video data is T = t1, t2, ..., t M , where t m This indicates the type of the m-th abnormal situation or potential threat;

[0040] The set of anomalies and the severity of potential threats in the video data is D = d1, d2, ..., d K , where d k This indicates the severity level, such as mild, moderate, severe, or critical.

[0041] The set of urgency levels of abnormal situations and potential threats is U = u1, u2, ..., u L , where u l Indicates the level of urgency (Level 1);

[0042] The model for judging anomalies and potential threats in video data is as follows:

[0043]

[0044] Where F represents the judgment function, whose input is the category label and target box of the video data, and whose output is the judgment result of the video data;

[0045] For each frame's category label Iterate through all categories tm ∈T, if Then it is assumed that there is an m-th anomaly or potential threat in the t-th frame, where α m The threshold representing the m-th abnormal situation or potential threat;

[0046] For the target bounding box in each frame Traverse all target boxes Based on the coordinates, size, shape, and location of the target bounding box, determine whether there are any abnormal situations or potential threats, including personnel, vehicles, equipment, and environment;

[0047] For the judgment result s of each frame t =s t,1 ,s t,2 ,...,s t,N Iterate through all anomalies or potential threats. t,n Based on its type, quantity, scope, and impact, determine its severity and urgency, and select the appropriate level from D and U, including minor, moderate, severe, and critical.

[0048] As a preferred embodiment of the intelligent monitoring method for power plant security described in this invention, the alarm and response model for abnormal situations and potential threats in the video data includes receiving the judgment results of the video data from the intelligent analysis model of the video data; and executing corresponding alarm and response actions based on the judgment results of the video data and the alarm and response rules, including triggering audible and visual alarms, intercepting drones, locking access control, notifying the management center and on-site personnel, and activating emergency plans.

[0049] As a preferred embodiment of the intelligent power plant security monitoring method described in this invention, the alarm and response model for abnormal situations and potential threats in the video data includes:

[0050] The alarm and response rules for video data are R = r1, r2, ..., r M , where r m =r m,1 ,r m,2 ,...,r m,K This represents the alarm and response rules for the m-th abnormal situation or potential threat, r. m,k This represents the k-th alarm and response action for the m-th abnormal situation or potential threat.

[0051] The alarm and response actions based on the video data are A = a1, a2, ..., a T , where a t =a t,1 ,a t,2 ,...,a t,L This represents the alarm and response actions in frame t, a t,lThis represents the l-th alarm and response action in the t-th frame. The alarm and response model for video data is as follows:

[0052] A = G(S,R)

[0053] Where G represents the alarm and response function, whose input is the judgment result of video data and the alarm and response rules, and whose output is the alarm and response actions of video data;

[0054] For the judgment result s of each frame t =s t,1 ,s t,2 ,...,s t,N Iterate through all anomalies or potential threats. t,n Based on its type, severity, and urgency, select the appropriate rule r from the alarm and response rules R. m ;

[0055] For each frame, the alarm and response actions are as follows: t =a t,1 ,a t,2 ,...,a t,L Iterate through all alarms and response rules. m Based on its alarm and response actions m,k , and perform the corresponding actions.

[0056] Another objective of this invention is to provide a system for intelligent monitoring of power plant security. This system, through modules such as video data acquisition equipment, intelligent video data analysis model, judgment model for abnormal situations and potential threats in video data, alarm and response model for abnormal situations and potential threats in video data, data visualization platform, and hydropower anti-terrorism emergency command system, realizes real-time monitoring, intelligent analysis, automatic alarm and emergency command of power plants, constructs a comprehensive prevention and control system, and ensures the safe operation of power plants.

[0057] To solve the above-mentioned technical problems, the present invention provides the following technical solution: a power plant security intelligent monitoring system, including video data acquisition equipment, video data intelligent analysis module, video data anomaly and potential threat judgment module, video data anomaly and potential threat alarm and response module, data visualization platform, and hydropower anti-terrorism emergency command module;

[0058] The video data acquisition equipment is used to collect video images and sensor data inside and outside the power station, so as to achieve all-round perception of the power station.

[0059] The intelligent analysis module for video data is used to process video data using deep learning, identify targets and events in the video, and realize intelligent analysis of the power station.

[0060] The video data anomaly and potential threat judgment module is used to make threshold judgments on the intelligent analysis results of video data, identify anomalies and potential threats in the video, and realize risk assessment of the power plant.

