Linkage early warning method based on security situation assessment and application thereof

Abstract

The invention provides a linkage early warning method based on security situation assessment and application thereof. The linkage early warning method comprises the following steps: data preprocessing and feature extraction; an abnormal evaluation model and a normal evaluation model are constructed, and training and parameter adjustment are carried out; evaluating the data acquired in real time by using a safety evaluation function, judging the current safety state of the hydraulic power plant, and outputting a safety parameter; constructing a decision function, and determining whether the current state is abnormal or not based on the safety parameters and environmental factors EnvFactors; dynamically selecting and determining to select an abnormal evaluation model and a normal evaluation model for real-time evaluation, and judging whether the current state needs early warning or not; when early warning is needed, an emergency command system and early warning information are linked to carry out early warning and emergency response, and possible safety risks of the hydraulic power plant are prompted. According to the invention, by constructing the security evaluation function and the decision function and combining the neural network model, the problem that the existing system lacks enough flexibility and adaptability and cannot effectively cope with variable network security threats is solved.

Description

Technical Field

[0001] The present invention relates to the field of security early warning, and in particular to a linkage early warning method based on security situation assessment and its application. Background Art

[0002] With the expansion of hydropower plant scale and the continuous advancement of technology, the requirements for hydropower plant operation safety are becoming higher and higher. However, the existing early warning systems often only focus on a single operating parameter or equipment status, lacking a comprehensive assessment of the overall safety situation of the hydropower plant. In a complex and changing operating environment, the monitoring of a single parameter often cannot accurately reflect the safety status of the hydropower plant, which can easily lead to missed reports or false reports, thereby increasing the risk of major accidents such as flooding of the plant.

[0003] In addition, the existing early warning system often has problems of slow response and low processing efficiency when dealing with emergencies. Due to the lack of rapid processing and analysis capabilities for large amounts of real-time data, the system is unable to identify potential safety hazards in a timely manner, resulting in the inability to take effective response measures in the first place. This not only increases the possibility of accidents, but may also have serious consequences for the operation of the hydropower plant.

[0004] Therefore, developing an early warning system that can comprehensively assess the safety situation of hydropower plants and improve the efficiency of handling emergencies has become an urgent issue to be solved in the current field of hydropower plant safety management. Summary of the invention

[0005] The main purpose of the present invention is to provide a linkage warning method based on security situation assessment and its application, so as to solve the problems of low efficiency, lack of flexibility and inability to respond to emergencies in real time when processing complex data in the prior art.

[0006] In order to solve the above technical problems, the technical solution adopted by the present invention is: a linkage early warning method based on security situation assessment, comprising the following steps: S1. Data preprocessing and feature extraction: normalize, filter noise, and reduce dimension of the original data, and extract features from the preprocessed data; S2. Construct an abnormal assessment model and a normal assessment model: input the processed feature data into the neural network for training, and adjust the network weights and thresholds so that the model can accurately distinguish between normal and abnormal states; input the data under normal and abnormal states into the convolutional neural network model respectively, and construct a safety assessment model under normal and abnormal states, that is, a normal assessment model for processing normal conditions and an abnormal assessment model for processing abnormal conditions. The cross entropy loss function is used to measure the difference between the model output and the actual label, and the Adam optimizer is used to optimize the network weights; S3. Evaluate safety situation: Use the safety evaluation function to evaluate the real-time collected data, determine the current safety status of the hydropower plant, and output a safety parameter ; S4. Constructing decision function: based on security parameters and environmental factors EnvFactors, through the decision function Determine whether the current state is abnormal; S5. Real-time determination of safety status: Dynamically select and determine the abnormal assessment model or the normal assessment model for real-time assessment through step S4 to determine whether the current status requires an early warning; S6, linkage warning: after it is determined that a warning is needed in step S5, the emergency command system and the warning information push system are linked to conduct warning and emergency response, and a warning signal is automatically issued to remind the hydropower plant of the possible safety risks; The above steps can be used to implement linkage warning of security situation assessment.

[0007] In the preferred embodiment, step S1 specifically includes the following sub-steps: S11. Raw data preprocessing: normalize the raw data, use Kalman filter to filter the noise of the raw data, and use principal component analysis technology to reduce the dimension of the data and extract the main features of the data; S12. Feature extraction: The preprocessed data is used to calculate the mean and standard deviation within the sliding window, extract the features reflecting the operating status of the hydropower plant, use time series analysis to extract trend and periodic features, and use frequency domain analysis methods to identify key frequency components.

[0008] In the preferred solution, in step S1, the raw data includes: water level sensor data, meteorological data, rainfall data and video monitoring data; Water level sensor data: water level data; Meteorological data: rainfall probability predicted by the meteorological department; Rainfall data: including rainfall amount and rainfall duration; Video surveillance data: Use video surveillance data to monitor the environment around the factory in real time, detect potential risk factors including infrastructure damage, human activity risks, and equipment failure and maintenance needs, and quantify them as video surveillance data.

[0009] In the preferred solution, in step S2, the processed feature data is input into the neural network for training, and a loss function is used to quantify the difference between the predicted probability and the actual label. The loss function is specifically expressed as: ; in, is the label of whether a flooding accident actually occurred, which is 0 or 1, and N is the number of training samples. The model predicts the The probability of a flooding accident occurring in a sample, , and They are The water level, rainfall and rainfall probability of each sample.

