Urban underground cable tunnel network fire hazard temperature prediction method and related device

Abstract

The invention belongs to the field of power grid safety prediction, and provides an urban underground cable tunnel network fire hazard temperature prediction method and a related device, and the method comprises the steps: collecting the environment data in a cable tunnel and the material data of a cable; establishing a three-dimensional finite element model comprising an electromagnetic field, a thermal field and a fluid field based on the environmental data and the material data, simulating temperature responses of the cable under different conditions, and obtaining sample data of cable core temperature; the method comprises the following steps: training a convolutional neural network (CNN) to extract features based on environment data and material data, processing time sequence data through a long short-term memory (LSTM) network, and constructing a CNN-LSTM hybrid model; and taking the obtained sample data of the cable core temperature as the input of the CNN-LSTM hybrid model to obtain a predicted value of the cable core temperature. The method not only improves the accuracy and real-time performance of prediction, but also enhances the adaptability and robustness of the model to a complex environment, and has wide application prospects and important social values.

Description

Technical Field

[0001] The present invention relates to the field of power grid safety prediction, and in particular to a method and related device for predicting fire temperature of an urban underground cable tunnel network. Background Art

[0002] With the rapid development of my country's economy, urban infrastructure construction has received increasing attention. In the field of power transmission, traditional overhead urban transmission lines are difficult to maintain and damage the city's image. Therefore, more and more cities are beginning to use underground cable tunnels as a solution for transmission lines. Cable tunnels are not only convenient for the expansion and maintenance of transmission lines, but also do not damage the ground landscape, effectively alleviating the pressure of shortage of ground space.

[0003] However, as an underground enclosed space, the fire problem of urban power cable tunnels cannot be ignored. Cable tunnels are characterized by closed space, a large number of potential combustibles, and difficulty in fire fighting. Once a fire occurs, it will cause serious consequences and seriously threaten the safe operation of the urban power grid. According to statistics, about 30% of cable tunnel fire accidents are caused by cable failures, and the remaining 70% of accidents are caused by external factors, such as welding sparks caused by construction, external fire spread and ignition, rodent damage, etc. Cable tunnel fire accidents have the characteristics of hidden fire points, difficult gas and heat diffusion, fast fire spread, narrow space, difficulty in fire fighting, serious losses, and difficult recovery.

[0004] Therefore, studying the causes and characteristics of cable tunnel fire accidents and analyzing the applicability of cable tunnel fire extinguishing systems are of great significance for preventing cable tunnel fires and reducing the losses and impacts caused by cable tunnel fire accidents. From the perspective of fire risk assessment, studying the fire risk and prevention and control issues of urban power cable tunnels can identify the key factors affecting the fire risk of cable tunnels, such as cable combustion properties, cable laying conditions, fire protection system configuration, and power failures. In addition, with the application of intelligent mobile inspection and fire extinguishing devices in tunnels, higher requirements are put forward for early warning and effective prevention and control of cable tunnel fires.

[0005] The existing technology mainly adopts analytical calculation method, numerical calculation method and experimental method. The analytical calculation method is mainly based on IEC standards and NM theory, which is suitable for simple laying environments and has the advantages of being fast and direct, but for complex laying conditions, the calculation results deviate from the actual situation; the numerical calculation method, which mostly uses the finite element method, can simulate the actual laying of the cable and is suitable for current carrying capacity calculation under complex conditions, but there are problems of model simplification and deviation from the actual situation. The experimental method is not universal due to its high cost. In addition, there is a real-time prediction system based on support vector machines. Although it can provide accurate and timely fire temperature prediction, the prediction error is large near the fire source; methods based on multi-physics field coupling models, such as electromagnetic-thermal-fluid coupling, can provide more comprehensive analysis, but the two-dimensional model limits its accuracy in simulating the actual operating state. Therefore, the existing technology has defects in predicting the fire temperature of urban underground cables, such as insufficient prediction accuracy, deviation from the actual situation of model simplification, and limitations of multi-physics field coupling models.

[0006] In summary, this study aims to propose a fire temperature prediction method for urban underground cable tunnel networks. Through in-depth analysis of cable tunnel fire risks and research on prediction technology, the accuracy of cable tunnel fire warning can be improved and the impact of fire accidents on urban power grids and people's lives can be reduced. Summary of the invention

[0007] The purpose of the present invention is to provide a method and related device for predicting fire temperature in an urban underground cable tunnel network to solve the above-mentioned problem.

[0008] To achieve the above object, the present invention adopts the following technical solutions: In a first aspect, the present invention provides a method for predicting fire temperature in an urban underground cable tunnel network, comprising: Collect environmental data inside the cable tunnel and material data of the cables; Based on environmental data and material data, a three-dimensional finite element model including electromagnetic field, thermal field and fluid field is established to simulate the temperature response of the cable under different conditions and obtain sample data of cable core temperature; Based on environmental data and material data, the convolutional neural network (CNN) is trained to extract features, and then the long short-term memory (LSTM) network is used to process time series data to build a CNN-LSTM hybrid model. The obtained sample data of cable core temperature is used as the input of the CNN-LSTM hybrid model to obtain the predicted value of the cable core temperature.

[0009] Furthermore, the collection of environmental data in the cable tunnel and material data of the cable includes: Ambient temperature, wind speed, humidity, and load current data in the cable tunnel; Collect the thermal conductivity, specific heat capacity and density parameters of the cable core, insulation layer and buffer layer materials.

