A marine disaster-causing organism monitoring system and method for a coastal nuclear power plant

Abstract

The application discloses a marine disaster-causing organism monitoring system and method for a coastal nuclear power plant, and particularly relates to the technical field of marine monitoring. The system comprises an optical imaging subunit, an acoustic detection subunit and an environmental parameter sensing subunit arranged at a water intake area of the coastal nuclear power plant, and is used to acquire multi-source heterogeneous data in the marine environment of the water intake area of the coastal nuclear power plant. The acquired multi-source heterogeneous data is subjected to fusion analysis and feature extraction, and species identification results, individual quantity and individual body length data are output. A biological disaster-causing quantity quantitative curve is constructed, and coupling analysis is performed on key operation parameters of a cold source system of the nuclear power plant. A comprehensive risk index caused by a current biological situation to the cold source system is calculated. According to the comprehensive risk index, graded early warning information is automatically generated. Based on the generated early warning information and the comprehensive risk index, automatic control instructions are provided for operation personnel of the nuclear power plant.

Description

Technical Field

[0001] This invention relates to the field of marine monitoring technology, and more specifically, to a marine disaster-causing organism monitoring system and method for use in coastal nuclear power plants. Background Technology

[0002] As an important clean energy base, the operational safety of coastal nuclear power plants is closely related to the marine environment. Nuclear power plants typically use seawater as a cooling medium, introducing seawater into the cooling system through intake structures. However, the marine environment contains various potentially hazardous organisms that pose a threat to the safe operation of nuclear power plants, among which the explosive aggregation of plankton such as jellyfish and krill is particularly prominent.

[0003] Currently, monitoring of organisms causing disasters in the cold sources of coastal nuclear power plants mainly relies on manual inspections, underwater video surveillance, or simple acoustic detection equipment. Manual inspections are inefficient, have limited coverage, and cannot be conducted in adverse weather conditions; traditional underwater video surveillance is easily affected by water turbidity and lighting conditions, and it is difficult to accurately count and measure the length of organisms when they are densely packed; while existing acoustic equipment can detect biomass, it is difficult to accurately identify species, especially jellyfish, krill, and other soft-bodied organisms. Summary of the Invention

[0004] In order to overcome the above-mentioned defects of the prior art, embodiments of the present invention provide a marine disaster-causing organism monitoring system and method for coastal nuclear power plants, so as to solve the problems mentioned in the background art.

[0005] To achieve the above objectives, the present invention provides the following technical solution: a marine disaster-causing organism monitoring system for coastal nuclear power plants, comprising a data acquisition module, an edge intelligent processing module, a risk quantification and early warning module, and an intelligent decision-making and control module;

[0006] The data acquisition module includes an optical imaging subunit, an acoustic detection subunit, and an environmental parameter sensing subunit, and is used to acquire multi-source heterogeneous data of the marine environment in the water intake area of ​​the coastal nuclear power plant.

[0007] The edge intelligent processing module is communicatively connected to the data acquisition module and is used to perform fusion analysis and feature extraction on multi-source heterogeneous data, and output species identification results, number of individuals and individual body length data.

[0008] The risk quantification and early warning module is connected to the edge intelligent processing module. Based on the data output by the edge intelligent processing module, it generates a quantitative curve of biological disaster amount and performs coupled analysis with the key operating parameters of the nuclear power plant cold source system to calculate the comprehensive risk index of the current biological situation on the cold source system. Based on the comprehensive risk index, it automatically generates graded early warning information.

[0009] The intelligent decision-making and control module is connected to the risk quantification and early warning module and the control interface of the nuclear power plant cold source system, respectively. Based on the early warning information and comprehensive risk index generated by the risk quantification and early warning module, it provides automated control commands to nuclear power plant operators.

[0010] Preferably, as a preferred embodiment of the marine disaster-causing biological monitoring system for a coastal nuclear power plant according to the present invention, the data acquisition module includes an optical imaging subunit, an acoustic detection subunit, and an environmental parameter sensing subunit, used to acquire multi-source heterogeneous data of the marine environment in the water intake area of ​​the coastal nuclear power plant;

[0011] The optical imaging subunit includes at least one high-definition underwater camera, equipped with an adaptive spectral supplementary light source and an ultrasonic self-cleaning device to ensure the stability of image quality under different lighting and water conditions. It is deployed in the underwater area of ​​the water intake and is used to continuously acquire high-definition video streams and still images containing disaster-causing biological targets.

[0012] The acoustic detection subunit includes multiple multi-frequency sonar devices deployed in the underwater area of ​​the water intake. These multi-frequency sonars are used to transmit and receive sound waves to detect organisms in the water, obtaining information on the location, density, and movement trajectories of the biological population. Specifically, this includes the following:

[0013] By emitting broadband acoustic pulses with different center frequencies and analyzing the echo signals received from the water body, the acoustic target intensity, volumetric scattering intensity, and radial velocity relative to the sonar of the organism are obtained. The acoustic target intensity is obtained by measuring the emitted sound source level. Received sound pressure level Dissemination loss The system gain G is calculated and reflects the ability of a biological system to scatter sound waves, with units of dB. The formula is: ,in, , R represents the acoustic target intensity, and R represents the target distance. The absorption attenuation coefficient (TS) shows that the TS values ​​of different organisms and their frequency dependence vary significantly at different frequencies.

[0014] The volumetric scattering intensity is obtained by summing the acoustic target intensities of all N detection targets within the sampling volume and taking the logarithm. It reflects the density of organisms per unit volume. The specific formula is as follows: ,in, It is the acoustic target intensity of the a-th detected target;

[0015] The radial motion speed is measured by measuring the frequency change of the echo signal relative to the transmitted signal. The calculation yields the following formula: Where c is the speed of sound in water. The center frequency of the emitted sound wave It is the radial velocity component of the target along the direction of sound wave propagation. Non-biological targets have extremely low radial velocities, while biological targets exhibit significant non-zero velocities. It can effectively distinguish between biological scatterers and non-biological background, thereby improving the reliability of target detection.

