Nuclear power station cable fire prediction method and system based on artificial intelligence

Abstract

The invention relates to a nuclear power station cable fire prediction method and system based on artificial intelligence, and the method comprises the steps: collecting cable operation state data and environment parameters in real time through a sensor, and extracting key features based on the cable operation state data; based on the environmental parameters and a pre-constructed fire-structure coupling model, obtaining space-time distribution of an internal temperature field of the structure system; superposing the spatial and temporal distribution of the internal temperature field of the structural system with a pre-constructed cable laying digital model to obtain a dynamic cable thermal risk map, and calculating a cable real-time failure probability based on a material tolerance temperature and thermal aging model; and based on the key features, the dynamic cable thermal risk map and the cable real-time failure probability, utilizing a pre-trained Bi-LSTM model to predict the cable fire occurrence probability in real time, and if the occurrence probability is greater than a preset threshold, giving out an early warning and notifying related personnel to take emergency measures. Compared with the prior art, comprehensive monitoring of the cable state can be realized.

Description

Technical Field

[0001] This invention relates to the field of fire safety technology, and in particular to a method and system for predicting cable fires in nuclear power plants based on artificial intelligence. Background Technology

[0002] Nuclear power plants are crucial energy supply facilities, and their safety is paramount. Currently, with the increasing number of nuclear power plants, the risk of fires and nuclear accidents is also rising. Statistics show that the probability of a fire at a nuclear power plant is 0.28 times per reactor. In 2010, the probability of a fire causing an accident in a nuclear safety-related system was 0.44. The greatest danger to nuclear power plants comes from fire, which has become one of the most realistic and direct threats to nuclear power plant safety.

[0003] As a core component for power transmission and signal control in nuclear power plants, cables are used in enormous quantities. Statistics show that the cable length required for every megawatt of installed capacity reaches 3,000 km. A fire in a nuclear power plant can not only damage equipment and cause power outages, but also trigger serious nuclear safety accidents, and even lead to radioactive material leaks, posing a significant threat to the environment and public health. In recent years, numerous nuclear power plant cable fires worldwide have caused severe economic losses and social impacts, highlighting the urgency of preventing and controlling cable fire risks in nuclear power plants. Traditional cable fire monitoring methods mainly rely on temperature sensors and smoke detectors, but these methods suffer from problems such as response lag, high false alarm rates, and the inability to provide early warnings. With the development of artificial intelligence technology, it has become possible to use machine learning and big data analytics to monitor and predict cable conditions in real time. Therefore, developing an AI-based method, system, and terminal for predicting nuclear power plant cable fires has significant practical implications.

[0004] Currently, the fire protection design of nuclear power plants is mainly based on design standards such as the "Code for Fire Protection Design of Nuclear Power Plants" (GB / T22158) and the "Guideline for Fire Hazard Analysis of Nuclear Power Plants" (EJ / T1194). However, because nuclear power plants are high-risk areas with high safety requirements, a fire could lead to the failure of critical equipment, affect the plant's safety systems, and increase the risk of nuclear leakage. Traditional fire detection and extinguishing systems suffer from slow response times and high false alarm rates, making it difficult to meet the high safety requirements of nuclear power plants. Although existing standards are relatively comprehensive, some shortcomings still exist in practical applications and research, mainly as follows: (1) Existing fire risk assessment methods mostly rely on qualitative analysis and lack quantitative tools, resulting in highly subjective results. Traditional methods rely on expert experience, make insufficient use of data, have simple models, are difficult to dynamically reflect complex environmental changes, have limited assessment accuracy, lack quantitative indicators, and cannot meet the needs of real-time monitoring and high-precision early warning for nuclear power plants.

[0005] (2) Existing fire detection technologies are not reliable enough in extreme environments such as high radiation and high temperature. Sensors are easily interfered with, resulting in a high rate of false alarms and false alarms, which makes it difficult to meet the high-precision monitoring needs of complex scenarios such as nuclear power plants.

[0006] (3) Existing fire extinguishing system designs are mostly based on conventional fire scenarios, and do not adequately consider fire scenarios unique to nuclear power plants (such as high radiation and cable trench fires), lacking specificity and adaptability, and are difficult to effectively cope with fire risks in complex environments.