[0010] In the preferred solution, in step S2, the Adam optimizer is used to adjust the network weights. The specific expression of the Adam optimizer is as follows: ; in, and is a bias-corrected version of the first- and second-order moment estimates of the gradients of water level, rainfall, and rainfall probability, Represents the model parameters, which are continuously updated by the Adam optimizer. Point to the The model parameter values ​​at step , Point to the The model parameter values ​​at step , represents the learning rate, Indicates the water level, Indicates rainfall, represents the probability of rainfall predicted by the meteorological department, Is a numeric value to prevent the denominator from being zero.

[0011] In the preferred solution, in step S3, the safety assessment function expression is as follows: ; in, represents the CNN model, represents the model parameters, Indicates the water level, Indicates rainfall, Indicates the probability of rainfall predicted by the meteorological department.

[0012] In the preferred solution, in step S4, a decision function is constructed To determine whether the real-time data is abnormal, the decision function Specifically expressed as: ; Based on the security parameters in step S3 and environmental factors EnvFactors, through the decision function Determine whether real-time data is abnormal; Among them, EnvFactors includes environmental factors such as time, season, and frequency of historical events. are the parameters of the decision function, including weights and thresholds; decision function This is done based on machine learning or heuristic rules to ensure that the most appropriate model is selected for evaluation in different situations.

[0013] In the preferred solution, in step S5, dynamically selecting and determining whether to use an abnormal assessment model or a normal assessment model for real-time assessment specifically includes: S51, according to step S4, whether the real-time data is abnormal; S52, if the real-time data is judged to be in an abnormal state, verify and evaluate it using an abnormal evaluation model; S53, if the real-time data is judged to be in a normal state, the normal evaluation model is used for verification and evaluation. If a normal result is obtained, the judgment is terminated and data collection continues. If an abnormal result is obtained, the abnormal evaluation model is used to continue the evaluation; S54. Output abnormal results and issue a warning.

[0014] The present invention also provides a linkage early warning system based on security situation assessment, which includes a data collection layer, a data processing layer and a decision-making layer; Data collection layer: real-time monitoring of various operating parameters of the hydropower plant, including water level sensor data, weather forecasts from the meteorological department, rainfall data, and video surveillance data; Data processing layer: pre-processing, feature extraction and pattern recognition of collected data to provide effective information for the decision-making layer; Decision-making layer: Based on the neural network safety situation assessment model, the safety status of the hydropower plant is assessed and early warning signals are issued; The system is used to implement a linkage early warning method based on security situation assessment.

[0015] In the preferred solution, the system also includes: a water level monitoring system, a video monitoring system, a rainfall warning system, a real-time data analysis system, an emergency command system and a warning information push system; The water level monitoring system: a water level sensor is installed in the plant to monitor whether there is water in the plant in real time. When the water level exceeds the preset safety threshold, the system automatically sends out an early warning signal; The rainfall warning system collects data from the meteorological department and predicts rainfall conditions in the future. When heavy rainfall is predicted, the system sends out a warning signal. The video monitoring system: cameras are arranged in important areas inside and outside the factory to monitor the real-time situation of the factory; Real-time data analysis system: collects water level, rainfall, and video surveillance data, and predicts the risk of possible factory flooding accidents through data analysis and processing. When the risk reaches a certain level, the system automatically triggers an early warning signal; The emergency command system: upon receiving the early warning signal, immediately starts the emergency command process and notifies the person to take measures; The warning information push system pushes warning information to relevant personnel via SMS, phone calls or APP to ensure timely awareness of dangerous situations.

[0016] The present invention also provides an application of a linkage early warning method based on security situation assessment, which is applied to the fields of hydrological monitoring and disaster prevention in hydropower stations, reservoirs, and river basin management. Through efficient integration and intelligent analysis of multi-source monitoring data, it can identify potential flood risks, equipment failures, and safety hazards at an early stage under complex and changeable natural environments and operating conditions, and achieve accurate early warning and rapid response to disasters. The application is implemented according to the linkage early warning method based on security situation assessment.

[0017] The present invention also provides an electronic device, comprising a memory and a processor, wherein the memory stores a computer program, and the processor is configured to run the computer program to execute a linkage warning method based on security situation assessment.

[0018] The present invention also provides a readable storage medium, in which a computer program is stored. The computer program includes a program code for controlling a process to execute a process, and the process includes a linkage early warning method based on security situation assessment.

[0019] The present invention provides a linkage early warning method based on security situation assessment and its application, which integrates multi-source data such as water level, meteorology, video surveillance, etc., and uses neural networks to construct abnormal assessment and normal assessment models through data preprocessing and feature extraction to distinguish normal and abnormal states of hydropower plants. The system collects operating parameters in real time. After data processing, the decision-making layer determines whether to trigger an early warning based on the safety assessment function, environmental factors and decision-making functions, and dynamically selects an applicable model for verification and evaluation. When an abnormality is confirmed, the emergency command and early warning push system are linked to start the response and notify relevant personnel. The whole system covers subsystems such as water level monitoring, rainfall early warning, video surveillance, etc., to achieve all-round monitoring, intelligent early warning and rapid response of safety risks in hydropower plants, significantly improve the efficiency of safety management, prevent problems before they occur, and ensure production safety and personnel life safety.