[0016] Using multi-sonar triangulation, the same target is detected at different spatial locations to obtain its three-dimensional coordinates. Time-series analysis of the sonar echo data is performed, and target association and tracking algorithms are used to match and link the same target detected at different times, constructing its continuous motion path over a period of time. Further steps include:

[0017] Deploy at least three multi-frequency sonar devices whose spatial locations are known and which are not collinear with each other. When multiple sonars detect a target, the time delay between the transmission and reception of the sound waves is measured. Calculate the target to the sonar distance , will sonar The coordinates are represented as The three-dimensional coordinates of the target are obtained by solving the system of equations using the least squares method. The system of equations consists of the measurement geometry of each sonar:

[0018] Where M is the number of sonars that detected the target, b is the sonar device index, and c is the speed of sound in water;

[0019] The known target set at time step k is obtained using sonar equipment. The set of new probe points with time step k+1 By processing the known target state using the Kalman filter algorithm, the state of the target at different time steps is obtained, and state estimation points are generated. For the same target, the state estimation points at consecutive time steps are connected to generate the real-time motion trajectory of the target.

[0020] The environmental parameter sensing subunit is deployed in the water body at the intake point and integrates a multi-parameter water quality sensor array, including a high-precision digital water temperature sensor, a conductivity sensor, a photoelectric scattering turbidity sensor, a three-dimensional ADCP flow velocity sensor, and a fluorescence chlorophyll a sensor, for real-time monitoring and transmission of marine environmental parameters, including water temperature, salinity, turbidity, flow velocity, and chlorophyll concentration data.

[0021] Preferably, as a preferred embodiment of the marine disaster-causing organism monitoring system for coastal nuclear power plants described in this invention, it includes an edge intelligent processing module and a data acquisition module that are communicatively connected for performing fusion analysis and feature extraction on multi-source heterogeneous data, and outputting species identification results, individual quantity, and individual body length data, specifically including the following:

[0022] It receives video streams and still images from the optical imaging subunit, sonar echo data from the acoustic detection subunit, and environmental parameter data from the environmental parameter sensing subunit. After spatiotemporal registration and correlation of the optical, acoustic, and environmental data, it extracts visual features, acoustic features, and environmental features, including:

[0023] A global coordinate system is established with the water intake center as the origin. For each optical image pixel, a ray is back-projected into the global coordinate system using the camera calibration matrix and pose matrix. For each acoustic detection point, its three-dimensional coordinates are... It exists directly in the global coordinate system; by calculating the perpendicular distance from the acoustic point to the optical ray and setting a threshold... Establish the correlation between "optical pixel area and acoustic detection point" ,in, Represents the first in an optical image One optical pixel, Represents the first in an optical image One acoustic detection point, Indicates the first The ray corresponding to the first optical pixel and the second The vertical distance between each acoustic detection point;

[0024] The acquired video streams and still images are preprocessed. A lightweight convolutional neural network model is pre-trained on a large marine life image dataset. For the target pest species, the pre-trained model is fine-tuned using a labeled target pest image dataset. The preprocessed images are then input into the fine-tuned CNN model, and the output of its last convolutional layer is extracted as a high-dimensional visual feature map. The feature map is then transformed into a one-dimensional feature vector using global average pooling. ,in, The eigenvector represents the first eigenvector. There are 1 element, where H and W are the height and width of the feature map, respectively, p represents the position index of the feature map in the height direction, and q represents the position index of the feature map in the width direction.

[0025] The acquired sonar echo data undergoes signal processing, including filtering and time-frequency analysis. For each detected and tracked acoustic target, a multidimensional feature vector is constructed. ,in, It is the acoustic target intensity. It is the multi-frequency TS difference. It is the radial velocity component. It is the volume scattering intensity;

[0026] The environmental parameter data provided by the environmental parameter sensing subunit is standardized and concatenated into an environmental state vector. ;

[0027] Employing a dual-stream neural network architecture, it integrates visual features from optical images with acoustic and environmental features to perform fine-grained classification tasks. This allows for high-accuracy differentiation of pest-causing organisms, outputting species identification results, individual quantity, and individual body length data. Further details include:

[0028] acoustic feature vectors and environment state vector The inputs are fed into their respective fully connected layers for dimension alignment and preliminary feature transformation, including acoustic feature transformation. With environmental feature transformation And by performing a concatenation operation, an early fusion feature vector is formed. The acoustic and environmental information is further extracted by passing it through multiple fully connected layers to obtain a fused feature representation of the acoustic-environment modalities. ,in, These are the learning weight matrices for acoustic and environmental features, respectively. These are the bias vectors for acoustic and environmental features, respectively, and ReLU is the activation function. , These are the weight matrix and bias vector of the fully connected layer, respectively.

[0029] Features that fuse visual features and acoustic-environment modalities Perform weighted fusion to obtain the final fused feature vector. And will finally fuse the feature vectors The input is fed into a classifier, which uses the Softmax activation function as the output layer to predict the probability distribution of the species of the causative organism. For the e-th organism category, the probability of the classifier's output layer is... for ,in, , These are the weights and biases of the classifier's output layer. This indicates that visual features are represented in the final fused feature vector. Weights in;

[0030] By minimizing the cross-entropy loss function, end-to-end training is performed on a large number of labeled multimodal datasets to optimize the parameters of the entire network. Where U is the number of samples and E is the number of categories. It is the true class label of sample u. It is the probability predicted by the model that sample u belongs to class e;

[0031] The category with the highest probability is selected as the final species identification result. A lightweight target detection model is used to detect and classify the bounding boxes of individual organisms causing disasters in optical images, and to track the targets in the detection results in consecutive video frames, outputting the number of individuals of different species detected in the field of view.

[0032] For successfully associated "optical pixel region - acoustic detection point" pairs, based on the acoustic detection point... The three-dimensional spatial position of the organism is determined, and the bounding box information obtained from target detection is used to convert the pixel size of the organism in the optical image into its actual physical length.