[0010] Furthermore, the three-dimensional finite element model including electromagnetic field, thermal field and fluid field is established based on the environmental data and material data, including: Using COMSOL multi-physics simulation software, a three-dimensional finite element model including electromagnetic field, thermal field and fluid field is established. According to the actual size and material properties of the cable tunnel, the geometric model of the tunnel and cable is constructed. Specifically: Heat conduction: According to Fourier's law, the heat flux density q is proportional to the rate of temperature change in that direction, and its function is shown as follows:

[0011] Where, q is the heat flux density in W / m 2 ; Ф is the heat flow through area A; λ is the thermal conductivity, the unit is W / (m·K); For thermal energy balance within the cable:

[0012] in, For the inflow of heat, To remove heat, The Joule heat generated by the cable core; Heat convection: The function is as follows:

[0013] in, It represents the temperature difference between the cable outer skin and the air in the tunnel; h is the heat transfer coefficient of the cable outer sheath surface, the unit is w / (m 2 *k); The calculation of the heat generated by the heat source and the increase in internal energy is shown as follows:

[0014] Among them, is the air density; is the heat capacity at constant pressure; Thermal radiation: The radiation energy that can be absorbed per unit time is shown in the following formula:

[0015] Where A is the area of ​​the object; T is the temperature of the object; When the surface temperature of the cable in the tunnel is 2T and the air temperature is 1T, the radiation heat between the cable and the air is as follows:

[0016] is the surface area of ​​the object; is the emissivity of air and does not exceed 1; The differential form of Maxwell's equations for the electromagnetic field in cables is shown below:

[0017] in, and are source variables, which are eddy current density and current density; The relationship between the parameters of Maxwell's equations and the relevant field quantities is expressed as follows: 𝐵 Among them, 𝜀 is the dielectric constant; 𝜇 is the magnetic permeability; 𝜎 is the electrical conductivity; according to Maxwell's equations, the vector magnetic potential equations of each layer of the cable are derived as shown below:

[0018] Among them, 𝐽 𝑠 is the current density of the cable core; e is the eddy current density of the metal sheath; A1-A4 represent the vector magnetic potential of the cable core, insulation layer, metal sheath layer and outer sheath layer respectively; The calculation formula for the loss of each layer of the cable is shown as follows:

[0019] In the formula, is the power loss per unit length of each layer; is the electrical conductivity of each material.

[0020] Furthermore, the simulation of the temperature response of the cable under different conditions to obtain sample data of the cable core temperature includes: Set its load current, initial temperature and boundary conditions, perform mesh generation, calculate the resistivity and dielectric loss angle parameters at the current temperature, calculate the dielectric loss angle and resistivity parameters at the current temperature; solve the temperature field and fluid field equations, calculate the unit temperature and wind speed, call the multi-physics field coupling module, calculate the temperature distribution and fluid distribution, and finally determine whether the temperature difference meets the requirements of the output results.

[0021] Furthermore, based on the environmental data and material data, a convolutional neural network (CNN) is trained to extract features, and then the time series data is processed by a long short-term memory (LSTM) network to construct a CNN-LSTM hybrid model, including: Use CNN to extract key features of temperature change trends and load current peaks from historical data, and design convolutional layers and pooling layers to extract temporal and spatial features; Use LSTM to process time series data, capture the long-term dependency of cable temperature changes, design the LSTM network structure, and determine the number of hidden layers and neurons; Use the collected data to train the CNN-LSTM model and evaluate the model performance through cross-validation method; adjust the model parameters of learning rate and batch size to optimize the model performance.

[0022] Furthermore, the obtained sample data of the cable core temperature is used as the input of the CNN-LSTM hybrid model to obtain the predicted value of the cable core temperature, including: Establish a COMSOL finite element transient simulation model for cables: determine its electromagnetic-thermal-fluid multi-physics field coupling, set cable and tunnel structural parameters, laying conditions, load current, boundary conditions and solver; Get sample data of cable core temperature: According to the cable transient model, input its ambient temperature, wind speed and load current data conditions, and obtain the temperature data of the cable core through transient research as the sample data of CNN-LSTM; Dataset division and preprocessing of sample data: The first 70% of the cable temperature sample data is used as the training set of the CNN-LSTM model, and the last 30% of the data is used as the validation set of the CNN-LSTM model. The original sample data of the cable core temperature data set is normalized, and X={X1(1),…,X1(n),…Xk(1),…,Xk(n)} is selected as the input variable of the prediction model, and Y={y(n),y(n+1),…,y(n+m)} is selected as the output variable; In this model, k is the number of measurable parameter types; n is the time series; m is the mth moment of the predicted cable core temperature; k, n, and m are natural numbers; Establish a prediction model: Select ReLU as the activation function, use the Adam function for optimization training with a learning rate of 0.001, use the mean absolute error MAE and determination coefficient R2 as the model performance evaluation indicators, and establish a CNN-LSTM cable core temperature prediction model; Through the operation of its CNN-LSTM hybrid model, the desired cable core temperature prediction value is obtained in the output layer.

[0023] Furthermore, an attention mechanism is introduced into the CNN-LSTM model. The correlation between time steps is calculated through a learning function, different weights are assigned, and the weights are normalized using the softmax function to obtain the attention weight of each time step.