[0061] The alarm and response module for abnormal situations and potential threats in the video data is used to execute rules based on the risk assessment results of the video data, trigger corresponding alarms and response measures, and realize automatic protection of the power station.

[0062] The data visualization platform is used to display the analysis, judgment, alarm and response information of video data to managers and operators through a graphical interface, so as to realize the visual management of the power station.

[0063] The hydropower anti-terrorism emergency command system is used to conduct emergency management through an emergency command platform based on alarm and response information from video data, thereby realizing anti-terrorism emergency command for the power station.

[0064] A computer device includes a memory and a processor, wherein the memory stores a computer program, characterized in that the processor executes the computer program to implement the steps of a power plant security intelligent monitoring method.

[0065] A computer-readable storage medium having a computer program stored thereon, characterized in that, when the computer program is executed by a processor, it implements the steps of a power plant security intelligent monitoring method.

[0066] The beneficial effects of this invention are: it improves the security level of the power plant, effectively preventing and responding to security risks such as personnel access management, equipment failure early warning, fire alarm, water / airspace protection, and counter-terrorism emergency response; it reduces the labor costs of the power plant, reduces the workload of manual inspection and video monitoring, and improves work efficiency and accuracy; it enhances the management capabilities of the power plant, realizes data visualization, intelligence, linkage and emergency response, and improves the operation quality and service level of the power plant. Attached Figure Description

[0067] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the description of the embodiments will be briefly introduced below. Obviously, the 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. Wherein:

[0068] Figure 1 This is a flowchart of a power plant security intelligent monitoring method provided in one embodiment of the present invention.

[0069] Figure 2This is a schematic diagram of a monitoring device for a power plant security intelligent monitoring method according to an embodiment of the present invention.

[0070] Figure 3 This is a structural diagram of a power plant security intelligent monitoring system provided in one embodiment of the present invention. Detailed Implementation

[0071] To make the above-mentioned objects, features, and advantages of the present invention more apparent and understandable, specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. 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 protection scope of the present invention.

[0072] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0073] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in different places in this specification does not necessarily refer to the same embodiment, nor is it a single or selective embodiment that is mutually exclusive with other embodiments.

[0074] This invention is described in detail with reference to the schematic diagrams. When detailing the embodiments of this invention, for ease of explanation, the cross-sectional views illustrating the device structure may be partially enlarged, not adhering to the usual scale. Furthermore, the schematic diagrams are merely examples and should not be construed as limiting the scope of protection of this invention. In actual fabrication, the three-dimensional spatial dimensions of length, width, and depth should be included.

[0075] Furthermore, in the description of this invention, it should be noted that the terms "upper," "lower," "inner," and "outer," etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. These terms are used solely for the convenience of describing the invention and for simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on the invention. In addition, the terms "first," "second," or "third" are used for descriptive purposes only and should not be construed as indicating or implying relative importance.

[0076] Unless otherwise explicitly specified and limited, the terms "installation," "connection," and "joining" in this invention should be interpreted broadly. For example, they can refer to fixed connections, detachable connections, or integral connections; similarly, they can refer to mechanical connections, electrical connections, or direct connections, or indirect connections through an intermediate medium, or internal connections between two components. Those skilled in the art can understand the specific meaning of the above terms in this invention based on the specific circumstances.

[0077] Example 1

[0078] Reference Figure 1 and Figure 2 As an embodiment of the present invention, a power plant security intelligent monitoring method is provided, comprising:

[0079] S1: Configure the video data acquisition device and transmit the video data to the intelligent analysis model of the video data;

[0080] Furthermore, the video data acquisition devices include full-color intelligent dome network cameras, global intelligent dome network cameras, full-color bullet AI network cameras, full-color intelligent dome surveillance cameras, AI intelligent infrared bullet cameras, full-color bullet intelligent network cameras, and dual-spectrum thermal imaging 8-inch dome cameras.

[0081] It should be noted that the acquisition equipment is equipped with 24*16T enterprise-grade hard drives, enabling video recordings from the video surveillance system to be stored for 90 days or more.

[0082] S2: The intelligent analysis model is used to preprocess, extract features, model time series, classify and detect video data to obtain the category labels and target boxes of the video data;

[0083] Furthermore, the intelligent analysis model for the video data includes the following components:

[0084] The video data is divided into several frames, and each frame is subjected to image enhancement, cropping, and scaling. A convolutional neural network (CNN) is used to extract features from each frame to obtain the feature vector of each frame. A recurrent neural network (RNN) is used to perform temporal modeling on the feature vector to obtain the temporal feature vector of the video data.