[0033] Preferably, as a preferred embodiment of the marine disaster-causing biological monitoring system for coastal nuclear power plants described in this invention, it includes a risk quantification and early warning module connected to an edge intelligent processing module. Based on the data output by the edge intelligent processing module, a quantitative curve of biological disaster-causing quantity is generated, and coupled with key operating parameters of the nuclear power plant's cold source system for analysis, the comprehensive risk index of the current biological situation on the cold source system is calculated. Based on the comprehensive risk index, graded early warning information is automatically generated, specifically including the following:

[0034] Based on species identification results, individual quantity, and individual body length data output by the edge intelligent processing module, a disaster quantification model for different disaster-causing organisms is constructed. For a single disaster-causing organism, a disaster intensity function is established based on its physical characteristics. Used to quantify the size of a single individual of this species at different body lengths The potential for causing disasters is below;

[0035] For a specific disaster-causing biological species j, the total disaster-causing amount at time point t Defined as the sum of the disaster-causing contributions of all individuals of that species: ,in, It represents the number of individuals of species j that caused the disaster, identified at time t. It is the body length of the i-th individual of this species at time t;

[0036] The total marine biological damage at the current moment is obtained by summing up all identified species that cause damage. Where J is the number of species of organisms causing the disaster. These are weighting coefficients, representing the potential hazard levels of different hazard-causing organisms to the cold source system, and... The changes over time were plotted to create a quantitative curve of biological hazard impact.

[0037] Obtain key operating parameters of the nuclear power plant's cooling system, including: filter pressure differential. Circulating water flow rate Condenser terminal temperature difference and circulating water pump power Based on the total biohazard potential and key operating parameters of the cold source system, a quantitative assessment of the comprehensive risk posed by the current biohazard situation to the cold source system is conducted, resulting in a comprehensive risk index. ,in, It is the first The standardized risk values ​​of key operating parameters at time t. It is the number of key operating parameters. It is the first Risk weights of key operating parameters It is the risk weight of total biological hazard, representing the direct risk that the total biological hazard itself poses to the system;

[0038] Based on the calculated comprehensive risk index The system categorizes early warning information into multiple levels and automatically generates corresponding early warning measures and reports, setting four incremental risk thresholds. To classify warning levels:

[0039] when The level 1 blue alert was triggered, indicating an increase in the amount of biologically hazardous substances.

[0040] when The level-two yellow alert was triggered, indicating that the amount of biological hazards continued to increase;

[0041] when The triggering of a Level III orange alert indicates that the biological threat level has reached a high level.

[0042] when The triggering of a Level IV red alert indicates that the amount of biological damage has reached a critical level.

[0043] Preferably, as a preferred embodiment of the marine disaster-causing organism monitoring system for coastal nuclear power plants according to the present invention, it includes the intelligent decision-making and control module connected to the risk quantification and early warning module and the control interface of the nuclear power plant's cold source system. Based on the early warning information and comprehensive risk index generated by the risk quantification and early warning module, it provides automated control commands to nuclear power plant operators, specifically including the following:

[0044] When a Level 1 blue alert is triggered, the monitoring frequency of the data acquisition module is automatically adjusted to continuously track and observe specific biological species and areas, and the warning cleaning mode of the bar screen cleaning equipment is activated, adjusting the cleaning frequency to once every 30 minutes for preventive cleaning to prevent biological aggregation.

[0045] When a Level II yellow alert is triggered, the operation mode of the bar screen cleaner should be adjusted to the enhanced cleaning mode, once every 15 minutes, and the standby circulating water pump should be checked to ensure it is ready to be started at any time.

[0046] When a Level 3 orange alert is triggered, all available bar screen cleaners and circulating water pumps will be started and operated at maximum frequency and intensity. Mandatory recommendations to reduce the operating load of the nuclear power unit will be issued to the operators, and a loud audible and visual alarm will be issued.

[0047] When a Level 4 red alert is triggered, all available decontamination and emergency cleanup measures shall be activated and carried out at maximum power and speed. Mandatory instructions shall be immediately issued to the operating personnel, and an emergency shutdown procedure shall be initiated to protect the safety of the nuclear reactor. The highest level of audible and visual alarms shall be issued, and all relevant personnel and external emergency agencies shall be notified through multiple channels.

[0048] This embodiment also provides a method for monitoring marine hazardous organisms in coastal nuclear power plants, specifically including:

[0049] By deploying optical imaging subunits, acoustic detection subunits, and environmental parameter sensing subunits in the water intake area, multi-source heterogeneous data of the marine environment in the water intake area of ​​the coastal nuclear power plant are acquired.

[0050] The obtained multi-source heterogeneous data are fused and analyzed, and features are extracted to output species identification results, number of individuals, and individual body length data.

[0051] Based on the output species identification results, individual quantity and individual body length data, a quantitative curve of biological hazard is constructed and coupled with the key operating parameters of the nuclear power plant cold source system for analysis. The comprehensive risk index of the current biological situation on the cold source system is calculated, and graded early warning information is automatically generated based on the comprehensive risk index.

[0052] Based on the generated early warning information and comprehensive risk index, automated control commands are provided to nuclear power plant operators.

[0053] On the other hand, the present invention provides a computer device, including a memory and a processor, wherein the memory stores a computer program, wherein when the computer program is executed by the processor, it implements a functional module of a marine disaster-causing organism monitoring system for a coastal nuclear power plant as described above.

[0054] On the other hand, the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein: when the computer program is executed by a processor, it implements a marine disaster-causing organism monitoring system for a coastal nuclear power plant as described above.

[0055] The technical effects and advantages provided by the present invention in the above technical solution are as follows:

[0056] By monitoring marine organisms causing disasters in the intake area of ​​nuclear power plants, quantifying their disaster risks, and deeply coupling them with the operating parameters of the nuclear power plant's cold source system, intelligent hierarchical early warning can be achieved, significantly improving the nuclear power plant's ability to respond to marine organism-related disasters and ensuring the safe and stable operation of the nuclear power plant. Attached Figure Description

[0057] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this invention. For those skilled in the art, other drawings can be obtained based on these drawings.

[0058] Figure 1 This is a flowchart of the method of the present invention. Detailed Implementation

[0059] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0060] In the description of this application, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of indicated technical features. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of the stated features. In the description of this application, "multiple" means two or more, unless otherwise explicitly specified.