[0007] (4) The existing fire safety management regulations are not implemented with varying degrees of effectiveness in practice. Some nuclear power plants have management loopholes, such as untimely equipment maintenance, insufficient personnel training, and incomplete emergency plans, which affect the overall fire prevention and control effect.

[0008] (5) Existing literature pays little attention to personnel training and emergency response. Some nuclear power plant personnel lack the necessary fire emergency response capabilities, resulting in low response efficiency when a fire occurs and increasing the risk of the accident escalating.

[0009] (6) Traditional studies often use simplified temperature rise curves (such as ISO standard curves) to simulate the fire development process. These curves are usually based on standard fire tests and are applicable to general building fire scenarios. However, nuclear power plant cable fires have special characteristics (such as high radiation, complex environment, and dense cable arrangement). Simplified curves cannot accurately reflect their dynamic characteristics and cannot truly reflect the fire scenario. In fact, due to the non-uniformity of temperature field distribution in nuclear power plant cable fires, the thermal response and electrical performance of cables will change significantly with location and time, leading to local overheating or abnormal electrical parameters, which in turn triggers a chain reaction. Using a uniform temperature field cannot accurately simulate the complex thermo-electric coupling effect in cable fires, nor can it reflect the phenomenon that cables far from the fire source may fail first due to the redistribution of internal forces, thus affecting the accuracy of fire risk assessment and prevention strategies. Summary of the Invention

[0010] The purpose of this invention is to overcome the shortcomings of the existing technology by providing a method and system for predicting nuclear power plant cable fires based on artificial intelligence. This method uses artificial intelligence, combined with multi-source data fusion and deep learning, to predict nuclear power plant cable fires.

[0011] The objective of this invention can be achieved through the following technical solutions: A method for predicting cable fires in nuclear power plants based on artificial intelligence, the method comprising: The cable operating status data and environmental parameters are collected in real time by sensors, and key features are extracted based on the cable operating status data. Based on the environmental parameters and the pre-constructed fire-structure coupling model, the spatiotemporal distribution of the temperature field inside the structural system is obtained; The spatiotemporal distribution of the internal temperature field of the structure system is superimposed with a pre-constructed digital model of cable laying to obtain a dynamic cable thermal risk map, and the real-time failure probability of the cable is calculated based on the material tolerance temperature and thermal aging model. Based on the aforementioned key features, dynamic cable thermal risk map, and real-time cable failure probability, a pre-trained Bi-LSTM model is used to predict the probability of cable fires in real time. If the probability of occurrence exceeds a preset threshold, an early warning is issued and relevant personnel are notified to take emergency measures.

[0012] Furthermore, after collecting the cable's operating status data and environmental parameters in real time through sensors, the collected data is preprocessed, including cleaning, noise reduction, and normalization. The key features extracted based on the cable's operating status data include temperature change rate, current fluctuation, and vibration anomaly.

[0013] Furthermore, the fire-structure coupling model includes a fire-side mathematical model that outputs the temperature field and heat flow field in the fire environment and a structure-side mathematical model that outputs the temperature field evolution inside the structure. The fire-side mathematical model and the structure-side mathematical model are connected through a coupling interface to achieve data exchange.

[0014] Furthermore, the pre-construction process of the fire-structure coupling model includes: Based on nuclear power plant building data and cable laying geometry data, fluid region models and solid region models are constructed respectively. The fluid region model is discretized using a computational fluid dynamics mesh, and the solid region model is discretized using a finite element mesh. A mathematical model of the fire side is established based on the discretized fluid region model and the conservation equations for mass, momentum, energy, and species transport. Based on the discretized solid region model, a three-dimensional unsteady heat conduction equation is used to construct a structural mathematical model. Establish point-to-point data mapping relationships between the discretized fluid region model and the discretized solid region model, consider convection and radiation heat transfer modes, obtain coupling interface conditions, and establish coupling interfaces. Based on the coupling interface, the fire-side mathematical model and the structural mathematical model are connected to obtain the fire-structure coupling model.