[0024] In a second aspect, the present invention provides a fire temperature prediction system for an urban underground cable tunnel network, comprising: Data acquisition module, used to collect environmental data in the cable tunnel and material data of the cable; The finite element model building module is used to build a three-dimensional finite element model including electromagnetic field, thermal field and fluid field based on environmental data and material data, simulate the temperature response of the cable under different conditions, and obtain sample data of the cable core temperature; The hybrid model building module is used to train the convolutional neural network (CNN) to extract features based on environmental data and material data, and then process the time series data through the long short-term memory network (LSTM) to build a CNN-LSTM hybrid model; The prediction output module is used to use the sample data of the cable core temperature obtained as the input of the CNN-LSTM hybrid model to obtain the predicted value of the cable core temperature.

[0025] In a third aspect, the present invention provides a computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the urban underground cable tunnel network fire temperature prediction method when executing the computer program.

[0026] In a fourth aspect, the present invention provides a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the steps of the urban underground cable tunnel network fire temperature prediction method.

[0027] Compared with the prior art, the present invention has the following technical effects: The fire temperature prediction method for urban underground cable tunnel network proposed in this paper integrates advanced technologies such as finite element simulation, deep learning and attention mechanism, aiming to achieve accurate prediction of cable tunnel fire temperature; By establishing a three-dimensional finite element model including electromagnetic field, thermal field and fluid field, the present invention can fully consider the complex physical environment in the cable tunnel, such as electromagnetic effect, heat conduction, convection and radiation, so as to more accurately simulate the temperature response of the cable under different conditions. This multi-dimensional and refined simulation provides a solid foundation for subsequent temperature prediction.

[0028] The CNN-LSTM hybrid model, which combines convolutional neural networks (CNN) and long short-term memory networks (LSTM), can effectively extract key features from historical data and capture the long-term dependencies of cable temperature changes. This deep learning method significantly improves the accuracy of temperature prediction, especially when dealing with nonlinear and non-stationary time series data.

[0029] The present invention uses a finite element model to quickly obtain sample data of cable core temperature as the input of the CNN-LSTM model, achieving a rapid response from data collection to temperature prediction. This real-time performance is crucial for fire prevention and emergency response, and can promptly detect potential safety hazards and provide support for decision-making.

[0030] By optimizing the structure and parameters of the CNN-LSTM model, such as adjusting the learning rate and batch size, the training efficiency and prediction speed of the model are further improved, making this method more efficient and feasible in practical applications.

[0031] The present invention introduces an attention mechanism into the CNN-LSTM model, which enables the model to automatically learn and focus on the most critical time steps and features for temperature prediction, thereby assigning different weights. This mechanism enhances the model's ability to capture important information and improves the accuracy and robustness of prediction.

[0032] The introduction of the attention mechanism also makes the model more flexible and efficient when processing long sequence data, avoiding problems such as information loss and gradient vanishing, and further improving the performance of the model.

[0033] By accurately predicting the fire temperature in cable tunnels, the present invention provides important data support for the safety management of power systems. When an abnormal temperature rise is predicted, timely intervention measures can be taken, such as increasing ventilation, reducing load, etc., to effectively prevent the occurrence of fires.

[0034] When a fire occurs, accurate temperature prediction can also help to quickly initiate and effectively implement emergency response, such as guiding rescue personnel to locate the source of the fire and assess the intensity of the fire, providing strong technical support for rescue work.

[0035] In summary, the fire temperature prediction method for urban underground cable tunnel networks proposed in this invention achieves high-precision and real-time prediction of cable tunnel fire temperature by combining advanced technologies such as finite element method, deep learning and attention mechanism, providing scientific basis and technical support for the safety management and fire prevention of power systems. This method not only improves the accuracy and real-time performance of the prediction, but also enhances the adaptability and robustness of the model to complex environments, and has broad application prospects and important social value. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] Figure 1 It is a flow chart of the present invention.

[0037] Figure 2 CNN-LSTM model structure framework.

[0038] Figure 3 Diagram of the input and output structure of the prediction model.

[0039] Figure 4 Attention mechanism structure. DETAILED DESCRIPTION

[0040] The present invention is further described below in conjunction with the accompanying drawings: Example 1, please refer to Figure 1 The present invention provides a method for predicting fire temperature of an urban underground cable tunnel network, comprising: Collect environmental data inside the cable tunnel and material data of the cables; Based on environmental data and material data, a three-dimensional finite element model including electromagnetic field, thermal field and fluid field is established to simulate the temperature response of the cable under different conditions and obtain sample data of cable core temperature; Based on environmental data and material data, the convolutional neural network (CNN) is trained to extract features, and then the long short-term memory (LSTM) network is used to process time series data to build a CNN-LSTM hybrid model. The obtained sample data of cable core temperature is used as the input of the CNN-LSTM hybrid model to obtain the predicted value of the cable core temperature.

[0041] By collecting environmental data in the cable tunnel and material data of the cable, the present invention can comprehensively consider various factors that affect the cable temperature and provide a rich information basis for temperature prediction.

[0042] Establishing a three-dimensional finite element model that includes electromagnetic fields, thermal fields, and fluid fields can accurately simulate the temperature response of the cable under different conditions, obtain accurate cable core temperature sample data, and provide a high-quality training set for subsequent deep learning predictions.