[0061] In the description of this application, the term "for example" is used to mean "used as an example, illustration, or description." Any embodiment described as "for example" in this application is not necessarily to be construed as being more preferred or advantageous than other embodiments. The following description is provided to enable any person skilled in the art to make and use the invention. Details are set forth in the following description for purposes of explanation. It should be understood that those skilled in the art will recognize that the invention can be made without using these specific details. In other instances, well-known structures and processes will not be described in detail to avoid obscuring the description of the invention with unnecessary detail. Therefore, the invention is not intended to be limited to the embodiments shown, but is consistent with the broadest scope of the principles and features disclosed in this application.

[0062] Example 1

[0063] This embodiment provides, for example Figure 1 The system shown is a marine disaster-causing organism monitoring system for coastal nuclear power plants, which specifically includes a data acquisition module, an edge intelligent processing module, a risk quantification and early warning module, and an intelligent decision-making and control module.

[0064] The data acquisition module includes an optical imaging subunit, an acoustic detection subunit, and an environmental parameter sensing subunit, and is used to acquire multi-source heterogeneous data of the marine environment in the water intake area of ​​the coastal nuclear power plant.

[0065] The edge intelligent processing module is communicatively connected to the data acquisition module and is used to perform fusion analysis and feature extraction on multi-source heterogeneous data, and output species identification results, number of individuals and individual body length data.

[0066] The risk quantification and early warning module is connected to the edge intelligent processing module. Based on the data output by the edge intelligent processing module, it generates a quantitative curve of biological disaster amount and performs coupled analysis with the key operating parameters of the nuclear power plant cold source system to calculate the comprehensive risk index of the current biological situation on the cold source system. Based on the comprehensive risk index, it automatically generates graded early warning information.

[0067] The intelligent decision-making and control module is connected to the risk quantification and early warning module and the control interface of the nuclear power plant cold source system, respectively. Based on the early warning information and comprehensive risk index generated by the risk quantification and early warning module, it provides automated control commands to nuclear power plant operators.

[0068] In this embodiment, the data acquisition module needs to be specifically described. The data acquisition module includes an optical imaging subunit, an acoustic detection subunit, and an environmental parameter sensing subunit, which is used to acquire multi-source heterogeneous data of the marine environment in the water intake area of ​​the coastal nuclear power plant.

[0069] The optical imaging subunit includes at least one high-definition underwater camera, equipped with an adaptive spectral supplementary light source and an ultrasonic self-cleaning device to ensure the stability of image quality under different lighting and water conditions. It is deployed in the underwater area of ​​the water intake and is used to continuously acquire high-definition video streams and still images containing disaster-causing biological targets.

[0070] The acoustic detection subunit includes multiple multi-frequency sonar devices deployed in the underwater area of ​​the water intake. These multi-frequency sonars are used to transmit and receive sound waves to detect organisms in the water, obtaining information on the location, density, and movement trajectories of the biological population. Specifically, this includes the following:

[0071] By emitting broadband acoustic pulses with different center frequencies and analyzing the echo signals received from the water body, the acoustic target intensity, volumetric scattering intensity, and radial velocity relative to the sonar of the organism are obtained. The acoustic target intensity is obtained by measuring the emitted sound source level. Received sound pressure level Dissemination loss The system gain G is calculated and reflects the ability of a biological system to scatter sound waves, with units of dB. The formula is: ,in, , R represents the acoustic target intensity, and R represents the target distance. The absorption attenuation coefficient (TS) shows that the TS values ​​of different organisms and their frequency dependence vary significantly at different frequencies.

[0072] The volumetric scattering intensity is obtained by summing the acoustic target intensities of all N detection targets within the sampling volume and taking the logarithm. It reflects the density of organisms per unit volume. The specific formula is as follows: ,in, It is the acoustic target intensity of the a-th detected target;

[0073] The radial motion speed is measured by measuring the frequency change of the echo signal relative to the transmitted signal. The calculation yields the following formula: Where c is the speed of sound in water. The center frequency of the emitted sound wave It is the radial velocity component of the target along the direction of sound wave propagation. Non-biological targets have extremely low radial velocities, while biological targets exhibit significant non-zero velocities. It can effectively distinguish between biological scatterers and non-biological background, thereby improving the reliability of target detection.

[0074] Using multi-sonar triangulation, the same target is detected at different spatial locations to obtain its three-dimensional coordinates. Time-series analysis of the sonar echo data is performed, and target association and tracking algorithms are used to match and link the same target detected at different times, constructing its continuous motion path over a period of time. Further steps include:

[0075] Deploy at least three multi-frequency sonar devices whose spatial locations are known and which are not collinear with each other. When multiple sonars detect a target, the time delay between the transmission and reception of the sound waves is measured. Calculate the target to the sonar distance , will sonar The coordinates are represented as The three-dimensional coordinates of the target are obtained by solving the system of equations using the least squares method. The system of equations consists of the measurement geometry of each sonar:

[0076] Where M is the number of sonars that detected the target, b is the sonar device index, and c is the speed of sound in water;

[0077] The known target set at time step k is obtained using sonar equipment. The set of new probe points with time step k+1 By processing the known target state using the Kalman filter algorithm, the state of the target at different time steps is obtained, and state estimation points are generated. For the same target, the state estimation points at consecutive time steps are connected to generate the real-time motion trajectory of the target.

[0078] The environmental parameter sensing subunit is deployed in the water body at the intake point and integrates a multi-parameter water quality sensor array, including a high-precision digital water temperature sensor, a conductivity sensor, a photoelectric scattering turbidity sensor, a three-dimensional ADCP flow velocity sensor, and a fluorescence chlorophyll a sensor, for real-time monitoring and transmission of marine environmental parameters, including water temperature, salinity, turbidity, flow velocity, and chlorophyll concentration data.