[0015] Furthermore, based on the aforementioned environmental parameters and the pre-constructed fire-structure coupling model, the process of obtaining the spatiotemporal distribution of the internal temperature field of the structural system includes: The environmental parameters and key features are compared with preset fluctuation thresholds. If any parameter or feature is greater than the preset fluctuation threshold, the state is determined to be abnormal. Based on the abnormal environmental parameters or key features, preset simulation data of the corresponding fire condition is called and input into the fire-structure coupling model to obtain the spatiotemporal distribution of the temperature field inside the structural system. Otherwise, the state is determined to be normal. Based on the environmental parameters and key features, preset simulation data of the normal condition is called and input into the fire-structure coupling model to obtain the spatiotemporal distribution of the temperature field inside the structural system.

[0016] Furthermore, the pre-setting process for the simulation data corresponding to fire conditions includes: Based on the functional layout of the target nuclear power plant, cable laying routes and safety specifications, safety objectives and performance indicators are determined, and preliminary design schemes are obtained by referring to historical fire cases. Based on the actual conditions of the buildings within the nuclear power plant, potential fire hazards are identified, potential fire scenarios are obtained, and fire conditions corresponding to different fire scenarios are designed according to the preliminary scheme. The fire conditions include different fire power, ventilation conditions, and fuel types. Based on the aforementioned fire conditions, the smoke trajectory, spatial temperature field, and convective heat transfer coefficient under different fire conditions are simulated using a fire dynamics model and output as simulation data for the corresponding fire conditions.

[0017] Furthermore, the digital model of cable laying is constructed based on the geometric drawings of the nuclear power plant building and cable laying. The digital model of cable laying includes the physical direction of the cable, laying location, density, joint distribution, and its relative position with the surrounding structure.

[0018] Furthermore, the process of calculating the real-time failure probability of the cable based on the material's tolerance temperature and thermal aging model includes: To obtain key properties of cable materials, including material temperature resistance, rated aging life at standard temperature, and thermal aging activation energy; Based on the thermal aging model, the accelerating effect of temperature on cable material aging is quantified, and the aging acceleration coefficient of the cable at the current temperature is obtained based on the aging rate at the standard temperature. Based on the cable's operating time and cumulative aging degree, the cumulative aging amount of the cable is calculated, and combined with the material's rated aging life and the cable's aging acceleration coefficient at the current temperature, the real-time remaining cable life is calculated. Based on the remaining lifespan and key characteristics of the cable in real time, the probability of real-time cable failure is calculated.

[0019] Furthermore, the Bi-LSTM model is periodically retrained using newly added real-time collected data and early warning feedback data to optimize model parameters.

[0020] An artificial intelligence-based nuclear power plant cable fire prediction system, the system comprising: Data acquisition module: Collects cable operating status data and environmental parameters in real time through sensors; Data processing module: preprocesses sensor-collected data and extracts key features based on the cable's operating status data; Model training module: pre-built fire-structure coupling model, pre-trained Bi-LSTM model; Fire prediction module: Based on the environmental parameters and the pre-built fire-structure coupling model, the spatiotemporal distribution of the internal temperature field of the structural system is obtained; the spatiotemporal distribution of the internal temperature field of the structural system is superimposed with the pre-built digital model of cable laying to obtain a dynamic cable thermal risk map, and the real-time failure probability of the cable is calculated based on the material tolerance temperature and thermal aging model; based on the key features, the dynamic cable thermal risk map and the real-time failure probability of the cable, the probability of cable fire occurrence is predicted in real time using a pre-trained Bi-LSTM model. Result output module: Outputs the probability of cable fire occurrence. If the probability of occurrence is greater than a preset threshold, an early warning is issued and relevant personnel are notified to take emergency measures.

[0021] Compared with the prior art, the beneficial effects of the present invention include: 1. This invention, through multi-source data fusion and deep learning technology, can identify early signs of cable failure before a fire occurs, accurately output the probability of fire occurrence, and significantly improve the accuracy and reliability of fire prediction. This invention constructs a dual quantitative system of dynamic cable thermal risk maps and real-time failure probabilities, replacing traditional qualitative analysis and expert experience judgment, providing objective data support for operation and maintenance decisions, and meeting the high-precision safety requirements of nuclear power plants. This invention also uses predictive maintenance to identify potential hazards in advance, reducing unplanned downtime and equipment wear, lowering the total lifecycle cost, and dynamically allocating firefighting resources based on risk levels, avoiding resource waste while ensuring safety.