[0043] The CNN-LSTM hybrid model constructed by combining the convolutional neural network (CNN) and the long short-term memory network (LSTM) can fully utilize the spatial feature extraction capability of CNN and the time series processing capability of LSTM to improve the accuracy of temperature prediction.

[0044] The present invention quickly obtains sample data of cable core temperature through a finite element model and uses a deep learning model to perform rapid prediction, thereby achieving efficient processing from data collection to temperature prediction and enhancing the real-time performance of the prediction.

[0045] Once the CNN-LSTM hybrid model is trained, it can quickly predict new input data to meet the needs of real-time monitoring and early warning.

[0046] By training the convolutional neural network (CNN) to extract features and then processing the time series data through the long short-term memory (LSTM) network, the CNN-LSTM hybrid model constructed by the present invention can learn the complex laws of cable temperature changes and improve the generalization ability of the model.

[0047] This hybrid model can adapt to different cable tunnel environments, load currents and boundary conditions, improving the applicability and reliability of the prediction method.

[0048] Accurate cable temperature prediction is crucial for fire prevention and emergency response. The prediction method provided by the present invention can timely detect abnormal increase in cable temperature and provide early warning information for fire prevention.

[0049] When a fire occurs, accurate temperature predictions can also help assess the fire, develop fire-fighting plans and guide rescue work, providing scientific decision-making support for emergency response.

[0050] In summary, the fire temperature prediction method for urban underground cable tunnel network provided by the present invention realizes accurate and real-time prediction of cable temperature by combining finite element simulation and deep learning technology, improves the accuracy and real-time performance of prediction, enhances the generalization ability of the model, and provides important decision support for fire prevention and emergency response. This method is of great significance for ensuring the safe operation of urban underground cable tunnels.

[0051] Embodiment 2, the present invention provides a method for predicting fire temperature of an urban underground cable tunnel network, specifically comprising: 1. Data Collection Collect environmental parameters of underground cable tunnels, including but not limited to ambient temperature, wind speed, load current, material parameters (cable core, insulation layer, buffer layer, etc.).

[0052] (II) Establishing finite element model Using COMSOL multi-physics simulation software, a three-dimensional finite element model including electromagnetic field, thermal field and fluid field is established: For the inside of a single-core power cable, the resistance loss in the cable core is the main heat source for the temperature difference between the cable and the external environment. Therefore, there is a heat conduction method to transfer heat energy from the cable core layer to the outer sheath layer. There is also a heat radiation transfer method between the outer sheath surface and the external environment, which dissipates heat energy to the external environment. This study assumes that there is air flow in the tunnel, generating forced convection for heat transfer and heat exchange.

[0053] Heat conduction: Heat conduction is a heat transfer phenomenon in the presence of temperature difference, where heat is transferred from high temperature to low temperature through the medium. According to Fourier's law, the heat flux density q is proportional to the rate of temperature change in that direction, and its function is shown in the following formula:

[0054] Where, q is the heat flux density in W / m 2; Ф is the heat flow through area A; λ is the thermal conductivity, the unit is W / (m·K). This formula is also known as Fourier's law, which shows that the larger the λ, the better its thermal conductivity. Therefore, the thermal conductivity is solid>liquid>gas.

[0055] From the law of conservation of energy, we know that the heat energy in the cable flows out and in through the cable core, and is transferred to each layer of the cable in turn, reaching a certain balance with its own medium.

[0056]

[0057] in, For the inflow of heat, To remove heat, The Joule heat generated by the cable core.

[0058] Thermal convection: Thermal convection refers to the process of heat transfer between fluids or solids when the fluid moves in nature, because there is a temperature difference between the two. Since the convection in the tunnel is basically derived from the density difference between the earth and the air in the tunnel, the natural convection in the tunnel can be described by the conservation theorem in the rectangular coordinate system. Its conservation laws include: the law of conservation of mass, the law of conservation of kinetic energy, and the law of conservation of energy. Based on the above three laws, Newton proposed the basic equation of the law of convective heat transfer. Its function is shown as follows:

[0059] in, It represents the temperature difference between the cable outer skin and the air in the tunnel; h is the heat transfer coefficient of the cable outer sheath surface, the unit is w / (m 2 *k).

[0060] Based on the basic equation, the heat generated by the heat source and the increase in internal energy are calculated as follows:

[0061] Among them, is the air density; is the heat capacity at constant pressure.

[0062] Thermal radiation: Thermal radiation appears in the form of electromagnetic waves. The heat generated by the internal energy of an object above 0°C is transferred through electromagnetic waves. The greater the internal energy, the higher the temperature generated and the more heat is radiated. Compared with the two heat transfer methods of thermal conduction and thermal convection, thermal radiation does not require the help of a medium. Even in a vacuum, heat can be transferred in the form of electromagnetic waves. For example, the heat transfer of the sun is through its own internal energy converted into electromagnetic waves, which are transferred to various objects through a vacuum for absorption, thereby forming their own energy. Through the definition of an idealized radiator - a black body, it can be understood that its absorption ratio is always 1 for all wavelengths. Therefore, the radiation energy that an idealized black body can absorb per unit time is shown in the following formula:

[0063] Among them, A is the area of ​​the object; T is the temperature of the object. Therefore, this formula shows that the radiation capacity of an object is proportional to the product of the surface area and temperature of the object.