[0079] In this embodiment, the edge intelligence processing is specifically described. The edge intelligence processing module is communicatively connected to the data acquisition module and is used to perform fusion analysis and feature extraction on multi-source heterogeneous data, outputting species identification results, individual quantity, and individual body length data. Specifically, it includes the following:

[0080] It receives video streams and still images from the optical imaging subunit, sonar echo data from the acoustic detection subunit, and environmental parameter data from the environmental parameter sensing subunit. After spatiotemporal registration and correlation of the optical, acoustic, and environmental data, it extracts visual features, acoustic features, and environmental features, including:

[0081] A global coordinate system is established with the water intake center as the origin. For each optical image pixel, a ray is back-projected into the global coordinate system using the camera calibration matrix and pose matrix. For each acoustic detection point, its three-dimensional coordinates are... It exists directly in the global coordinate system; by calculating the perpendicular distance from the acoustic point to the optical ray and setting a threshold... Establish the correlation between "optical pixel area and acoustic detection point" ,in, Represents the first in an optical image One optical pixel, Represents the first in an optical image One acoustic detection point, Indicates the first The ray corresponding to the first optical pixel and the second The vertical distance between each acoustic detection point;

[0082] The acquired video streams and still images were preprocessed, including denoising, image enhancement, size normalization, and color space conversion. A lightweight convolutional neural network model was pre-trained on a large marine life image dataset. For the target pest species, the pre-trained model was fine-tuned using a labeled target pest image dataset. The pre-processed images were then input into the fine-tuned CNN model, and the output of the last convolutional layer was extracted as a high-dimensional visual feature map. The feature map is then transformed into a one-dimensional feature vector using global average pooling. ,in, The eigenvector represents the first eigenvector. There are 1 element, where H and W are the height and width of the feature map, respectively, p represents the position index of the feature map in the height direction, and q represents the position index of the feature map in the width direction.

[0083] The acquired sonar echo data undergoes signal processing, including filtering and time-frequency analysis. For each detected and tracked acoustic target, a multidimensional feature vector is constructed. ,in, It is the acoustic target intensity. It is the multi-frequency TS difference, which reflects the target's frequency response characteristics. It is the radial velocity component. It is the volume scattering intensity;

[0084] The environmental parameter data provided by the environmental parameter sensing subunit is standardized and concatenated into an environmental state vector. ;

[0085] Employing a dual-stream neural network architecture, it integrates visual features from optical images with acoustic and environmental features to perform fine-grained classification tasks. This allows for high-accuracy differentiation of pest-causing organisms, outputting species identification results, individual quantity, and individual body length data. Further details include:

[0086] acoustic feature vectors and environment state vector The inputs are fed into their respective fully connected layers for dimension alignment and preliminary feature transformation, including acoustic feature transformation. With environmental feature transformation And by performing a concatenation operation, an early fusion feature vector is formed. The acoustic and environmental information is further extracted by passing it through multiple fully connected layers to obtain a fused feature representation of the acoustic-environment modalities. ,in, These are the learning weight matrices for acoustic and environmental features, respectively. These are the bias vectors for acoustic and environmental features, respectively, and ReLU is the activation function. , These are the weight matrix and bias vector of the fully connected layer, respectively.

[0087] Features that fuse visual features and acoustic-environment modalities Perform weighted fusion to obtain the final fused feature vector. And will finally fuse the feature vectors The input is fed into a classifier, which uses the Softmax activation function as the output layer to predict the probability distribution of the species of the causative organism. For the e-th organism category, the probability of the classifier's output layer is... for ,in, , These are the weights and biases of the classifier's output layer. This indicates that visual features are represented in the final fused feature vector. Weights in;

[0088] By minimizing the cross-entropy loss function, end-to-end training is performed on a large number of labeled multimodal datasets to optimize the parameters of the entire network. Where U is the number of samples and E is the number of categories. It is the true class label of sample u. It is the probability predicted by the model that sample u belongs to class e;

[0089] The category with the highest probability is selected as the final species identification result. A lightweight target detection model is used to detect and classify the bounding boxes of individual organisms causing disasters in optical images, and to track the targets in the detection results in consecutive video frames, outputting the number of individuals of different species detected in the field of view.

[0090] For successfully associated "optical pixel region - acoustic detection point" pairs, based on the acoustic detection point... The three-dimensional spatial position of the organism is determined, and the bounding box information obtained from target detection is used to convert the pixel size of the organism in the optical image into its actual physical length.

[0091] In this embodiment, the risk quantification and early warning module needs to be specifically described. This module is connected to the edge intelligent processing module. Based on the data output by the edge intelligent processing module, it generates a quantitative curve of biological hazard, and performs coupled analysis with key operating parameters of the nuclear power plant's cold source system to calculate the comprehensive risk index of the current biological situation on the cold source system. Based on the comprehensive risk index, it automatically generates graded early warning information, specifically including the following:

[0092] Based on species identification results, individual quantity, and individual body length data output by the edge intelligent processing module, a disaster quantification model for different disaster-causing organisms is constructed. For a single disaster-causing organism, a disaster intensity function is established based on its physical characteristics. Used to quantify the size of a single individual of this species at different body lengths The potential for causing disasters is below;

[0093] For a specific disaster-causing biological species j, the total disaster-causing amount at time point t Defined as the sum of the disaster-causing contributions of all individuals of that species: ,in, It represents the number of individuals of species j that caused the disaster, identified at time t. It is the body length of the i-th individual of this species at time t;

[0094] The total marine biological damage at the current moment is obtained by summing up all identified species that cause damage. Where J is the number of species of organisms causing the disaster. These are weighting coefficients, representing the potential hazard levels of different hazard-causing organisms to the cold source system, and... The changes over time were plotted to create a quantitative curve of biological hazard impact.

[0095] Obtain key operating parameters of the nuclear power plant's cooling system, including: filter pressure differential. Circulating water flow rate Condenser terminal temperature difference and circulating water pump power Based on the total biohazard potential and key operating parameters of the cold source system, a quantitative assessment of the comprehensive risk posed by the current biohazard situation to the cold source system is conducted, resulting in a comprehensive risk index. ,in, It is the first The standardized risk values ​​of key operating parameters at time t. It is the number of key operating parameters. It is the first Risk weights of key operating parameters It is the risk weight of total biological hazard, representing the direct risk that the total biological hazard itself poses to the system;

[0096] Based on the calculated comprehensive risk index The system categorizes early warning information into multiple levels and automatically generates corresponding early warning measures and reports, setting four incremental risk thresholds. To classify warning levels:

[0097] when The triggering of a Level 1 Blue Alert indicates an increase in biological hazards, but the impact on the operation of the cold source system is minimal, and the system is in a stable operating state. It is recommended to strengthen monitoring and analyze potential risks.