[0022] 2. This invention proposes a fire-structure coupling model that uses fluid dynamics and finite element mesh mapping to transform the temperature field and heat flow field of the fire simulation into the real-time temperature distribution of the cable structure. The coupling interface conditions simultaneously consider convection and radiation heat transfer, accurately calculate the thermal response of the cable at any location and at any time, and solve the problem of the correlation between environmental fire and cable failure.

[0023] 3. This invention addresses the challenges of nuclear power plants with high radiation, dense cabling, and uneven temperature fields by employing a fire-structure coupling model and customized fire condition design. This solves the problem that traditional simplified temperature rise curves cannot accurately simulate real fires, ensuring the reliability of predictions under extreme conditions.

[0024] 4. This invention can continuously collect real-time monitoring data and early warning feedback results, and regularly update the Bi-LSTM model and fire condition simulation data to continuously improve prediction accuracy and adapt to dynamic scenarios such as equipment aging and changes in operating conditions.

[0025] 5. In the early stage, this invention designs and simulates fire conditions to pre-set simulation data of fire conditions; in the middle stage, it achieves accurate risk prediction through fire-structure coupling model and Bi-LSTM model; in the later stage, it forms a closed loop of basic construction-real-time prediction-emergency response by issuing early warnings and notifying relevant personnel to take emergency measures, thus realizing an essential leap from passive protection to active prevention and control. Attached Figure Description

[0026] Figure 1 This is a flowchart of the method of the present invention; Figure 2 This is a system structure diagram of the present invention. Detailed Implementation

[0027] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0028] Example 1 This embodiment discloses an artificial intelligence-based method for predicting cable fires in nuclear power plants. The method is as follows: Figure 1 As shown, it includes steps S1-S4.

[0029] The specific steps are described below: Step S1: Collect cable operating status data and environmental parameters in real time through sensors, and extract key features based on the cable operating status data.

[0030] Sensors placed around the cable include temperature sensors, current sensors, and vibration sensors. The real-time data collected on the cable's operating status includes temperature, current, and vibration frequency.

[0031] After collecting real-time data on cable operation status and environmental parameters through sensors, the collected data is preprocessed, including cleaning, noise reduction, and normalization.

[0032] Key features extracted from cable operating status data include temperature change rate, current fluctuation, and vibration anomalies.

[0033] Step S2: Based on environmental parameters and the pre-built Fire Structures Coupled Interface (FSCI) model, obtain the spatiotemporal distribution of the temperature field inside the structural system.

[0034] The fire-structure coupling model includes a fire-side mathematical model that outputs the temperature field and heat flow field in the fire environment, and a structure-side mathematical model that outputs the evolution of the temperature field inside the structure. The fire-side mathematical model and the structure-side mathematical model are connected through a coupling interface and can communicate with each other.

[0035] The pre-construction process of the fire-structure coupling model includes: Based on nuclear power plant building data and cable laying geometry data, fluid region models and solid region models are constructed respectively. The fluid region model is discretized using computational fluid dynamics (CFD) meshes, and the solid region model is discretized using finite element method (FEM) meshes. A mathematical model of the fire side is established based on the discretized fluid region model and the conservation equations for mass, momentum, energy, and species transport. Based on the discretized solid region model, a three-dimensional unsteady heat conduction equation is used to construct a structural mathematical model. Establish point-to-point data mapping relationships between the discretized fluid region model and the discretized solid region model, consider convection and radiation heat transfer modes, obtain coupling interface conditions, and establish coupling interfaces. A fire-structure coupling model is obtained by connecting the fire-side mathematical model and the structure-side mathematical model through a coupling interface.

[0036] The core energy equation expression of the fire detection mathematical model is: in, For gas density, h For enthalpy, u It is a velocity vector. k Thermal conductivity, T For temperature, This represents the heat source of combustion. The model outputs the temperature field in the fire environment. and thermal flow field.