[0085] Fully connected layers and activation functions are used to classify and detect temporal feature vectors, resulting in category labels and bounding boxes for video data.

[0086] The video data is V = v1, v2, ..., v T , where v t Let t represent the t-th frame, and T represent the total number of frames in the video data; the convolutional neural network is a CNN, and its parameters are θ. CNN Its output is CNN(v t ;θCNN ), representing the feature vector of the t-th frame;

[0087] Recurrent neural networks, or long short-term memory networks, are RNNs, and their parameters are θ. RNN Its output is RNN(CNN(v) t ;θ CNN );θ RNN ), representing the temporal feature vector of frame t;

[0088] The fully connected layer is FC, and its parameter is θ. FC Its output is FC(RNN(CNN(v) t ;θ CNN );θ RNN );θ FC ), representing the category label and target bounding box of frame t;

[0089] The intelligent analysis model for video data is as follows:

[0090]

[0091] Where θ=θ CNN ,θ RNN ,θ FC This represents the total parameters of the model;

[0092] The training objective of the model is to minimize the loss function for classifying and detecting video data. This loss function consists of cross-entropy loss and mean squared error loss, as follows:

[0093]

[0094] Where Y = y1, y2, ..., y T y represents the true label of the video data. t =y t,1 ,y t,2 ,...,y t,C ,y t,1 ,y t,2 ,...,y t,B y represents the true label of frame t. t,c Indicates whether frame t belongs to class c, y t,b p represents the coordinates of the b-th bounding box in frame t; t =p t,1 ,p t,2 ,...,p t,C ,p t,1 ,p t,2 ,...,p t,B p represents the predicted label for frame t. t,c p represents the probability that frame t belongs to class c. t,bLet C represent the coordinates of the b-th bounding box in frame t; let C represent the number of categories in the video data; let B represent the number of bounding boxes in the video data; and let λ represent the balance coefficient of the loss function.

[0095] The model parameters are updated using stochastic gradient descent (SGD) to minimize the loss function, as follows:

[0096]

[0097] Where α represents the learning rate. This represents the gradient of the loss function with respect to the model parameters;

[0098] Using the trained model parameters, intelligent analysis is performed on the input video data, outputting the video data's category labels and bounding boxes, as follows:

[0099]

[0100] in Represents the predicted labels for video data. This represents the predicted label for frame t. Let represent the probability that frame t belongs to class c. This represents the coordinates of the b-th bounding box in frame t; This represents the parameters of the trained model.

[0101] S3: Use the judgment model of anomalies and potential threats in video data to judge the category labels and target boxes of video data, and obtain the judgment results of video data;

[0102] Furthermore, the model for judging anomalies and potential threats in the video data includes determining whether anomalies and potential threats exist based on the category labels of the video data;

[0103] When the category label of the video data includes "flame", "smoke" and "explosion", it indicates that there is an abnormal situation of fire or explosion in the video. If the probability of the category label exceeds the set threshold, the abnormal situation is determined to be true.

[0104] When the category label of video data includes "drone", "airplane" or "helicopter", it indicates that there is a potential threat of airspace intrusion in the video. If the probability of the category label exceeds the set threshold, the potential threat is determined to be valid.

[0105] When the category labels of video data include "people", "vehicles" and "animals", it indicates that there is a potential threat of ground intrusion in the video. If the probability of the category label exceeds the set threshold, the potential threat is determined to be valid.

[0106] Based on the bounding boxes in the video data, determine whether there are any anomalies or potential threats;

[0107] Based on the category labels and bounding boxes of the video data, a comprehensive assessment is made to determine whether there are any abnormal situations or potential threats, as well as their severity and urgency.

[0108] If there are no abnormalities or potential threats in the category labels and bounding boxes of the video data, it means that the situation in the video is normal and no action needs to be taken.

[0109] If there are abnormalities or potential threats in the category labels or target boxes of video data, it indicates that the situation in the video is abnormal. It is necessary to judge its severity and urgency based on its type, quantity, scope, and impact, and select the appropriate level from the four levels of minor, general, serious, and critical.

[0110] If the category label of the video data is "flame" or "explosion", and the size, shape, and location characteristics of the target box indicate that the fire or explosion range is large and the impact is wide, it means that there is a serious or critical fire or explosion anomaly in the video, and fire extinguishing or evacuation emergency measures need to be taken immediately.