[0064] When the surface temperature of the cable in the tunnel is 2T and the air temperature is 1T, since the cable is wrapped by air, the radiation heat between the cable and the air is as follows:

[0065] is the surface area of ​​the object; is the emissivity of air and does not exceed 1.

[0066] When a single-core high-voltage power cable flows through a load AC current, the material of the metal part of the cable will generate eddy current loss, which is caused by the cable core and the metal sheath layer in the alternating magnetic field. From a macroscopic perspective, the electromagnetic phenomenon can be equivalent to the Maxwell equations. Its basic variables include: magnetic field intensity H, electric field intensity E, magnetic induction intensity B and electric displacement vector D. Its source variables include current density J and charge density p, so the differential form of the Maxwell equations of the cable electromagnetic field is shown as follows:

[0067] in, and are source variables, which are eddy current density and current density respectively.

[0068] To characterize macroscopic electromagnetic properties, the relationship between the parameters of Maxwell's equations and related field quantities can be expressed as follows: 𝐵 Among them, 𝜀 is the dielectric constant; 𝜇 is the magnetic permeability; 𝜎 is the electrical conductivity. According to Maxwell's equations, the vector magnetic potential equations of each layer of the cable are derived as shown below:

[0069] Among them, 𝐽𝑠 is the current density of the cable core; 𝐽e is the eddy current density of the metal sheath. A1-A4 represent the vector magnetic potential of the cable core, insulation layer, metal sheath layer and outer sheath layer respectively.

[0070] The calculation formula for the loss of each layer of the cable is shown as follows:

[0071] In the formula, is the power loss per unit length of each layer; is the electrical conductivity of each material.

[0072] The flow of air in the fluid field is generally divided into natural convection and forced convection. Natural convection depends on the distribution of temperature w in the air. When there is an uneven distribution, the cold flow will drop and the hot flow will rise, resulting in natural convection in the tunnel. However, ventilation devices are usually arranged in the tunnel to achieve forced convection to dissipate heat from the cables. Forced convection must first distinguish between laminar flow and turbulent flow. Laminar flow flows smoothly and slowly in a straight line in the tunnel, while turbulent flow changes its motion trajectory with different amplitudes as the flow velocity changes. The faster the flow velocity, the greater the swing amplitude and the more irregular the flow. Therefore, it is necessary to judge the size of the Reynolds number. Before analyzing the fluid field, it is necessary to determine whether the fluid in the tunnel is laminar or turbulent.

[0073]

[0074] In the formula, is the air velocity; is the tunnel diameter; is the air density, in kg / m 2 ; is the dynamic viscosity, the unit is Pas. When the Reynolds number is less than 2300, it is generally considered that the fluid in the tunnel is laminar flow, and when the Reynolds number is greater than 4000, it is generally considered that the fluid in the tunnel is mainly turbulent flow.

[0075] Simulate the temperature distribution of cables in normal operation and under fire conditions: According to the basic principles of temperature field, electromagnetic field, and fluid field, the equations are described, and a three-dimensional tunnel cable cluster geometry model is established. According to the differential equations, the corresponding physical fields are selected, such as turbulence in the fluid field, etc., the solution area is unitized, and the finite element calculation equations of each physical field are established. The load current, initial temperature, and boundary conditions are set. Meshing is performed to ensure the accuracy of the finite element calculation, and the resistivity, dielectric loss angle and other parameters at the current temperature are calculated. Calculate the dielectric loss angle and resistivity and other parameters at the current temperature. Solve the temperature field and fluid field equations, and calculate the unit temperature and wind speed. Call the multi-physics field coupling module to calculate the temperature distribution and fluid distribution, and finally determine whether the temperature difference meets the requirements of the output results.

[0076] (III) Building a CNN-LSTM model Based on the collected data, a convolutional neural network (CNN) is trained to extract features, and then the time series data is processed through a long short-term memory network (LSTM) to construct a CNN-LSTM hybrid model, such as Figure 2 shown.

[0077] Since LSTM can solve problems related to time series, but because it cannot consider spatial position and is a short time series, its accuracy is not high, and there will be problems such as gradient disappearance, which makes it impossible to predict the cable temperature; and it is difficult for CNN to add time series. For this reason, CNN is combined with LSTM, and the advantages of CNN in feature extraction and dimensionality reduction, as well as the extraction of higher-order features from the input information and the function of filtering useless information are integrated with the LSTM time series to improve the prediction accuracy and speed up data processing. By establishing a CNN-LSTM combined model, the first half is the CNN network, in which the convolution layer is used to extract features, and the second half is the LSTM prediction, which is used to extract the time features of the cable data. Finally, the LSTM and convolution layers are combined through the fully connected layer to improve the memory module capacity of the LSTM network and avoid the prediction lag problem caused by the large amount of data. Finally, the output layer is used to generate the final prediction output.

[0078] In practical applications, the data obtained through the model may be unreliable due to problems such as boundary condition settings and grid generation. Deleting these data may result in missing data. Therefore, the missing data can be compensated by filling or deleting them.

[0079] 1. Deletion method: The deletion method is to delete the incomplete data set and save the remaining complete data for model training. This method is suitable for when the characteristic factors of the data set are small and the impact on model training is low. If the characteristic factors of the data set are large, it will affect the prediction value of the model, thereby increasing the prediction error.