[0098] when The triggering of a Level II Yellow Alert indicates that the amount of biological disasters continues to increase and some operating parameters of the cold source system have deviated from the normal range. It is recommended to activate the emergency plan, including increasing the frequency of bar cleaning, reducing the load on some parts of the system, and starting the backup circulating water pump.

[0099] when The triggering of a Level III orange alert indicates that the biological disaster has reached a high level and the operating parameters of the cold source system are significantly abnormal, requiring immediate emergency cleanup operations.

[0100] when The system triggered a Level IV red alert, indicating that the biohazard level had reached a critical point, the cold source system was under serious threat, and there was a risk of equipment failure. The system was required to take the highest level of emergency response measures immediately, including emergency shutdown and comprehensive decontamination.

[0101] In this embodiment, the intelligent decision-making and control module needs to be specifically described. This module is connected to the risk quantification and early warning module and the control interface of the nuclear power plant's cold source system. Based on the early warning information and comprehensive risk index generated by the risk quantification and early warning module, it provides automated control commands to the nuclear power plant operators, specifically including the following:

[0102] When a Level 1 blue alert is triggered, the monitoring frequency of the data acquisition module is automatically adjusted to continuously track and observe specific biological species and areas, and the warning cleaning mode of the bar screen cleaning equipment is activated, adjusting the cleaning frequency to once every 30 minutes for preventive cleaning to prevent biological aggregation.

[0103] When a Level II yellow alert is triggered, the operation mode of the bar screen cleaner should be adjusted to the enhanced cleaning mode, once every 15 minutes, and the standby circulating water pump should be checked to ensure it is ready to be started at any time.

[0104] When a Level 3 orange alert is triggered, all available bar screen cleaners and circulating water pumps will be started and operated at maximum frequency and intensity. Mandatory recommendations to reduce the operating load of the nuclear power unit will be issued to the operators, and a loud audible and visual alarm will be issued.

[0105] When a Level 4 red alert is triggered, all available decontamination and emergency cleanup measures shall be activated and carried out at maximum power and speed. Mandatory instructions shall be immediately issued to the operating personnel, and an emergency shutdown procedure shall be initiated to protect the safety of the nuclear reactor. The highest level of audible and visual alarms shall be issued, and all relevant personnel and external emergency agencies shall be notified through multiple channels.

[0106] Example 2

[0107] This embodiment provides a method for monitoring marine hazardous organisms in coastal nuclear power plants, specifically including:

[0108] By deploying optical imaging subunits, acoustic detection subunits, and environmental parameter sensing subunits in the water intake area, multi-source heterogeneous data of the marine environment in the water intake area of ​​the coastal nuclear power plant are acquired.

[0109] The obtained multi-source heterogeneous data are fused and analyzed, and features are extracted to output species identification results, number of individuals, and individual body length data.

[0110] Based on the output species identification results, individual quantity and individual body length data, a quantitative curve of biological hazard is constructed and coupled with the key operating parameters of the nuclear power plant cold source system for analysis. The comprehensive risk index of the current biological situation on the cold source system is calculated, and graded early warning information is automatically generated based on the comprehensive risk index.

[0111] Based on the generated early warning information and comprehensive risk index, automated control commands are provided to nuclear power plant operators.

[0112] The computer device can be a terminal, comprising a processor, memory, communication interface, display screen, and input devices connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The communication interface is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, carrier networks, NFC (Near Field Communication), or other technologies. The display screen can be an LCD screen or an e-ink screen. The input devices can be a touch layer covering the display screen, buttons, a trackball, or a touchpad on the computer device's casing, or an external keyboard, touchpad, or mouse.

[0113] This embodiment also provides a storage medium storing a computer program. When executed by a processor, the program implements the functional modules of a marine disaster-causing organism monitoring system for a coastal nuclear power plant as proposed in the above embodiments. The storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as Static Random Access Memory (SRAM), Electrically Erasable Programmable Read-Only Memory (EEPROM), Erasable Programmable Read Only Memory (EPROM), Programmable Red-Only Memory (PROM), Read-Only Memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk.

[0114] The foregoing has only described certain exemplary embodiments of the present invention by way of illustration. Undoubtedly, those skilled in the art can modify the described embodiments in various ways without departing from the spirit and scope of the present invention. Therefore, the foregoing drawings and descriptions are illustrative in nature and should not be construed as limiting the scope of protection of the claims of the present invention.