[0111] If the category label of the video data is "drone" or "airplane", and the number, proportion, and relationship characteristics of the target boxes indicate that there are many airspace intruders and they are close by, it means that there is a serious or critical potential threat of airspace intrusion in the video, and it is necessary to take immediate emergency measures such as drone interception or alarm.

[0112] If the category label of the video data is "personnel" or "vehicle", and the coordinates, size, shape, and location characteristics of the target box indicate that a ground intruder has entered a restricted or sensitive area, it means that there is a general or serious potential threat of ground intrusion in the video, and it is necessary to take emergency measures such as locking the access control or notifying the on-site personnel in a timely manner.

[0113] If the category label of the video data is "animals", and the coordinates, size, shape, position, and other features of the target box indicate that the animals have not caused damage or impact on the power station equipment, it means that there is a minor potential threat of ground intrusion in the video, and regular repelling or cleaning preventive measures are needed.

[0114] Furthermore, the model for judging anomalies and potential threats in the video data also includes the target bounding box of the video data. in This represents the target bounding box in frame t. This represents the coordinates of the b-th bounding box in frame t;

[0115] The results of judging anomalies and potential threats in the video data are S = s1, s2, ..., s T , where s t =st,1 ,s t,2 ,...,s t,N s represents the judgment result of frame t. t,n This indicates the type, severity, and urgency of the nth anomaly or potential threat in frame t.

[0116] The set of anomalies and potential threats in video data is T = t1, t2, ..., t M , where t m This indicates the type of the m-th abnormal situation or potential threat;

[0117] The set of anomalies and the severity of potential threats in the video data is D = d1, d2, ..., d K , where d k This indicates the severity level, such as mild, moderate, severe, or critical.

[0118] The set of urgency levels of abnormal situations and potential threats is U = u1, u2, ..., u L , where u l Indicates the level of urgency (Level 1);

[0119] The model for judging anomalies and potential threats in video data is as follows:

[0120]

[0121] Where F represents the judgment function, whose input is the category label and target box of the video data, and whose output is the judgment result of the video data;

[0122] For each frame's category label Iterate through all categories t m ∈T, if Then it is assumed that there is an m-th anomaly or potential threat in the t-th frame, where α m The threshold representing the m-th abnormal situation or potential threat;

[0123] For the target bounding box in each frame Traverse all target boxes Based on the coordinates, size, shape, and location of the target bounding box, determine whether there are any abnormal situations or potential threats, including personnel, vehicles, equipment, and environment;

[0124] For the judgment result s of each frame t =s t,1 ,s t,2 ,...,s t,N Iterate through all anomalies or potential threats. t,nBased on its type, quantity, scope, and impact, determine its severity and urgency, and select the appropriate level from D and U, including minor, moderate, severe, and critical.

[0125] S4: Utilize the alarm and response model for abnormal situations and potential threats in video data to execute the judgment results and alarm and response rules of video data, and obtain the alarm and response actions for video data;

[0126] Furthermore, the alarm and response model for abnormal situations and potential threats in the video data includes receiving the judgment results of the video data from the intelligent analysis model of the video data; and executing corresponding alarm and response actions based on the judgment results of the video data and alarm and response rules, including triggering audible and visual alarms, intercepting drones, locking access control, notifying the management center and on-site personnel, and activating emergency plans.

[0127] It should be noted that if the judgment result of the video data is "fire" or "explosion", and its severity is "serious" or "critical", then the first alarm and response rule will be executed, which includes the following actions: triggering an audible and visual alarm, issuing a warning signal to people in the surrounding area, and reminding them to stay away from the danger zone;

[0128] Notify the management center and on-site personnel, send them the judgment results, alarm and response information of the video data, and require them to take emergency measures such as fire extinguishing or evacuation immediately;

[0129] Activate the emergency response plan and, based on the power plant's fire or explosion emergency plan, automatically or manually control the power plant's equipment, systems, and access routes to ensure the power plant's safe and stable operation.

[0130] If the video data indicates "airspace intrusion" and its severity is "serious" or "critical," then the second alarm and response rule will be executed. This rule includes the following actions: triggering an audible and visual alarm to issue a warning signal to nearby personnel, alerting them to the abnormal situation in the air; notifying the management center and on-site personnel, sending them the video data's assessment result, alarm, and response information, and requesting them to closely monitor the movement and intent of the airspace intruder; and initiating drone interception, automatically or manually controlling the drone according to the power station's airspace defense system to identify, track, intercept, and drive away the airspace intruder.