[0020] The beneficial effects of the present invention are: (1) Multi-source data fusion and in-depth analysis: By integrating information from multiple sources such as water level sensor data, meteorological data, and video surveillance data, we can achieve all-round monitoring of the operating status of the hydropower plant. Professional pre-processing and feature extraction of these data fully utilizes the advantages of multi-source data and improves the comprehensiveness and accuracy of situational awareness. (2) Accurate safety situation assessment: construct abnormal assessment models and normal assessment models, and use neural network technology to accurately distinguish between normal and abnormal conditions of hydropower plants; (3) Real-time warning and intelligent decision-making: Decision-making functions are constructed based on safety parameters and environmental factors to realize intelligent judgment of real-time data, which can timely identify potential risks. Real-time verification and evaluation are carried out by dynamically selecting evaluation models to ensure the accuracy and timeliness of warnings. Once an abnormal state is determined, the emergency command system and the early warning information push system are immediately linked to quickly start the emergency plan and notify relevant personnel, which significantly shortens the risk response time. (4) Systematic risk prevention and control system: A linkage early warning system is built, which includes data collection layer, data processing layer, decision-making layer, as well as subsystems such as water level monitoring, rainfall warning, video monitoring, real-time data analysis, emergency command, and early warning information push. This forms a complete closed loop from data acquisition, processing, analysis to decision-making and response, effectively improving the integrity and coordination of risk management in hydropower plants; (5) Significantly improve the level of safety management: The application of the present invention significantly enhances the hydropower plant's ability to identify, warn and respond to various safety risks, helps prevent major safety accidents, ensures the continuity and stability of production operations, thereby reducing economic losses caused by downtime, failures or accidents, and protecting personnel lives. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] The present invention will be further described below in conjunction with the accompanying drawings and embodiments: Figure 1 It is a flow chart of the linkage early warning method based on security situation assessment of the present invention; Figure 2 It is an overall flow chart of the linkage early warning system based on security situation assessment of the present invention. DETAILED DESCRIPTION

[0022] Example 1 like Figure 1~2 As shown, a linkage warning method based on security situation assessment includes the following steps: S1. Data preprocessing and feature extraction: normalize, filter noise, and reduce dimension of the original data, and extract features from the preprocessed data; S2. Construct an abnormal assessment model and a normal assessment model: input the processed feature data into the neural network for training, and adjust the network weights and thresholds so that the model can accurately distinguish between normal and abnormal states; input the data under normal and abnormal states into the convolutional neural network model respectively, and construct a safety assessment model under normal and abnormal states, that is, a normal assessment model for processing normal conditions and an abnormal assessment model for processing abnormal conditions. The cross entropy loss function is used to measure the difference between the model output and the actual label, and the Adam optimizer is used to optimize the network weights; S3. Evaluate safety situation: Use the safety evaluation function to evaluate the real-time collected data, determine the current safety status of the hydropower plant, and output a safety parameter ; S4. Constructing decision function: based on security parameters and environmental factors EnvFactors, through the decision function Determine whether the current state is abnormal; S5. Real-time determination of safety status: Dynamically select and determine the abnormal assessment model or the normal assessment model for real-time assessment through step S4 to determine whether the current status requires an early warning; S6, linkage warning: after it is determined that a warning is needed in step S5, the emergency command system and the warning information push system are linked to conduct warning and emergency response, and a warning signal is automatically issued to remind the hydropower plant of the possible safety risks; The above steps can be used to implement linkage warning of security situation assessment.

[0023] The present invention proposes a linkage early warning method based on safety situation assessment, which integrates multi-source data such as water level, meteorology, and video surveillance. First, the original data is preprocessed and feature extracted, and then the processed feature data is input into the neural network for training, so that the model can accurately distinguish between normal and abnormal states, respectively use the normal data and abnormal data in the sample library to build a normal assessment model and an abnormal assessment model, use the safety assessment function combined with environmental factors to build a decision function, judge whether the real-time data is abnormal, and thus select a suitable model for evaluation, accurately distinguish the normal and abnormal states of the hydropower plant, judge whether to trigger an early warning, and realize real-time assessment and judgment of the safety situation. When the abnormality is confirmed, the emergency command and early warning push system are linked to start the response and notify relevant personnel, forming a complete closed loop from data acquisition, processing, analysis to decision-making and response. The method significantly improves the risk identification, early warning and response capabilities of the hydropower plant, effectively prevents safety accidents, ensures continuous and stable production, reduces economic losses, protects personnel safety, and has significant social and economic benefits.

[0024] In the preferred embodiment, step S1 specifically includes the following sub-steps: S11. Raw data preprocessing: normalize the raw data, use Kalman filter to filter the noise of the raw data, and use principal component analysis technology to reduce the dimension of the data and extract the main features of the data; S12. Feature extraction: The preprocessed data is used to calculate the mean and standard deviation within the sliding window, extract the features reflecting the operating status of the hydropower plant, use time series analysis to extract trend and periodic features, and use frequency domain analysis methods to identify key frequency components.

[0025] In the above step S11, normalization processing is used to unify the scale of data of different dimensions or amplitude ranges, eliminate the interference caused by unit and size differences between data, and facilitate the convergence and comparison of neural network models; Kalman filtering is used to remove random noise in the original data through a recursive filtering algorithm to improve data quality and ensure that the model training focuses on real signals rather than meaningless fluctuations; and principal component analysis (PCA) reduces the dimension through linear transformation, extracts the main eigenvectors of the data, compresses the data set while retaining most of the information, simplifies the model input, reduces the computational burden, and improves the model generalization ability.