[0080] 2. Filling method: Select the median or mean in mathematical statistics to replace the missing eigenvalues. This method is suitable for data groups with more characteristic factors.

[0081] 3. Data normalization: Most of the feature dimensions and numerical values ​​in the data set are different, which will greatly affect the model and even make it impossible to complete the training. Therefore, the original data set needs to be preprocessed with data standardization before inputting it into the model, which can improve the model training effect. The commonly used methods are:

[0082] (1) Min-max normalization: This method normalizes and scales the original data to between [0,1]. The biggest disadvantage of the normalization method is that if the maximum and minimum values ​​in the data set are very different, this method will not be able to eliminate the dimension effect and will affect the normalization effect. As shown in the following formula:

[0083] (2) Z-Score Standardization: Z-Score standardization is to transform the data, including the original data 𝑋𝑚𝑒𝑎𝑛 and standard deviation This method is also called standard deviation normalization method. Compared with min-max normalization method, this method can eliminate the dimension effect. As shown in the following formula:

[0084] In order to evaluate the prediction ability of the CNN-LSTM hybrid model for cable temperature, three error evaluation indicators, namely mean absolute error (MAE), determination coefficient (R2) and root mean square error (RMSE), were selected to measure the prediction accuracy of the cable temperature prediction model.

[0085] 1. Mean absolute error (MAE) The mean absolute error represents the mean of the absolute deviation between the predicted value and the true value of the cable temperature. The smaller the MAE value, the better the effect of the cable temperature prediction model. As shown in the following formula:

[0086] 2. Determination coefficient R-Squared, the denominator represents the sum of the squared difference between the true value and the mean, and the numerator is the sum of the squared difference. According to R2, the value range is [0,1]. When R2 is 0, it means that the model fitting effect is very poor, on the contrary, if it is 1, the fitting effect is good. And the larger the R2, the better the model fitting effect. As shown in the following formula:

[0087] 3. The root mean square error measures the deviation between the predicted value and the true value, and is more sensitive to outliers in the data. The RMSE is the square root of the ratio of the square of the deviation between the predicted value and the true value to the number of observations n. The smaller the RMSE, the better the model effect. As shown in the following formula:

[0088] 4. Temperature forecast: The temperature data of the finite element model is input into the CNN-LSTM model to predict the temperature change during a fire in an underground cable tunnel. The specific method is as follows: 1. Establish a COMSOL finite element cable transient simulation model: determine its electromagnetic-thermal-fluid multi-physics field coupling, set the cable and tunnel structure parameters, laying conditions, load current, boundary conditions and solver, etc.

[0089] 2. Obtain sample data of cable core temperature: According to the cable transient model, input data conditions such as ambient temperature, wind speed, and load current. Through transient research, obtain the temperature data of the cable core within 5 days as sample data for CNN-LSTM.

[0090] 3. Data set division and preprocessing of data samples: The first 70% of the cable temperature sample data is used as the training set of the CNN-LSTM model. The last 30% of the data is used as the validation set of the CNN-LSTM model. The original sample data of the cable core temperature data set is normalized. Select X={X1(1),…,X1(n),…Xk(1),…,Xk(n)} as the input variable of the prediction model, and Y={y(n),y(n+1),…,y(n+m)} as the output variable, such as Figure 3 shown.

[0091] In this model, k is the number of measurable parameters, such as outer sheath temperature, ambient temperature, load, time, electromagnetic intensity, etc.; n is the time series; m is the mth moment of predicting the cable core temperature. k, n, and m are natural numbers.

[0092] 4. Establish a prediction model: Select ReLU as the activation function, use the Adam function for optimization training with a learning rate of 0.001, use the mean absolute error (MAE) and determination coefficient (R2) as the model performance evaluation indicators, and establish a CNN-LSTM cable core temperature prediction model.

[0093] 5. Output cable core prediction: Through the operation of its CNN-LSTM hybrid model, the required cable core temperature prediction value is obtained in the output layer.

[0094] (V) Model optimization: The attention mechanism is introduced based on the CNN-LSTM model to optimize model performance and improve prediction accuracy.

[0095] The attention mechanism is a research based on natural language processing, image detection, speech recognition, computer and other technologies. It can give different weights to different parts of different things, thereby reducing the role and influence of irrelevant parts. This section uses the Attention mechanism to assign different weights to the output hidden layer vectors in the CNN-LSTM network to optimize the performance of cable core temperature prediction. The correlation between each Key is calculated by learning function a, and the similarity score mi of the hidden layer Hi output at each time step is obtained. Then, the similarity score mi is normalized by the softmax function to obtain the attention weight ai of each Hi. Finally, the product of ai and Hi is calculated and summed to obtain the context vector c.

[0096] H1, H2, ..., Hn represent the hidden layer vector output by the LSTM neural network at each time step, as the key, and n is the time step. The calculation process is shown in the following formula:

[0097] (VI) Experimental verification: The purpose of the experimental verification is to verify the accuracy and practicality of the proposed underground cable tunnel fire temperature prediction method. By simulating the fire situation in the actual cable tunnel, collecting temperature data, and comparing it with the model prediction results, the effectiveness of the model is evaluated.