[0131] Furthermore, the alarm and response model for anomalies and potential threats in the video data includes...

[0132] The alarm and response rules for video data are R = r1, r2, ..., r M , where r m =r m,1 ,r m,2 ,...,rm,K This represents the alarm and response rules for the m-th abnormal situation or potential threat, r. m,k This represents the k-th alarm and response action for the m-th abnormal situation or potential threat.

[0133] The alarm and response actions based on the video data are A = a1, a2, ..., a T , where a t =a t,1 ,a t,2 ,...,a t,L This represents the alarm and response actions in frame t, a t,l This represents the l-th alarm and response action in the t-th frame. The alarm and response model for video data is as follows:

[0134] A = G(S,R)

[0135] Where G represents the alarm and response function, whose input is the judgment result of video data and the alarm and response rules, and whose output is the alarm and response actions of video data;

[0136] For the judgment result s of each frame t =s t,1 ,s t,2 ,...,s t,N Iterate through all anomalies or potential threats. t,n Based on its type, severity, and urgency, select the appropriate rule r from the alarm and response rules R. m ;

[0137] For each frame, the alarm and response actions are as follows: t =a t,1 ,a t,2 ,...,a t,L Iterate through all alarms and response rules. m Based on its alarm and response actions m,k , and perform the corresponding actions.

[0138] S5: Display the judgment results, alarms, and response information of video data to managers and operators through a data visualization platform.

[0139] It should be noted that the data visualization platform includes a graphical interface, which is divided into multiple areas, including:

[0140] The real-time playback area for video data is used to display video images inside and outside the power plant, as well as category labels and target boxes for the video data;

[0141] The video data analysis results area displays the judgment results of the video data, as well as the type, severity, and urgency of anomalies and potential threats;

[0142] The video data alarm and response information area is used to display the alarm and response actions, as well as the status and results of the alarm and response.

[0143] The emergency command platform area is used to interact with the hydropower anti-terrorism emergency command system to realize anti-terrorism emergency command of the power station.

[0144] The data visualization platform also includes function buttons for controlling video data playback, pause, fast forward, rewind, screenshot, recording, and other operations, as well as adjusting parameters such as resolution, clarity, brightness, and contrast of the video data, and switching video data acquisition devices and angles.

[0145] The data visualization platform also includes prompts to provide advice and guidance to managers and operators:

[0146] If the video data indicates a "fire" or "explosion" and its severity is "serious" or "critical," a red warning message will be displayed in the alarm and response information area of ​​the video data, prompting "Please take immediate emergency measures such as fire extinguishing or evacuation."

[0147] If the judgment result of the video data is "airspace intrusion" and its severity is "serious" or "critical", then a yellow warning message will be displayed in the alarm and response information area of ​​the video data, prompting "Please take emergency measures such as drone interception or alarm immediately";

[0148] If the video data indicates "ground intrusion" and its severity is "moderate" or "severe", a blue warning message will be displayed in the alarm and response information area of ​​the video data, prompting "Please take emergency measures such as locking the access control or notifying on-site personnel in a timely manner".

[0149] If the video data indicates "ground intrusion" and its severity is "minor," a green message will be displayed in the alarm and response information area of ​​the video data, prompting "Please take preventive measures such as driving away or cleaning regularly."

[0150] Example 2

[0151] The second embodiment of the present invention differs from the first embodiment in that:

[0152] If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, essentially, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0153] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as a sequenced list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by, or in conjunction with, an instruction execution system, apparatus, or device (such as a computer-based system, a processor-included system, or other system that can fetch and execute instructions from, an instruction execution system, apparatus, or device). For the purposes of this specification, "computer-readable medium" can be any means that can contain, store, communicate, propagate, or transmit programs for use by, or in conjunction with, an instruction execution system, apparatus, or device.

[0154] More specific examples of computer-readable media (a non-exhaustive list) include: electrical connections (electronic devices) having one or more wires, portable computer disk drives (magnetic devices), random access memory (RAM), read-only memory (ROM), erasable and editable read-only memory (EPROM or flash memory), fiber optic devices, and portable optical disc read-only memory (CDROM). Furthermore, computer-readable media can even be paper or other suitable media on which the program can be printed, because the program can be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, interpreting, or otherwise processing as necessary, and then stored in computer memory.