[0026] In the above step S12, the sliding window statistics are used to calculate the mean and standard deviation of the data within a certain time window, so as to capture the short-term fluctuation characteristics of the operation status of the hydropower plant and reflect the local stability and abnormal changes. Time series analysis is used to reveal the trend and periodic pattern of the data evolution over time, which helps to understand the long-term laws and seasonal influences of the operation of the hydropower plant and provide the model with valuable clues about future state changes. In addition, frequency domain analysis can identify key frequency components in the data, such as low-frequency trends, medium- and high-frequency fluctuations, etc. These frequency characteristics are often associated with specific operating problems or fault modes, which helps the model distinguish between normal and abnormal states.

[0027] In the preferred solution, in step S1, the raw data includes: water level sensor data, meteorological data, rainfall data and video monitoring data; Water level sensor data: water level data; Meteorological data: rainfall probability predicted by the meteorological department; Rainfall data: including rainfall amount and rainfall duration; Video surveillance data: Use video surveillance data to monitor the environment around the factory in real time, detect potential risk factors including infrastructure damage, human activity risks, and equipment failure and maintenance needs, and quantify them as video surveillance data.

[0028] The system integrates water level sensor data, meteorological data, rainfall data, and video surveillance data to achieve multi-dimensional and all-round monitoring of the safety situation of hydropower plants, ensure comprehensive coverage of risk factors, improve the accuracy and sensitivity of risk identification, provide rich and diverse inputs for neural network model training, and strongly support safety situation assessment and early warning, thereby enhancing the safety management and emergency response capabilities of hydropower plants.

[0029] In the preferred solution, in step S2, the processed feature data is input into the neural network for training, and a loss function is used to quantify the difference between the predicted probability and the actual label. The loss function is specifically expressed as: ; in, is the label of whether a flooding accident actually occurred, which is 0 or 1, and N is the number of training samples. The model predicts the The probability of a flooding accident occurring in a sample, , and They are The water level, rainfall and rainfall probability of each sample.

[0030] The neural network training is performed on the processed feature data using a specific loss function. The advantage is that it is goal-oriented training and focuses on flood risk identification. It considers multiple factors comprehensively, including the impact of water level, rainfall, and rainfall probability on flood risk. It also quantitatively evaluates model performance, minimizes losses to approach the ideal prediction state, and accurately distinguishes between flood accidents under normal and abnormal conditions.

[0031] In the preferred solution, in step S2, the Adam optimizer is used to adjust the network weights. The specific expression of the Adam optimizer is as follows: ; in, and is a bias-corrected version of the first- and second-order moment estimates of the gradients of water level, rainfall, and rainfall probability, Represents the model parameters, which are continuously updated by the Adam optimizer. Point to the The model parameter values ​​at step , Point to the The model parameter values ​​at step , represents the learning rate, Indicates the water level, Indicates rainfall, represents the probability of rainfall predicted by the meteorological department, Is a numeric value to prevent the denominator from being zero.

[0032] The Adam optimizer is used to adjust the network weights. Its advantages lie in its adaptive learning rate, which combines the momentum term with RMSProp to accelerate training and improve convergence efficiency. It also has efficient memory usage, requiring only the storage of moving average values, saving resources, and is suitable for large-scale network training. It also has stability and robustness, introducing a bias correction term to mitigate the gradient estimation bias and avoid excessive learning rates in the early stages of training.

[0033] In the preferred solution, in step S3, the security situation function expression is as follows: ; in, represents the CNN model, represents the model parameters, Indicates the water level, Indicates rainfall, Indicates the probability of rainfall predicted by the meteorological department.

[0034] The above safety situation function, with the CNN model as the core, combines key data such as water level, rainfall and rainfall probability to achieve accurate quantitative assessment of the current safety status of the hydropower plant. This function plays a core role in safety assessment in the overall early warning system. Through deep learning technology, it conducts in-depth fusion analysis of multiple factors to ensure the comprehensiveness and accuracy of the assessment, and can dynamically respond to risk changes. The output safety parameters and environmental parameters together construct a decision function , so as to further select the normal evaluation model or the abnormal evaluation model for real-time data evaluation.

[0035] In the preferred solution, in step S4, a decision function is constructed To determine whether the real-time data is abnormal, the decision function Specifically expressed as: ; Based on the security parameters in step S3 and environmental factors EnvFactors, through the decision function Determine whether real-time data is abnormal; Among them, EnvFactors includes environmental factors such as time, season, and frequency of historical events. are the parameters of the decision function, including weights and thresholds; decision function This is done based on machine learning or heuristic rules to ensure that the most appropriate model is selected for evaluation in different situations.

[0036] The decision function plays a key role in model selection in the linkage early warning method. It dynamically combines real-time safety parameters with environmental factors, including time, season, and historical event frequency, to ensure that the early warning system selects the most suitable assessment model, that is, the normal or abnormal assessment model, under different operating environments and risk conditions, thereby enhancing the system's environmental adaptability and risk identification accuracy. By dynamically adjusting the assessment strategy, the early warning system can flexibly switch between normal / abnormal models, avoiding over-calculation at low risk and ensuring in-depth analysis at high risk. In addition, the decision function can also optimize resource allocation and improve the operating efficiency of the early warning system, especially maintaining stability and response speed when processing large amounts of real-time data and complex analysis tasks. Based on machine learning or heuristic rules, the decision function can continuously learn and improve the accuracy of model selection, further improving the risk identification ability and prediction accuracy of the early warning system, and reducing false positives and false negatives.