[0037] The governing equations of the structural mathematical model are three-dimensional unsteady-state heat conduction equations, which describe the temperature field inside the structure. evolution: in, , and These are the density, specific heat capacity, and thermal conductivity of the structural material, respectively.

[0038] The expression for the coupling interface condition is: in, The net heat flux received by the structural surface. The convective heat transfer coefficient, This is the Stefan-Boltzmann constant. The surface emissivity of the structure.

[0039] The process of obtaining the spatiotemporal distribution of the internal temperature field of a structural system based on environmental parameters and a pre-built fire-structure coupling model includes: The environmental parameters and key features are compared with preset fluctuation thresholds. If any parameter or feature is greater than the preset fluctuation threshold, the state is determined to be abnormal. Based on the abnormal environmental parameters or key features, the preset simulation data of the corresponding fire condition is called and input into the fire-structure coupling model to obtain the spatiotemporal distribution of the temperature field inside the structural system. Otherwise, the state is determined to be normal. Based on the environmental parameters and key features, the preset simulation data of the normal condition is called and input into the fire-structure coupling model to obtain the spatiotemporal distribution of the temperature field inside the structural system.

[0040] Specifically, the process of obtaining the spatiotemporal distribution of the internal temperature field of the structural system within the fire-structure coupling model is as follows: the fluid region temperature obtained from fire simulation calculations is obtained through the FSCI interface of the fire-structure coupling model FSCI. and convective heat transfer coefficient As a spatiotemporally varying load, it is mapped from the CFD mesh nodes to the corresponding boundary nodes of the FEM mesh; Apply boundary conditions to the above mapping. and Substituting the coupling interface condition formula, the net heat flux acting on the structure surface is calculated. And apply it as a non-uniform, transient boundary condition to the structural side model; After applying loads and boundary conditions from the fire side, the three-dimensional unsteady heat conduction equations on the structural side are solved, and the temperature values ​​at any location and at any time within the structural system are calculated throughout the entire analysis period. This refers to the spatiotemporal distribution of the temperature field within the structural system.

[0041] The pre-setting process for simulation data corresponding to fire conditions includes: Based on the functional layout of the target nuclear power plant, cable laying routes and safety specifications, safety objectives and performance indicators are determined, and preliminary design schemes are obtained by referring to historical fire cases. Based on the actual conditions of the buildings within the nuclear power plant, potential fire hazards are identified, potential fire scenarios are obtained, and fire conditions corresponding to different fire scenarios are obtained according to the preliminary design plan. Fire conditions include different fire power, ventilation conditions, and fuel types. Typical potential fire scenarios include cable short circuit fire scenarios, external fire source igniting cable scenarios, and fire scenarios in the confined space of cable trenches. Based on fire conditions, the smoke trajectory, spatial temperature field, and convective heat transfer coefficient under different fire conditions are simulated using a fire dynamics model and output as simulation data for the corresponding fire conditions.

[0042] The fire dynamics model is a simulation model generated by FDS software and built based on real-world data.

[0043] Step S3: Overlay the spatiotemporal distribution of the internal temperature field of the structural system with the pre-constructed digital model of cable laying to obtain a dynamic cable thermal risk map, and calculate the real-time failure probability of the cable based on the material tolerance temperature and thermal aging model.

[0044] The digital model of cable laying is constructed based on the geometric drawings of the nuclear power plant building and cable laying. The digital model of cable laying includes the physical direction of the cable, laying location, density, joint distribution and its relative position with the surrounding structure.

[0045] The dynamic cable thermal risk map is updated synchronously and dynamically based on the sensor data acquisition frequency and the spatiotemporal distribution update frequency of the internal temperature field of the structural system.