[0155] It should be understood that various parts of the present invention can be implemented in hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented in software or firmware stored in memory and executed by a suitable instruction execution system. For example, if implemented in hardware, as in another embodiment, it can be implemented using any one or a combination of the following techniques known in the art: discrete logic circuits having logic gates for implementing logical functions on data signals, application-specific integrated circuits (ASICs) having suitable combinational logic gates, programmable gate arrays (PGAs), field-programmable gate arrays (FPGAs), etc.

[0156] Example 3

[0157] Reference Figure 3 This is the third embodiment of the present invention. This embodiment provides a power plant security intelligent monitoring system, including a video data acquisition device, a video data intelligent analysis module, a video data anomaly and potential threat judgment module, a video data anomaly and potential threat alarm and response module, a data visualization platform, and a hydropower anti-terrorism emergency command module.

[0158] The video data acquisition equipment is used to collect video images and sensor data inside and outside the power station, so as to achieve all-round perception of the power station.

[0159] The intelligent analysis module for video data is used to process video data using deep learning, identify targets and events in the video, and realize intelligent analysis of the power station.

[0160] The video data anomaly and potential threat judgment module is used to make threshold judgments on the intelligent analysis results of video data, identify anomalies and potential threats in the video, and realize risk assessment of the power plant.

[0161] The alarm and response module for abnormal situations and potential threats in the video data is used to execute rules based on the risk assessment results of the video data, trigger corresponding alarms and response measures, and realize automatic protection of the power station.

[0162] The data visualization platform is used to display the analysis, judgment, alarm and response information of video data to managers and operators through a graphical interface, so as to realize the visual management of the power station.

[0163] The hydropower anti-terrorism emergency command system is used to conduct emergency management through an emergency command platform based on alarm and response information from video data, thereby realizing anti-terrorism emergency command for the power station.

[0164] Example 4

[0165] The fourth embodiment of the present invention provides a method for intelligent monitoring of power plant security.

[0166] This example compares the effectiveness of our proposed method and existing technical solutions in power plant security, including indicators such as power generation revenue, operation and maintenance costs, production safety, and management level. The theoretical experiment involved two power plants of the same size and under the same conditions, using our proposed method and existing technical solutions respectively for security management for one year, recording and analyzing the data.

[0167] The data from the theoretical experiments are presented in tabular form as follows:

[0168]

[0169]

[0170] Electricity revenue refers to the total electricity generated by a power plant in a year multiplied by the electricity price, reflecting the economic benefits of the power plant. This method increases electricity revenue by 11.1% compared to existing technical solutions, indicating that this method can effectively improve the power plant's power generation efficiency, reduce power plant losses and failures, and increase the power plant's profits.

[0171] Operation and maintenance costs refer to the human, material, and financial resources invested by a power plant within one year, reflecting the power plant's operational efficiency. This method reduces operation and maintenance costs by 33.3% compared to existing technical solutions, indicating that this method can effectively reduce the power plant's manpower requirements, improve the power plant's automation level, and save the power plant's resources and funds.

[0172] Unplanned outages refer to the number of times a power plant abnormally stops generating electricity due to faults, accidents, attacks, or other reasons within a year, reflecting the power plant's safety performance. This method reduces the number of unplanned outages by 50% compared to existing technical solutions, demonstrating its effectiveness in preventing and responding to abnormal situations and potential threats to the power plant, thus ensuring its safe operation.

[0173] Inspection efficiency refers to the number of inspections a power plant completes in a year divided by the time spent on each inspection, reflecting the power plant's management efficiency. This method improves inspection efficiency by 100% compared to existing technical solutions, demonstrating that it effectively utilizes intelligent analysis of video data and equipment such as drones to improve the quality and speed of power plant inspections and optimize power plant management processes.

[0174] The security level of a power plant refers to the comprehensive score obtained by the power plant within one year according to security assessment standards, reflecting the power plant's prevention and control capabilities. This method improves the power plant's security level by 12.5% ​​compared to existing technical solutions, indicating that this method can effectively build a comprehensive prevention and control system, enhance the power plant's prevention, alarm, response, and command capabilities, and strengthen the power plant's security.