[0037] In the preferred solution, in step S5, dynamically selecting and determining whether to use an abnormal assessment model or a normal assessment model for real-time assessment specifically includes: S51, according to step S4, whether the real-time data is abnormal; S52, if the real-time data is judged to be in an abnormal state, verify and evaluate it using an abnormal evaluation model; S53, if the real-time data is judged to be in a normal state, the normal evaluation model is used for verification and evaluation. If a normal result is obtained, the judgment is terminated and data collection continues. If an abnormal result is obtained, the abnormal evaluation model is used to continue the evaluation; S54. Output abnormal results and issue a warning.

[0038] The present invention also provides a linkage early warning system based on security situation assessment, which includes a data collection layer, a data processing layer and a decision-making layer.

[0039] Data collection layer: Real-time monitoring of various operating parameters of the hydropower plant, including water level sensor data, weather forecasts from the meteorological department, rainfall data, and video surveillance data.

[0040] Data processing layer: pre-processes, extracts features and recognizes patterns on the collected data to provide effective information for the decision-making layer.

[0041] Decision-making layer: Based on the neural network safety situation assessment model, the safety status of the hydropower plant is evaluated and early warning signals are issued.

[0042] The system is used to implement a linkage early warning method based on security situation assessment.

[0043] The linkage early warning system of the present invention is composed of a data acquisition layer, a data processing layer and a decision-making layer, which can realize comprehensive monitoring, accurate evaluation and timely early warning of the safety situation of the hydropower plant; the data acquisition layer monitors various operating parameters in real time, covering multi-source data such as water level, weather, rainfall, video surveillance, etc., to ensure the comprehensiveness of risk information; the data processing layer improves data quality through preprocessing, feature extraction and pattern recognition, explores deep risk associations, and provides effective information for decision-making. The decision-making layer evaluates the safety situation based on the neural network model, accurately judges the risk level, and issues early warning signals in a timely manner. The system has constructed a complete chain from data acquisition, processing to decision-making, realizing the intelligent and refined management of the safety situation of the hydropower plant, effectively preventing risks, improving emergency response efficiency, and ensuring the safe and stable operation of the hydropower plant.

[0044] In the preferred solution, the system also includes: a water level monitoring system, a video monitoring system, a rainfall warning system, a real-time data analysis system, an emergency command system and a warning information push system.

[0045] The water level monitoring system: a water level sensor is set in the factory to monitor in real time whether there is water in the room. When the water level exceeds a preset safety threshold, the system automatically sends out an early warning signal.

[0046] The rainfall warning system collects data from the meteorological department and predicts rainfall conditions in the future. When heavy rainfall is predicted, the system sends out a warning signal.

[0047] The video monitoring system: cameras are arranged in important areas inside and outside the factory to monitor the real-time situation of the factory.

[0048] Real-time data analysis system: collects water level, rainfall, and video surveillance data, and predicts the risk of possible factory flooding accidents through data analysis and processing. When the risk reaches a certain level, the system automatically triggers an early warning signal.

[0049] The emergency command system: upon receiving the early warning signal, immediately initiates the emergency command process and notifies the taking of measures.

[0050] The warning information push system pushes warning information to relevant personnel via SMS, phone calls or APP to ensure timely awareness of dangerous situations.

[0051] Each system constitutes a comprehensive, multi-level linkage early warning system, which significantly improves the safety early warning and emergency response capabilities of the hydropower plant; the water level monitoring system monitors the water level in the plant in real time, automatically warns when it exceeds the threshold, and promptly detects the risk of flooding. The rainfall early warning system uses meteorological data to predict rainfall, warns of water level rises that may be caused by heavy rainfall, and takes precautions in advance. The video surveillance system intuitively displays the on-site situation and assists in identifying other risk factors; the real-time data analysis system integrates multi-source data, predicts the risk of flooding accidents, and accurately triggers early warnings. The emergency command system quickly starts the emergency process and guides personnel to take countermeasures; the linkage alarm system realizes internal and external information sharing and linkage, expands the scope of early warning, and requests external support. The early warning information push system ensures that the early warning information is transmitted to relevant personnel in a timely manner, improving the response speed; in summary, this linkage early warning system realizes comprehensive risk monitoring, accurate prediction, rapid response and effective communication, and effectively ensures the safe operation of the hydropower plant.

[0052] Example 2 Further illustrate with reference to Example 1, Figure 1-2 As shown, a linkage warning method based on security situation assessment is as follows: Step 1: Build a network security situation indicator system The raw data include: water level sensor data, meteorological data, rainfall data and video surveillance data.

[0053] Water level sensor data: Water level data; CNN predicts possible water level rise by analyzing trends and abnormal patterns in sensor data.

[0054] Meteorological data: The probability of rainfall predicted by the meteorological service; meteorological data, such as the predicted amount and duration of rainfall, can also be included in the analysis to assess weather conditions that may affect water levels.

[0055] Rainfall data: including rainfall amount and duration; analysis of rainfall data, especially heavy rainfall events, to estimate potential impacts on reservoir levels.

[0056] Video surveillance data: Use video surveillance data to monitor the environment around the plant in real time, detect water level changes and other potential risk factors, and quantify them as video surveillance data.

[0057] Step 2: Build a neural network security situation assessment model (1) Data preprocessing: normalize and filter the raw data to reduce noise interference and improve the accuracy of the evaluation. Use the Kalman filter to filter the noise of the raw data to improve the data quality. Use the principal component analysis (PCA) technology to reduce the dimension of the data to improve the efficiency and accuracy of subsequent processing.