[0098] Preliminary preparation: Set up the ignition source: Under the premise of ensuring safety, select an area in the cable tunnel as the experimental ignition source. Use a controllable ignition device (electric heating wire or burner) to simulate the fire caused by cable fault.

[0099] Experimental cable layout: Select a cable with the same (or similar) material and structure as the actual operating cable as the experimental cable. Place the experimental cable near the ignition source to ensure that the flame can directly contact the cable.

[0100] Installation of temperature measurement system: High temperature resistant temperature sensors are installed at different positions of the experimental cable to monitor the temperature changes on the surface and inside of the cable in real time.

[0101] Experimental protection device configuration: Configure necessary fire-fighting equipment (automatic sprinkler system, gas fire extinguishing system, etc.) to ensure the safety of the experiment. Set up emergency evacuation routes and safety warning signs to ensure the safety of the experimenters.

[0102] Data logging equipment: Install data logging equipment, including data loggers and computer systems, to record temperature sensor data in real time.

[0103] Experimental process: Simulate fire: Start the ignition source to simulate the cable tunnel fire. Observe the spread of flames and smoke to ensure that the experimental conditions match the fire scenario predicted by the model.

[0104] Data collection: Monitor the data of temperature sensors in real time and record the temperature changes of flames and smoke. Record any abnormal conditions during the experiment, including sensor failure or changes in experimental conditions.

[0105] Safety monitoring: During the experiment, the safety of the experimental area is continuously monitored. Once any dangerous signal is detected, the protection device is immediately activated and the experiment is terminated.

[0106] Data analysis and verification of experimental results: Data comparison: The flame temperature data collected in the experiment were compared with the model prediction results. The prediction errors, including mean absolute error (MAE) and root mean square error (RMSE), were calculated to evaluate the accuracy of the model prediction.

[0107] Model adjustment: Adjust and optimize the model based on experimental data to improve the accuracy of prediction. If there is a significant difference between the experimental results and the model prediction, analyze the possible reasons and adjust the model parameters.

[0108] Verification report: Write an experimental verification report, record the experimental process, data analysis results and model adjustments in detail. Propose suggestions for model improvement to provide reference for future experiments and model applications.

[0109] In yet another embodiment of the present invention, a system for predicting the fire temperature of an urban underground cable tunnel network is provided, which can be used to implement the above-mentioned method for predicting the fire temperature of an urban underground cable tunnel network. Specifically, the system includes: Data acquisition module, used to collect environmental data in the cable tunnel and material data of the cable; The finite element model building module is used to build a three-dimensional finite element model including electromagnetic field, thermal field and fluid field based on environmental data and material data, simulate the temperature response of the cable under different conditions, and obtain sample data of the cable core temperature; The hybrid model building module is used to train the convolutional neural network (CNN) to extract features based on environmental data and material data, and then process the time series data through the long short-term memory network (LSTM) to build a CNN-LSTM hybrid model; The prediction output module is used to use the sample data of the cable core temperature obtained as the input of the CNN-LSTM hybrid model to obtain the predicted value of the cable core temperature.

[0110] The division of modules in the embodiments of the present invention is schematic and is only a logical function division. There may be other division methods in actual implementation. In addition, each functional module in each embodiment of the present invention may be integrated into one processor, or may exist physically separately, or two or more modules may be integrated into one module. The above-mentioned integrated modules may be implemented in the form of hardware or in the form of software functional modules.

[0111] In another embodiment of the present invention, a computer device is provided, the computer device comprising a processor and a memory, the memory being used to store a computer program, the computer program comprising program instructions, and the processor being used to execute the program instructions stored in the computer storage medium. The processor may be a central processing unit (CPU), or may be other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components, etc., which are the computing core and control core of the terminal, which are suitable for implementing one or more instructions, and are specifically suitable for loading and executing one or more instructions in the computer storage medium to implement the corresponding method flow or corresponding function; the processor described in the embodiment of the present invention can be used for the operation of the urban underground cable tunnel network fire temperature prediction method.

[0112] In another embodiment of the present invention, the present invention further provides a storage medium, specifically a computer-readable storage medium (Memory), which is a memory device in a computer device for storing programs and data. It is understandable that the computer-readable storage medium here can include both built-in storage media in the computer device and, of course, extended storage media supported by the computer device. The computer-readable storage medium provides a storage space, which stores the operating system of the terminal. In addition, one or more instructions suitable for being loaded and executed by the processor are also stored in the storage space, and these instructions can be one or more computer programs (including program codes). It should be noted that the computer-readable storage medium here can be a high-speed RAM memory or a non-volatile memory, such as at least one disk memory. The processor can load and execute one or more instructions stored in the computer-readable storage medium to implement the corresponding steps of the fire temperature prediction method for the urban underground cable tunnel network in the above embodiment.

[0113] It will be appreciated by those skilled in the art that embodiments of the present invention may be provided as methods, systems, or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Furthermore, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0114] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0115] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0116] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0117] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the above embodiments, ordinary technicians in the relevant field should understand that the specific implementation methods of the present invention can still be modified or replaced by equivalents. Any modification or equivalent replacement that does not depart from the spirit and scope of the present invention should be covered within the scope of protection of the claims of the present invention.