[0175] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A method for intelligent security monitoring of a power station, characterized in that, include: Configure video data acquisition equipment to transmit video data to an intelligent video data analysis model; The intelligent analysis model is used to preprocess, extract features, perform temporal modeling, classify and detect video data to obtain category labels and target boxes for the video data. The anomaly and potential threat judgment model of video data is used to judge the category label and target box of the video data to obtain the judgment result of the video data; The alarm and response model for anomalies and potential threats in video data is used to execute the judgment results and alarm and response rules of video data, thereby obtaining alarm and response actions for video data; The judgment results, alarms, and response information of video data are displayed to managers and operators through a data visualization platform; The intelligent analysis model for the video data includes dividing the video data into several frames, performing image enhancement, cropping, and scaling on each frame; extracting features from each frame using a convolutional neural network to obtain a feature vector for each frame; and performing temporal modeling on the feature vector using a recurrent neural network to obtain a temporal feature vector for the video data. Fully connected layers and activation functions are used to classify and detect temporal feature vectors, resulting in category labels and bounding boxes for video data. Video data is ,in Indicates the first frame, This represents the total number of frames in the video data; a convolutional neural network is... Its parameters are Its output is , indicating the first The feature vector of the frame; Recurrent neural networks are Its parameters are Its output is , indicating the first Temporal feature vector of a frame; Fully connected layer is Its parameters are Its output is , indicating the first The frame's category label and target bounding box; The intelligent analysis model for video data is as follows: ； in This represents the total parameters of the model; The training objective of the model is to minimize the loss function for classifying and detecting video data. This loss function consists of cross-entropy loss and mean squared error loss, as follows: ； in Represents the true label of video data. Indicates the first The true label of the frame, Indicates the first Does the frame belong to the first? kind, Indicates the first The first frame The coordinates of the target bounding box; Indicates the first Predicted label of frame, Indicates the first Frame belongs to the The probability of a class Indicates the first The first frame The coordinates of the target bounding box; Indicates the number of categories in the video data. This indicates the number of target bounding boxes in the video data. Represents the balance coefficient of the loss function; The model parameters are updated using stochastic gradient descent (SGD) to minimize the loss function, as follows: ； in Indicates the learning rate. This represents the gradient of the loss function with respect to the model parameters; Using the trained model parameters, intelligent analysis is performed on the input video data, outputting the video data's category labels and bounding boxes, as follows: ； in Represents the predicted labels for video data. Indicates the first Predicted label of frame, Indicates the first Frame belongs to the The probability of a class Indicates the first The first frame The coordinates of the target bounding box; This represents the parameters of the trained model; The model for judging anomalies and potential threats in video data includes determining whether anomalies and potential threats exist based on the category labels of the video data. When the category label of the video data includes "flame", "smoke" and "explosion", it indicates that there is an abnormal situation of fire or explosion in the video. If the probability of the category label exceeds the set threshold, the abnormal situation is determined to be true. When the category label of video data includes "drone", "airplane" and "helicopter", it indicates that there is a potential threat of airspace intrusion in the video. If the probability of the category label exceeds the set threshold, the potential threat is determined to be valid. When the category labels of video data include "people", "vehicles" and "animals", it indicates that there is a potential threat of ground intrusion in the video. If the probability of the category label exceeds a set threshold, the potential threat is determined to be valid. Based on the bounding box in the video data, determine whether there are any anomalies or potential threats; Based on the category labels and bounding boxes of the video data, a comprehensive assessment is made to determine whether there are any abnormal situations or potential threats, as well as their severity and urgency. If there are no abnormalities or potential threats in the category labels and bounding boxes of the video data, it means that the situation in the video is normal and no action needs to be taken. If there are abnormalities or potential threats in the category labels or target boxes of video data, it indicates that the situation in the video is abnormal. It is necessary to judge its severity and urgency based on its type, quantity, scope, and impact, and select the appropriate level from the four levels of minor, general, serious, and critical. If the category label of the video data is "flame" or "explosion", and the size, shape, and location characteristics of the target box indicate that the fire or explosion range is large and the impact is wide, it means that there is a serious or critical fire or explosion anomaly in the video, and fire extinguishing or evacuation emergency measures need to be taken immediately. If the category label of the video data is "drone" or "airplane", and the number, proportion, and relationship characteristics of the target boxes indicate that there are many airspace intruders and they are close at hand, it means that there is a serious or critical potential threat of airspace intrusion in the video, and it is necessary to take immediate emergency measures such as drone interception or alarm. If the category label of the video data is "personnel" or "vehicle", and the coordinates, size, shape, and location characteristics of the target box indicate that a ground intruder has entered a restricted or sensitive area, it means that there is a general or serious potential threat of ground intrusion in the video, and it is necessary to take emergency measures such as locking the access control or notifying the on-site personnel in a timely manner. If the category label of the video data is "animals", and the coordinates, size, shape, and location characteristics of the target box indicate that the animals have not caused damage or impact on the power station equipment, it means that there is a minor potential threat of ground intrusion in the video, and regular driving or cleaning preventive measures are needed. The alarm and response model for abnormal situations and potential threats in the video data includes receiving the judgment results of the video data from the intelligent analysis model of the video data; and executing corresponding alarm and response actions based on the judgment results of the video data and alarm and response rules, including triggering audible and visual alarms, intercepting drones, locking access control, notifying the management center and on-site personnel, and activating emergency plans. The alarm and response models for anomalies and potential threats in the video data include: The alarm and response rules for video data are as follows ,in Indicates the first Alarm and response rules for various abnormal situations or potential threats. Indicates the first The first abnormal situation or potential threat One alarm and response action; Alarm and response actions based on video data ,in Indicates the first Frame alarm and response actions, Indicates the first The first frame The alarm and response model for video data is as follows: ； in This represents the alarm and response function, whose input is the judgment result of video data and the alarm and response rules, and whose output is the alarm and response actions of video data; The judgment result for each frame Traverse all anomalies or potential threats Based on its type, severity, and urgency, from the alarm and response rules Select the corresponding rule ; Alarm and response actions for each frame Iterate through all alarms and response rules. Based on its alarm and response actions , and perform the corresponding actions.