[0058] (2) Feature extraction: Extract representative features from the processed data, such as sliding mean, standard deviation, etc., to reflect the operating status of the hydropower plant. Use time series analysis to extract time-related features in the data, such as trends and periodicity, and use frequency domain analysis methods to identify key frequency components in the data.

[0059] (3) Training the neural network: Input the processed feature data into the neural network for training, adjust the network weights and thresholds so that the neural network can correctly identify normal and abnormal states. Convolutional neural networks (CNNs) are selected to process and identify complex data patterns.

[0060] (4) Constructing an abnormal assessment model and a normal assessment model: The data under normal and abnormal conditions are respectively input into the convolutional neural network model to construct a safety assessment model under normal and abnormal conditions, that is, a normal assessment model for processing normal conditions and an abnormal assessment model for processing abnormal conditions. The cross entropy loss function is used to measure the difference between the model output and the actual label, and the Adam optimizer is used to optimize the network weights.

[0061] In the present invention, convolutional neural network (CNN) is selected to process simple and complex data patterns respectively. CNN effectively extracts features from input data through its multi-layer structure. These layers include convolutional layers, pooling layers and fully connected layers, each of which performs specific operations on the input data to extract and learn features.

[0062] Convolutional layers: These layers use multiple small receptive regions to extract local features of the input data through convolution operations, where each convolution kernel focuses on capturing a specific aspect of the data, such as edges or textures.

[0063] Pooling Layer: Pooling layers are used to reduce the spatial dimension of the data while keeping important information, which helps reduce the amount of computation and prevent over-pooling.

[0064] Fully connected layer: After the data passes through a series of convolution and pooling layers, the fully connected layer is used to integrate the learned features and perform classification or regression tasks.

[0065] During training, we use the cross entropy loss function to measure the difference between the CNN output and the actual label. The cross entropy loss function L can be expressed as: ; in, is the label of whether a flooding accident actually occurred, which is 0 or 1, and N is the number of training samples. represents the probability of a flooding accident occurring in the 𝑖th sample predicted by the model, , and They are The water level, rainfall and rainfall probability of each sample.

[0066] In addition, the Adam optimizer is used to adjust the network weights. The Adam optimizer combines the advantages of momentum and RMSprop and is suitable for dealing with large-scale and non-convex optimization problems. The specific formula of the Adam optimizer is: ; in, and is a bias-corrected version of the first- and second-order moment estimates of the gradients of water level, rainfall, and rainfall probability, Represents the model parameters, which are continuously updated by the Adam optimizer. Point to the The model parameter values ​​at step , Point to the The model parameter values ​​at step , represents the learning rate, Indicates the water level, Indicates rainfall, represents the probability of rainfall predicted by the meteorological department, Is a numeric value to prevent the denominator from being zero.

[0067] By combining the powerful feature extraction capabilities of CNN and the cross-entropy loss function with the efficient optimization strategy of the Adam optimizer, the performance and accuracy of the model can be effectively improved when processing complex network security data.

[0068] Among them, the convolution layer and the pooling layer work together on the input data to extract key features.

[0069] Convolutional layer (Conv): ; is the input data, here is the water level , rainfall and probability of rainfall A combination of and are the weights and bias parameters of the convolutional layer; Represents a convolution operation, which is used to extract features of the input data.

[0070] Activation function (ReLU): ; This is a nonlinear function used to increase the expressiveness of the model, allowing it to learn more complex features. The ReLU activation function increases the model's ability to recognize nonlinear relationships.

[0071] Pooling layer (Pool): ; The pooling layer performs dimensionality reduction on Z to reduce the amount of computation while retaining important features.

[0072] (5) Evaluate safety situation: Use the safety evaluation function to evaluate the real-time collected data, determine the current safety status of the hydropower plant, and output a safety parameter ; About safety status evaluation: The safety status evaluation functions are expressed as ; This function comprehensively analyzes the water level , rainfall and probability of rainfall To assess the security status.

[0073] The security assessment function can be expressed as: .

[0074] It is a sigmoid activation function that maps parameters to between 0 and 1, indicating the probability of security risk. Comprehensively consider multiple input parameters to predict the safety risks of hydropower plants under normal and abnormal conditions. and are the weights and biases of the scoring function, is the output of the last layer of the CNN model, where represents the CNN model, Represents the parameters of the model. The function outputs a safety parameter that reflects the current safety status of the hydropower plant.

[0075] (6) Constructing decision functions: based on security parameters and environmental factors EnvFactors, through the decision function Determine whether the current state is abnormal; decision function Specifically expressed as: ; This decision function is based on the security parameter and environmental factors EnvFactors to select the appropriate model. EnvFactors can include environmental factors such as time, season, frequency of historical events, etc. are the parameters of the decision function, which may include weights and thresholds, used to determine whether to use a standard model or a complex model. The decision function This can be implemented based on machine learning or heuristic rules to ensure that the most appropriate model is selected for evaluation in different situations.

[0076] (7) Real-time determination of safety status: Through step S4, dynamically select and determine the abnormal assessment model and the normal assessment model for real-time assessment to determine whether the current status requires an early warning.

[0077] (8) Joint warning: After it is determined in step S5 that a warning is needed, the emergency command system and the warning information push system are linked to carry out warning and emergency response, and a warning signal is automatically issued to alert the hydropower plant of possible safety risks.