[0046] The process of calculating the real-time failure probability of cables based on material resistance temperature and thermal aging models includes: To obtain key properties of cable materials, including material temperature resistance, rated aging life at standard temperature, and thermal aging activation energy; Based on a thermal aging model, the accelerating effect of temperature on cable material aging is quantified, and the aging acceleration coefficient at the current temperature is obtained based on the aging rate at a standard temperature. In this embodiment, the industry-standard Arrhenius thermal aging model is used to quantify the accelerating effect of temperature on cable material aging, expressed as: in, For frequency factors, The gas constant is... For the real-time temperature of the cable, The aging rate at standard temperature. It is a thermal aging activation energy. The aging acceleration factor of the cable at the current temperature; Based on the cable's service life and cumulative aging degree, the cumulative aging amount of the cable is calculated, and the expression is as follows: in, This represents the cumulative aging amount of the cable. The cable has been in operation for a certain period of time. The cable's historical temperature variation function is calculated back from historical sensor monitoring data and the FSCI model. The remaining life of the cable in real time is calculated by combining the rated aging life of the material and the aging acceleration factor of the cable at the current temperature. The expression is as follows: in, The thermal aging rate at standard temperature. The rated cumulative aging limit of the material is obtained based on the material's rated aging life. Real-time cable remaining life; Based on the real-time remaining cable life and key characteristics, the expression for calculating the real-time failure probability of the cable is as follows: in, This represents the real-time failure probability of the cable. These are correction coefficients related to key anomaly features, and their values ​​are obtained through training on historical data. The rated aging life of the material.

[0047] Step S4: Based on key features, dynamic cable thermal risk map and real-time cable failure probability, a pre-trained Bi-LSTM model is used to predict the probability of cable fire occurrence in real time. If the probability of occurrence is greater than a preset threshold, an early warning is issued and relevant personnel are notified to take emergency measures.

[0048] In this embodiment, the early warning is specifically divided into: Low risk, probability 10%-30%: The system pushes a prompt message to the operation and maintenance platform, marking the location of high-risk cable sections based on the cable thermal risk map, and recommends that operation and maintenance personnel focus on inspection.

[0049] Medium risk, probability 30%-60%: Activate audible and visual warnings, notify the maintenance supervisor via system pop-ups and SMS, and simultaneously display real-time cable data curves, thermal risk maps, and failure probability trends to guide maintenance personnel in on-site troubleshooting.

[0050] High risk, probability ≥60%: Immediately activate the highest level of early warning, link the emergency command system, notify the emergency team to arrive, and automatically push the preset emergency plan.

[0051] The Bi-LSTM model is trained based on historical data. Specifically, the Bi-LSTM model is a deep learning model. In this embodiment, it is a bidirectional long short-term memory recurrent neural network used to learn the complex nonlinear relationship between cable status and fire risk.

[0052] Maintenance personnel record the verification results after the warning, and the emergency response team records data such as whether a fire has occurred, the time difference between the warning and the actual occurrence, and the rationality of the allocation of firefighting resources, and feeds them back to the system database.

[0053] The Bi-LSTM model is periodically retrained using newly acquired real-time data, early warning feedback data, and newly emerging hazard data to optimize model parameters, improve adaptability to new operating conditions, such as parameter changes caused by equipment aging and the impact of extreme weather, and reduce false alarm and false negative rates.

[0054] When updating the Bi-LSTM model, newly identified fire hazards and new fire cases are also incorporated to supplement the design fire conditions, update the simulation data of the preset fire conditions in the FSCI model, expand the coverage of temperature field simulation, and ensure that the model can accurately reflect the thermal response of cables under various complex scenarios.

[0055] Example 2 This embodiment, based on Embodiment 1 above, discloses an artificial intelligence-based nuclear power plant cable fire prediction system. The system is as follows: Figure 2 As shown, it includes: Data acquisition module: Collects cable operating status data and environmental parameters in real time through sensors; Data processing module: preprocesses sensor-acquired data and extracts key features based on cable operating status data; preprocessing includes cleaning, noise reduction, and normalization. Model training module: pre-built fire-structure coupling model, pre-trained Bi-LSTM model; Fire prediction module: Based on environmental parameters and a pre-built fire-structure coupling model, the spatiotemporal distribution of the internal temperature field of the structural system is obtained; the spatiotemporal distribution of the internal temperature field of the structural system is superimposed with a pre-built digital model of cable laying to obtain a dynamic cable thermal risk map, and the real-time failure probability of the cable is calculated based on the material tolerance temperature and thermal aging model; based on key features, the dynamic cable thermal risk map and the real-time failure probability of the cable, a pre-trained Bi-LSTM model is used to predict the probability of cable fire in real time. Results output module: Outputs the probability of cable fire occurrence. If the probability of occurrence is greater than a preset threshold, an early warning will be issued and relevant personnel will be notified to take emergency measures.