2. The intelligent monitoring method for power plant security as described in claim 1, characterized in that: The video data acquisition devices include full-color intelligent dome network cameras, global intelligent dome network cameras, full-color tube AI network cameras, full-color intelligent dome surveillance cameras, AI intelligent infrared bullet cameras, full-color tube intelligent network cameras, and dual-spectrum thermal imaging 8-inch dome cameras.

3. The intelligent monitoring method for power plant security as described in claim 2, characterized in that: The model for judging anomalies and potential threats in video data also includes the target bounding box of the video data. ,in Indicates the first The target bounding box of the frame, Indicates the first The first frame The coordinates of the target bounding box; The assessment results regarding anomalies and potential threats in the video data are as follows: ,in Indicates the first The result of the frame judgment, Indicates the first The first frame The type, severity, and urgency of each unusual situation or potential threat; The set of anomalies and potential threats in video data is as follows: ,in Indicates the first Types of abnormal situations or potential threats; The set of anomalies and the severity of potential threats in video data is as follows: ,in Indicates the first The severity level is categorized as mild, moderate, severe, and critical. The set of urgency levels for abnormal situations and potential threats is as follows: ,in Indicates the first Level of urgency; The model for judging anomalies and potential threats in video data is as follows: ； in This represents a judgment function, whose input is the category label and bounding box of the video data, and whose output is the judgment result of the video data; For each frame's category label Iterate through all categories ,like Then it is considered that the first Frame exists in the first Such abnormal situations or potential threats, among which Indicates the first Thresholds for abnormal situations or potential threats; For the target bounding box of each frame Traverse all target boxes Based on the coordinates, size, shape, and location of the target bounding box, determine whether there are any abnormal situations or potential threats, including personnel, vehicles, equipment, and environment; The judgment result for each frame Traverse all anomalies or potential threats Based on its type, quantity, scope, and impact, its severity and urgency are determined, and then... and Choose the appropriate level, including minor, moderate, severe, or critical.

4. A system employing the intelligent power station security monitoring method as described in any one of claims 1 to 3, characterized in that: It includes video data acquisition equipment, intelligent video data analysis module, video data anomaly and potential threat judgment module, video data anomaly and potential threat alarm and response module, data visualization platform, and hydropower anti-terrorism emergency command module; The video data acquisition equipment is used to collect video images and sensor data inside and outside the power station, so as to achieve all-round perception of the power station. The intelligent analysis module for video data is used to process video data using deep learning, identify targets and events in the video, and realize intelligent analysis of the power station. The video data anomaly and potential threat judgment module is used to make threshold judgments on the intelligent analysis results of video data, identify anomalies and potential threats in the video, and realize risk assessment of the power plant. The alarm and response module for abnormal situations and potential threats in the video data is used to execute rules based on the risk assessment results of the video data, trigger corresponding alarms and response measures, and realize automatic protection of the power station. The data visualization platform is used to display the analysis, judgment, alarm and response information of video data to managers and operators through a graphical interface, so as to realize the visual management of the power station. The hydropower anti-terrorism emergency command system is used to conduct emergency management through an emergency command platform based on alarm and response information from video data, thereby realizing anti-terrorism emergency command of the power station.

5. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the method according to any one of claims 1 to 3.

6. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 3.

Images