[0078] Through this mechanism, our system can flexibly adjust the assessment model according to the actual situation, ensuring accurate and efficient safety assessment under different operating conditions. This not only improves the response speed of the early warning system, but also enhances the ability to identify potential risks, greatly improving the overall safety management level of the hydropower plant.

[0079] Step 3: Linking comprehensive systems for early warning Water level monitoring system: A water level sensor is installed in the plant to monitor whether there is water in the room in real time. When the water level exceeds the preset safety threshold, the system automatically sends an early warning signal and transmits it to the system.

[0080] Rainfall warning system: collects data from the meteorological department and predicts rainfall conditions in the future. When heavy rainfall is predicted, the system sends a warning signal and transmits it to the system.

[0081] Video surveillance system: Cameras are placed in important areas inside and outside the factory to monitor the real-time situation of the factory.

[0082] Real-time data analysis system: collects data such as water level, rainfall, video surveillance, etc., and predicts the risk of possible factory flooding accidents through data analysis and processing. When the risk reaches a certain level, the system automatically triggers an early warning signal.

[0083] Emergency command system: After receiving the early warning signal, the emergency command process is immediately initiated to notify relevant departments and personnel to take measures, such as starting drainage equipment and organizing personnel evacuation.

[0084] Early warning information push system: Early warning information will be pushed to relevant personnel and persons in charge in a timely manner through SMS, phone calls, APP, etc., to ensure that everyone can understand the dangerous situation in a timely manner.

[0085] According to the early warning results, take corresponding emergency measures. The purpose of early warning is to timely discover and deal with network security risks. When the early warning is triggered, according to the early warning results, emergency measures such as strengthening network security protection, restricting network access rights, and conducting security drills can be taken to ensure network security.

[0086] Example 3 Further illustrated in combination with Example 1 and Example 2, an application of a linkage early warning method based on security situation assessment is applied to the fields of hydrological monitoring and disaster prevention in hydropower stations, reservoirs, and river basin management. Through efficient integration and intelligent analysis of multi-source monitoring data, it can identify potential flood risks, equipment failures, and safety hazards at an early stage under complex and changeable natural environments and operating conditions, and achieve accurate early warning and rapid response to disasters. The application is implemented according to the linkage early warning method based on security situation assessment.

[0087] Example 4 Further described in combination with Example 1 and Example 2, an electronic device includes a memory and a processor, wherein a computer program is stored in the memory, and the processor is configured to run the computer program to execute a linkage warning method based on security situation assessment.

[0088] Example 5 Further illustrated in combination with Example 1 and Example 2, a readable storage medium stores a computer program, wherein the computer program includes a program code for controlling a process to execute a process, wherein the process includes a linkage warning method based on security situation assessment.

[0089] The above embodiments are only preferred technical solutions of the present invention and should not be regarded as limiting the present invention. The protection scope of the present invention shall be the technical solutions recorded in the claims, including equivalent replacement solutions of the technical features in the technical solutions recorded in the claims. That is, equivalent replacement improvements within this scope are also within the protection scope of the present invention.

Claims

1. A linkage early warning method based on security situation assessment, characterized by: The following steps are involved: S1. Data preprocessing and feature extraction: normalize, filter noise, and reduce dimensionality of the original data, and extract features from the preprocessed data; S2. Construct an abnormal assessment model and a normal assessment model: input the processed feature data into the neural network for training, and adjust the network weights and thresholds so that the model can accurately distinguish between normal and abnormal states; input the data under normal and abnormal states into the convolutional neural network model respectively, and construct a safety assessment model under normal and abnormal states, that is, a normal assessment model for processing normal conditions and an abnormal assessment model for processing abnormal conditions. The cross entropy loss function is used to measure the difference between the model output and the actual label, and the Adam optimizer is used to optimize the network weights; S3. Evaluate safety situation: Use the safety evaluation function to evaluate the real-time collected data, determine the current safety status of the hydropower plant, and output a safety parameter ; S4. Constructing decision function: based on security parameters and environmental factors EnvFactors, through the decision function Determine whether the current state is abnormal; S5. Real-time determination of safety status: Dynamically select and determine the abnormal assessment model or the normal assessment model for real-time assessment through step S4 to determine whether the current status requires an early warning; S6, linkage warning: after it is determined that a warning is needed in step S5, the emergency command system and the warning information push system are linked to conduct warning and emergency response, and a warning signal is automatically issued to remind the hydropower plant of the possible safety risks; The above steps can be used to implement linkage warning of security situation assessment.

2. The linkage early warning method based on security situation assessment according to claim 1 is characterized in that: The step S1 specifically includes the following sub-steps: S11. Raw data preprocessing: normalize the raw data, use Kalman filter to filter the noise of the raw data, and use principal component analysis technology to reduce the dimension of the data and extract the main features of the data; S12. Feature extraction: The preprocessed data is used to calculate the mean and standard deviation within the sliding window, extract the features reflecting the operating status of the hydropower plant, use time series analysis to extract trend and periodic features, and use frequency domain analysis methods to identify key frequency components.

3. The linkage early warning method based on security situation assessment according to claim 1 is characterized in that: In step S1, the raw data includes: water level sensor data, meteorological data, rainfall data and video monitoring data; Water level sensor data: water level data; Meteorological data: rainfall probability predicted by the meteorological department; Rainfall data: including rainfall amount and rainfall duration; Video surveillance data: Use video surveillance data to monitor the environment around the factory in real time, detect potential risk factors including infrastructure damage, human activity risks, and equipment failure and maintenance needs, and quantify them as video surveillance data.