[0056] For details regarding the above modules, please refer to the relevant descriptions and effects in Example 1 for further understanding.

[0057] Example 3 Based on Embodiment 1, this embodiment provides an electronic device, including: one or more processors and a memory, wherein the memory stores one or more programs, the one or more programs including instructions for executing the aforementioned artificial intelligence-based nuclear power plant cable fire prediction method.

[0058] At the hardware level, the electronic device includes a processor, internal bus, network interface, memory, and non-volatile memory, and may also include other hardware required for business operations. The processor reads the corresponding computer program from the non-volatile memory into memory and then runs it to implement the aforementioned artificial intelligence-based nuclear power plant cable fire prediction method. Of course, in addition to software implementation, this invention does not exclude other implementation methods, such as logic devices or a combination of hardware and software, etc. That is to say, the execution subject of the following processing flow is not limited to individual logic units, but can also be hardware or logic devices.

[0059] Memory may include non-persistent storage in computer-readable media, such as random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM. Memory is an example of computer-readable media.

[0060] Computer-readable media include both permanent and non-permanent, removable and non-removable media that can store information using any method or technology. Information can be computer-readable instructions, data structures, modules of programs, or other data. Examples of computer storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, disk storage or other magnetic storage devices, or any other non-transferable medium that can be used to store information accessible by a computing device. As defined herein, computer-readable media does not include transient computer-readable media, such as modulated data signals and carrier waves.

[0061] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any person skilled in the art can easily conceive of various equivalent modifications or substitutions within the technical scope disclosed in the present invention, and these modifications or substitutions should all be covered within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A method for predicting cable fires in nuclear power plants based on artificial intelligence, characterized in that, The method includes: The cable operating status data and environmental parameters are collected in real time by sensors, and key features are extracted based on the cable operating status data. Based on the environmental parameters and the pre-constructed fire-structure coupling model, the spatiotemporal distribution of the temperature field inside the structural system is obtained; The spatiotemporal distribution of the internal temperature field of the structure system is superimposed with a pre-constructed digital model of cable laying to obtain a dynamic cable thermal risk map, and the real-time failure probability of the cable is calculated based on the material tolerance temperature and thermal aging model. Based on the aforementioned key features, dynamic cable thermal risk map, and real-time cable failure probability, a pre-trained Bi-LSTM model is used to predict the probability of cable fires in real time. If the probability of occurrence exceeds a preset threshold, an early warning is issued and relevant personnel are notified to take emergency measures.

2. The method for predicting nuclear power plant cable fires based on artificial intelligence according to claim 1, characterized in that, After collecting real-time data on the cable's operating status and environmental parameters through sensors, the collected data is preprocessed, including cleaning, noise reduction, and normalization. The key features extracted from cable operating status data include temperature change rate, current fluctuation, and vibration anomaly.

3. The method for predicting nuclear power plant cable fires based on artificial intelligence according to claim 1, characterized in that, The fire-structure coupling model includes a fire-side mathematical model that outputs the temperature field and heat flow field in the fire environment and a structure-side mathematical model that outputs the temperature field evolution inside the structure. The fire-side mathematical model and the structure-side mathematical model are connected through a coupling interface and achieve data exchange.

4. The method for predicting nuclear power plant cable fires based on artificial intelligence according to claim 3, characterized in that, The pre-construction process of the fire-structure coupling model includes: Based on nuclear power plant building data and cable laying geometry data, fluid region models and solid region models are constructed respectively. The fluid region model is discretized using a computational fluid dynamics mesh, and the solid region model is discretized using a finite element mesh. A mathematical model of the fire side is established based on the discretized fluid region model and the conservation equations for mass, momentum, energy, and species transport. Based on the discretized solid region model, a three-dimensional unsteady heat conduction equation is used to construct a structural mathematical model. Establish point-to-point data mapping relationships between the discretized fluid region model and the discretized solid region model, consider convection and radiation heat transfer modes, obtain coupling interface conditions, and establish coupling interfaces. Based on the coupling interface, the fire-side mathematical model and the structural mathematical model are connected to obtain the fire-structure coupling model.