Claims

1. A marine disaster-causing biological monitoring system for coastal nuclear power plants, characterized in that: It includes a data acquisition module, an edge intelligence processing module, a risk quantification and early warning module, and an intelligent decision-making and control module; The data acquisition module includes an optical imaging subunit, an acoustic detection subunit, and an environmental parameter sensing subunit, and is used to acquire multi-source heterogeneous data of the marine environment in the water intake area of ​​the coastal nuclear power plant. The edge intelligent processing module is communicatively connected to the data acquisition module and is used to perform fusion analysis and feature extraction on multi-source heterogeneous data, and output species identification results, number of individuals and individual body length data. The risk quantification and early warning module is connected to the edge intelligent processing module. Based on the data output by the edge intelligent processing module, it generates a quantitative curve of biological hazard, and performs coupled analysis with key operating parameters of the nuclear power plant's cold source system to calculate the comprehensive risk index of the current biological situation on the cold source system. Based on the comprehensive risk index, it automatically generates graded early warning information, specifically including the following: Based on species identification results, individual quantity, and individual body length data output by the edge intelligent processing module, a disaster quantification model for different disaster-causing organisms is constructed. For a single disaster-causing organism, a disaster intensity function is established based on its physical characteristics. Used to quantify the size of a single individual of this species at different body lengths The potential for causing disasters is below; For a specific disaster-causing biological species j, the total disaster-causing amount at time point t Defined as the sum of the disaster-causing contributions of all individuals of that species: ,in, It represents the number of individuals of species j that caused the disaster, identified at time t. It is the body length of the i-th individual of this species at time t; The total marine biological damage at the current moment is obtained by summing up all identified species that cause damage. Where J is the number of species of organisms causing the disaster. These are weighting coefficients, and... The changes over time were plotted to create a quantitative curve of biological hazard impact. Obtain key operating parameters of the nuclear power plant's cooling system, including: filter pressure differential. Circulating water flow rate Condenser terminal temperature difference and circulating water pump power Based on the total biohazard potential and key operating parameters of the cold source system, a quantitative assessment of the comprehensive risk posed by the current biohazard situation to the cold source system is conducted, resulting in a comprehensive risk index. ,in, It is the first The standardized risk values ​​of key operating parameters at time t. It is the number of key operating parameters. It is the first Risk weights of key operating parameters It is the risk weight of total biological hazard, representing the direct risk that the total biological hazard itself poses to the system; The intelligent decision-making and control module is connected to the risk quantification and early warning module and the control interface of the nuclear power plant cold source system, respectively. Based on the early warning information and comprehensive risk index generated by the risk quantification and early warning module, it provides automated control commands to nuclear power plant operators.

2. The marine disaster-causing biological monitoring system for coastal nuclear power plants according to claim 1, characterized in that: The optical imaging subunit is deployed in the underwater area of ​​the water intake and is used to continuously acquire high-definition video streams and still images containing hazard-causing biological targets. The acoustic detection subunit, deployed in the underwater area of ​​the water intake, utilizes multi-band sonar to transmit and receive sound waves to detect organisms in the water and obtain information on the location, density, and movement trajectory of the biological population. Specifically, it includes the following: By emitting broadband acoustic pulses with different center frequencies and analyzing the echo signals received from the water body, the acoustic target intensity, volumetric scattering intensity, and radial velocity relative to the sonar of the organism are obtained. The acoustic target intensity is obtained by measuring the emitted sound source level. Received sound pressure level Dissemination loss The system gain G is calculated using the following formula: ,in, , R represents the acoustic target intensity, and R represents the target distance. The absorption attenuation coefficient; The volumetric scattering intensity is obtained by summing the acoustic target intensities of all N detection targets within the sampling volume and taking the logarithm. It reflects the density of organisms per unit volume. The specific formula is as follows: ,in, It is the acoustic target intensity of the a-th detected target; The radial motion speed is measured by measuring the frequency change of the echo signal relative to the transmitted signal. The calculation yields the following formula: Where c is the speed of sound in water. The center frequency of the emitted sound wave It is the radial velocity component of the target along the direction of sound wave propagation; By using multi-sonar triangulation, the same target is detected at different spatial locations to obtain the target's three-dimensional coordinates. The sonar echo data is then analyzed over time. Through target association and tracking algorithms, the same target detected at different times is matched and linked to construct its continuous motion path over a period of time. The environmental parameter sensing subunit is deployed in the water body at the intake point and integrates a multi-parameter water quality sensor array, including a high-precision digital water temperature sensor, a conductivity sensor, a photoelectric scattering turbidity sensor, a three-dimensional ADCP flow velocity sensor, and a fluorescence chlorophyll a sensor, for real-time monitoring and transmission of marine environmental parameters, including water temperature, salinity, turbidity, flow velocity, and chlorophyll concentration data.

3. A marine disaster-causing biological monitoring system for a coastal nuclear power plant according to claim 2, characterized in that: The method involves using multi-sonar triangulation to detect the same target at different spatial locations, acquiring the target's three-dimensional coordinates, performing time-series analysis on the sonar echo data, and matching and linking the same target detected at different times using target association and tracking algorithms to construct its continuous motion path over a period of time. This further includes: Deploy at least three multi-frequency sonar devices whose spatial locations are known and which are not collinear with each other. When multiple sonars detect a target, the time delay between the transmission and reception of the sound waves is measured. Calculate the target to the sonar distance , will sonar The coordinates are represented as The three-dimensional coordinates of the target are obtained by solving the system of equations using the least squares method. The system of equations consists of the measurement geometry of each sonar: Where M is the number of sonars that detected the target, b is the sonar device index, and c is the speed of sound in water; The known target set at time step k is obtained using sonar equipment. The set of new probe points with time step k+1 By processing the known target state using the Kalman filter algorithm, the state of the target at different time steps is obtained, and state estimation points are generated. For the same target, the state estimation points at consecutive time steps are connected to generate the real-time motion trajectory of the target.

4. A marine disaster-causing biological monitoring system for a coastal nuclear power plant according to claim 1, characterized in that: The edge intelligent processing module is communicatively connected to the data acquisition module and is used to perform fusion analysis and feature extraction on multi-source heterogeneous data, outputting species identification results, individual quantity, and individual body length data, specifically including the following: It receives video streams and still images from the optical imaging subunit, sonar echo data from the acoustic detection subunit, and environmental parameter data from the environmental parameter sensing subunit. After spatiotemporal registration and correlation of the optical, acoustic, and environmental data, it extracts visual features, acoustic features, and environmental features, including: A global coordinate system is established with the water intake center as the origin. For each optical image pixel, a ray is back-projected into the global coordinate system using the camera calibration matrix and pose matrix. For each acoustic detection point, its three-dimensional coordinates are... It exists directly in the global coordinate system; by calculating the perpendicular distance from the acoustic point to the optical ray and setting a threshold... Establish the correlation between "optical pixel region and acoustic detection point" ,in, Represents the first in an optical image One optical pixel, Represents the first in an optical image One acoustic detection point, Indicates the first The ray corresponding to the first optical pixel and the second The vertical distance between each acoustic detection point; The acquired video streams and still images were preprocessed. A lightweight convolutional neural network model was pre-trained on a large marine life image dataset. For the target pest species, the pre-trained model was fine-tuned using a labeled target pest image dataset. The preprocessed images were then input into the fine-tuned CNN model, and the output of the last convolutional layer was extracted as a high-dimensional visual feature map. The feature map is then transformed into a one-dimensional feature vector using global average pooling. ,in, The eigenvector represents the first eigenvector. There are 1 element, where H and W are the height and width of the feature map, respectively, p represents the position index of the feature map in the height direction, and q represents the position index of the feature map in the width direction; The acquired sonar echo data undergoes signal processing, including filtering and time-frequency analysis. For each detected and tracked acoustic target, a multidimensional feature vector is constructed. ,in, It is the acoustic target intensity. It is the multi-frequency TS difference. It is the radial velocity component. It is the volume scattering intensity; The environmental parameter data provided by the environmental parameter sensing subunit is standardized and concatenated into an environmental state vector. ; Employing a dual-stream neural network architecture, it integrates visual features from optical images with acoustic and environmental features to perform fine-grained classification tasks, distinguishing the types of disaster-causing organisms with high accuracy, and outputting species identification results, individual quantity, and individual body length data.