4. The linkage early warning method based on security situation assessment according to claim 1 is characterized in that: In step S2, the processed feature data is input into the neural network for training, and a loss function is used to quantify the difference between the predicted probability and the actual label. The loss function is specifically expressed as: ； in, is the label of whether a flooding accident actually occurred, which is 0 or 1, and N is the number of training samples. The model predicts the The probability of a flooding accident occurring in a sample, , and They are The water level, rainfall and rainfall probability of each sample.

5. The linkage early warning method based on security situation assessment according to claim 1 is characterized in that: In step S2, the Adam optimizer is used to adjust the network weights. The specific expression of the Adam optimizer is as follows: ； in, and is a bias-corrected version of the first- and second-order moment estimates of the gradients of water level, rainfall, and rainfall probability, Represents the model parameters, which are continuously updated by the Adam optimizer. Point to the The model parameter values ​​at step , Point to the The model parameter values ​​at step , represents the learning rate, Indicates the water level, Indicates rainfall, represents the probability of rainfall predicted by the meteorological department, Is a numeric value to prevent the denominator from being zero.

6. The linkage early warning method based on security situation assessment according to claim 1 is characterized in that: In step S3, the security assessment function expression is as follows: ； in, represents the CNN model, represents the model parameters, Indicates the water level, Indicates rainfall, Indicates the probability of rainfall predicted by the meteorological department.

7. The linkage early warning method based on security situation assessment according to claim 1 is characterized by: In step S4, a decision function is constructed. To determine whether the real-time data is abnormal, the decision function Specifically expressed as: ； Based on the security parameters in step S3 and environmental factors EnvFactors, through the decision function Determine whether real-time data is abnormal; Among them, EnvFactors includes environmental factors such as time, season, and frequency of historical events. are the parameters of the decision function, including weights and thresholds; the decision function This is done based on machine learning or heuristic rules to ensure that the most appropriate model is selected for evaluation in different situations.

8. The linkage early warning method based on security situation assessment according to claim 1 is characterized by: In step S5, dynamically selecting and determining whether to use an abnormal assessment model or a normal assessment model for real-time assessment specifically includes: S51, according to step S4, whether the real-time data is abnormal; S52: If the real-time data is judged to be in an abnormal state, the abnormal evaluation model is used for verification and evaluation; S53, if the real-time data is judged to be in a normal state, the normal evaluation model is used for verification and evaluation. If a normal result is obtained, the judgment is terminated and data collection continues. If an abnormal result is obtained, the abnormal evaluation model is used to continue the evaluation; S54. Output abnormal results and issue a warning.

9. A linkage early warning system based on security situation assessment, characterized by: The system includes data collection layer, data processing layer and decision-making layer; Data collection layer: real-time monitoring of various operating parameters of the hydropower plant, including water level sensor data, weather forecasts from the meteorological department, rainfall data, and video surveillance data; Data processing layer: pre-processing, feature extraction and pattern recognition of collected data to provide effective information for the decision-making layer; Decision-making layer: Based on the neural network safety situation assessment model, the safety status of the hydropower plant is assessed and early warning signals are issued; The system is used to implement the linkage warning method based on security situation assessment as described in any one of claims 1-8.

10. The linkage early warning system based on security situation assessment according to claim 9 is characterized in that: The system also includes: a water level monitoring system, a video monitoring system, a rainfall warning system, a real-time data analysis system, an emergency command system and a warning information push system; The water level monitoring system: a water level sensor is installed in the plant to monitor whether there is water in the plant in real time. When the water level exceeds the preset safety threshold, the system automatically sends out an early warning signal; The rainfall warning system collects data from the meteorological department and predicts rainfall conditions in the future. When heavy rainfall is predicted, the system sends out a warning signal. The video monitoring system: cameras are arranged in important areas inside and outside the factory to monitor the real-time situation of the factory; Real-time data analysis system: collects water level, rainfall, and video surveillance data, and predicts the risk of possible factory flooding accidents through data analysis and processing. When the risk reaches a certain level, the system automatically triggers an early warning signal; The emergency command system: upon receiving the early warning signal, immediately starts the emergency command process and notifies the person to take measures; The warning information push system pushes warning information to relevant personnel via SMS, phone calls or APP to ensure timely awareness of dangerous situations.

11. An application of a linkage early warning method based on security situation assessment, characterized in that: Applied to the field of hydrological monitoring and disaster prevention in hydropower stations, reservoirs, and river basin management, through efficient integration and intelligent analysis of multi-source monitoring data, it can identify potential flood risks, equipment failures, and safety hazards at an early stage under complex and changeable natural environments and operating conditions, and achieve accurate early warning and rapid response to disasters. The application is implemented according to the linkage early warning method based on security situation assessment as described in any one of claims 1 to 8.

12. An electronic device comprising a memory and a processor, characterized in that: A computer program is stored in the memory, and the processor is configured to run the computer program to execute the linkage warning method based on security situation assessment as described in any one of claims 1 to 8.

13. A readable storage medium, characterized in that: A computer program is stored in a readable storage medium, wherein the computer program includes a program code for controlling a process to execute a process, wherein the process includes the linkage early warning method based on security situation assessment according to any one of claims 1 to 8.

Images