5. The method for predicting nuclear power plant cable fires based on artificial intelligence according to claim 1, characterized in that, Based on the aforementioned environmental parameters and the pre-constructed fire-structure coupling model, the process of obtaining the spatiotemporal distribution of the internal temperature field of the structural system includes: The environmental parameters and key features are compared with preset fluctuation thresholds. If any parameter or feature is greater than the preset fluctuation threshold, the state is determined to be abnormal. Based on the abnormal environmental parameters or key features, preset simulation data of the corresponding fire condition is called and input into the fire-structure coupling model to obtain the spatiotemporal distribution of the temperature field inside the structural system. Otherwise, the state is determined to be normal. Based on the environmental parameters and key features, preset simulation data of the normal condition is called and input into the fire-structure coupling model to obtain the spatiotemporal distribution of the temperature field inside the structural system.

6. The method for predicting nuclear power plant cable fires based on artificial intelligence according to claim 5, characterized in that, The pre-setting process for the simulation data corresponding to fire conditions includes: Based on the functional layout of the target nuclear power plant, cable laying routes and safety specifications, safety objectives and performance indicators are determined, and preliminary design schemes are obtained by referring to historical fire cases. Based on the actual conditions of the buildings within the nuclear power plant, potential fire hazards are identified, potential fire scenarios are obtained, and fire conditions corresponding to different fire scenarios are designed according to the preliminary scheme. The fire conditions include different fire power, ventilation conditions, and fuel types. Based on the aforementioned fire conditions, the smoke trajectory, spatial temperature field, and convective heat transfer coefficient under different fire conditions are simulated using a fire dynamics model and output as simulation data for the corresponding fire conditions.

7. The method for predicting nuclear power plant cable fires based on artificial intelligence according to claim 1, characterized in that, The digital model of cable laying is constructed based on the geometric drawings of the nuclear power plant building and cable laying. The digital model of cable laying includes the physical direction of the cable, laying location, density, joint distribution and its relative position with the surrounding structure.

8. The method for predicting nuclear power plant cable fires based on artificial intelligence according to claim 1, characterized in that, The process of calculating the real-time failure probability of the cable based on the material's temperature tolerance and thermal aging model includes: To obtain key properties of cable materials, including material temperature resistance, rated aging life at standard temperature, and thermal aging activation energy; Based on the thermal aging model, the accelerating effect of temperature on cable material aging is quantified, and the aging acceleration coefficient of the cable at the current temperature is obtained based on the aging rate at the standard temperature. Based on the cable's operating time and cumulative aging degree, the cumulative aging amount of the cable is calculated, and combined with the material's rated aging life and the cable's aging acceleration coefficient at the current temperature, the real-time remaining cable life is calculated. Based on the remaining lifespan and key characteristics of the cable in real time, the probability of real-time cable failure is calculated.

9. The method for predicting nuclear power plant cable fires based on artificial intelligence according to claim 1, characterized in that, The Bi-LSTM model is periodically retrained using newly added real-time collected data and early warning feedback data to optimize model parameters.

10. A nuclear power plant cable fire prediction system based on artificial intelligence, characterized in that, The system includes: Data acquisition module: Collects cable operating status data and environmental parameters in real time through sensors; Data processing module: preprocesses sensor-collected data and extracts key features based on the cable's operating status data; Model training module: pre-built fire-structure coupling model, pre-trained Bi-LSTM model; Fire prediction module: Based on the environmental parameters and the pre-built fire-structure coupling model, the spatiotemporal distribution of the internal temperature field of the structural system is obtained; the spatiotemporal distribution of the internal temperature field of the structural system is superimposed with the pre-built digital model of cable laying to obtain a dynamic cable thermal risk map, and the real-time failure probability of the cable is calculated based on the material tolerance temperature and thermal aging model; based on the key features, the dynamic cable thermal risk map and the real-time failure probability of the cable, the probability of cable fire occurrence is predicted in real time using a pre-trained Bi-LSTM model. Result output module: Outputs the probability of cable fire occurrence. If the probability of occurrence is greater than a preset threshold, an early warning is issued and relevant personnel are notified to take emergency measures.

Images