5. A marine disaster-causing biological monitoring system for a coastal nuclear power plant according to claim 4, characterized in that: The system employs a dual-stream neural network architecture, fusing visual features from optical images with acoustic and environmental features to perform fine-grained classification tasks. This allows for high-accuracy differentiation of causative organisms, outputting species identification results, individual quantity, and individual body length data. Further components include: acoustic feature vectors and environment state vector The inputs are fed into their respective fully connected layers for dimension alignment and preliminary feature transformation, including acoustic feature transformation. With environmental feature transformation And by performing a concatenation operation, an early fusion feature vector is formed. By passing it through multiple fully connected layers, a fused feature representation of the acoustic-environment modalities is obtained. ,in, These are the learning weight matrices for acoustic and environmental features, respectively. These are the bias vectors for acoustic and environmental features, respectively, and ReLU is the activation function. , These are the weight matrix and bias vector of the fully connected layer, respectively. Features that fuse visual features and acoustic-environment modalities Perform weighted fusion to obtain the final fused feature vector. And will finally fuse the feature vectors The input is fed into a classifier, which uses the Softmax activation function as the output layer to predict the probability distribution of the species of the causative organism. For the e-th organism category, the probability of the classifier's output layer is... for ,in, , These are the weights and biases of the classifier's output layer. This indicates that visual features are represented in the final fused feature vector. Weights in; By minimizing the cross-entropy loss function, end-to-end training is performed on a large number of labeled multimodal datasets to optimize the parameters of the entire network. Where U is the number of samples and E is the number of categories. It is the true class label of sample u. It is the probability predicted by the model that sample u belongs to class e; The category with the highest probability is selected as the final species identification result. A lightweight target detection model is used to detect and classify the bounding boxes of individual organisms causing disasters in optical images, and to track the targets in the detection results in consecutive video frames, outputting the number of individuals of different species detected in the field of view. For successfully associated "optical pixel region - acoustic detection point" pairs, based on the acoustic detection point... The three-dimensional spatial position of the organism is determined, and the bounding box information obtained from target detection is used to convert the pixel size of the organism in the optical image into its actual physical length.

6. A marine disaster-causing biological monitoring system for a coastal nuclear power plant according to claim 1, characterized in that: The comprehensive risk index automatically generates tiered early warning information, which specifically includes the following: Based on the calculated comprehensive risk index The system categorizes early warning information into multiple levels and automatically generates corresponding early warning measures and reports, setting four incremental risk thresholds. To classify warning levels: when The level 1 blue alert was triggered, indicating an increase in the amount of biologically hazardous substances. when The level-two yellow alert was triggered, indicating that the amount of biological hazards continued to increase; when The triggering of a Level III orange alert indicates that the biological threat level has reached a high level. when The triggering of a Level IV red alert indicates that the amount of biological damage has reached a critical level.

7. A marine disaster-causing biological monitoring system for a coastal nuclear power plant according to claim 1, characterized in that: The intelligent decision-making and control module is connected to the risk quantification and early warning module and the control interface of the nuclear power plant cold source system, and specifically includes the following: When a Level 1 blue alert is triggered, the monitoring frequency of the data acquisition module is automatically adjusted to continuously track and observe specific biological species and areas, and the warning and cleaning mode of the bar screen cleaning equipment is activated. When a Level II yellow alert is triggered, the operation mode of the bar screen cleaning machine should be adjusted to the enhanced cleaning mode, and the standby circulating water pump should be checked to ensure it is ready to be started at any time. When a Level 3 orange alert is triggered, all bar screen cleaning machines and circulating water pumps will be started and run at maximum frequency and intensity. Mandatory recommendations to reduce the operating load of the nuclear power unit will be issued to the operators, and a loud audible and visual alarm will be issued. When a Level 4 red alert is triggered, all cleanup and emergency cleanup measures are initiated, and operations are carried out at maximum power and speed. Mandatory instructions are immediately issued to operating personnel, and an emergency shutdown procedure is initiated to protect the safety of the nuclear reactor. The highest level of audible and visual alarms are issued, and all relevant personnel and external emergency agencies are notified through multiple channels.

8. A method for monitoring marine hazardous organisms in a coastal nuclear power plant, applied to a marine hazardous organism monitoring system for a coastal nuclear power plant as described in any one of claims 1-7, characterized in that: Specifically, it includes: By deploying optical imaging subunits, acoustic detection subunits, and environmental parameter sensing subunits in the water intake area, multi-source heterogeneous data of the marine environment in the water intake area of ​​the coastal nuclear power plant are acquired. The obtained multi-source heterogeneous data are fused and analyzed, and features are extracted to output species identification results, number of individuals, and individual body length data. Based on the output species identification results, individual quantity and individual body length data, a quantitative curve of biological hazard is constructed and coupled with the key operating parameters of the nuclear power plant cold source system for analysis. The comprehensive risk index of the current biological situation on the cold source system is calculated, and graded early warning information is automatically generated based on the comprehensive risk index. Based on the generated early warning information and comprehensive risk index, automated control commands are provided to nuclear power plant operators.

Images