Claims

1. A method for predicting fire temperature in an urban underground cable tunnel network, characterized in that: include: Collect environmental data inside the cable tunnel and material data of the cables; Based on environmental data and material data, a three-dimensional finite element model including electromagnetic field, thermal field and fluid field is established to simulate the temperature response of the cable under different conditions and obtain sample data of cable core temperature; Based on environmental data and material data, the convolutional neural network (CNN) is trained to extract features, and then the long short-term memory (LSTM) network is used to process time series data to build a CNN-LSTM hybrid model. The obtained sample data of cable core temperature is used as the input of the CNN-LSTM hybrid model to obtain the predicted value of the cable core temperature.

2. The method for predicting fire temperature of an urban underground cable tunnel network according to claim 1, characterized in that: The collection of environmental data in the cable tunnel and material data of the cable includes: Ambient temperature, wind speed, humidity, and load current data in the cable tunnel; Collect the thermal conductivity, specific heat capacity and density parameters of the cable core, insulation layer and buffer layer materials.

3. The method for predicting fire temperature of an urban underground cable tunnel network according to claim 1, characterized in that: The three-dimensional finite element model including electromagnetic field, thermal field and fluid field is established based on the environmental data and material data, including: Using COMSOL multi-physics simulation software, a three-dimensional finite element model including electromagnetic field, thermal field and fluid field is established. According to the actual size and material properties of the cable tunnel, the geometric model of the tunnel and cable is constructed. Specifically: Heat conduction: According to Fourier's law, the heat flux density q is proportional to the rate of temperature change in that direction, and its function is shown as follows: Where, q is the heat flux density in W / m 2 ; Ф is the heat flow through area A; λ is the thermal conductivity, the unit is W / (m·K); For thermal energy balance within the cable: in, For the inflow of heat, To remove heat, The Joule heat generated by the cable core; Heat convection: The function is as follows: in, It represents the temperature difference between the cable outer skin and the air in the tunnel; h is the heat transfer coefficient of the cable outer sheath surface, the unit is w / (m 2 *k); The calculation of the heat generated by the heat source and the increase in internal energy is shown as follows: Among them, is the air density; is the heat capacity at constant pressure; Thermal radiation: The radiation energy that can be absorbed per unit time is shown in the following formula: Where A is the area of ​​the object; T is the temperature of the object; When the surface temperature of the cable in the tunnel is 2T and the air temperature is 1T, the radiation heat between the cable and the air is as follows: is the surface area of ​​the object; is the emissivity of air and does not exceed 1; The differential form of Maxwell's equations for the electromagnetic field in cables is shown below: in, and are source variables, which are eddy current density and current density; The relationship between the parameters of Maxwell's equations and the relevant field quantities is expressed as follows: 𝐵 Among them, 𝜀 is the dielectric constant; 𝜇 is the magnetic permeability; 𝜎 is the electrical conductivity; according to Maxwell's equations, the vector magnetic potential equations of each layer of the cable are derived as shown below: Among them, 𝐽 𝑠 is the current density of the cable core; e is the eddy current density of the metal sheath; A1-A4 represent the vector magnetic potential of the cable core, insulation layer, metal sheath layer and outer sheath layer respectively; The calculation formula for the loss of each layer of the cable is shown as follows: In the formula, is the power loss per unit length of each layer; is the electrical conductivity of each material.

4. The method for predicting fire temperature of an urban underground cable tunnel network according to claim 3 is characterized in that: The simulation of the temperature response of the cable under different conditions to obtain sample data of the cable core temperature includes: Set its load current, initial temperature and boundary conditions, perform mesh generation, calculate the resistivity and dielectric loss angle parameters at the current temperature, calculate the dielectric loss angle and resistivity parameters at the current temperature; solve the temperature field and fluid field equations, calculate the unit temperature and wind speed, call the multi-physics field coupling module, calculate the temperature distribution and fluid distribution, and finally determine whether the temperature difference meets the requirements of the output results.

5. The method for predicting fire temperature of an urban underground cable tunnel network according to claim 1, characterized in that: Based on the environmental data and material data, the convolutional neural network CNN is trained to extract features, and then the time series data is processed by the long short-term memory network LSTM to construct a CNN-LSTM hybrid model, including: Use CNN to extract key features of temperature change trends and load current peaks from historical data, and design convolutional layers and pooling layers to extract temporal and spatial features; Use LSTM to process time series data, capture the long-term dependency of cable temperature changes, design the LSTM network structure, and determine the number of hidden layers and neurons; Use the collected data to train the CNN-LSTM model and evaluate the model performance through cross-validation method; adjust the model parameters of learning rate and batch size to optimize the model performance.

6. The method for predicting fire temperature of an urban underground cable tunnel network according to claim 1, characterized in that: The obtained cable core temperature sample data is used as the input of the CNN-LSTM hybrid model to obtain the predicted value of the cable core temperature, including: Establish a COMSOL finite element transient simulation model for cables: determine its electromagnetic-thermal-fluid multi-physics field coupling, set cable and tunnel structural parameters, laying conditions, load current, boundary conditions and solver; Get sample data of cable core temperature: According to the cable transient model, input its ambient temperature, wind speed and load current data conditions, and obtain the temperature data of the cable core through transient research as the sample data of CNN-LSTM; Dataset division and preprocessing of sample data: The first 70% of the cable temperature sample data is used as the training set of the CNN-LSTM model, and the last 30% of the data is used as the validation set of the CNN-LSTM model. The original sample data of the cable core temperature data set is normalized, and X={X1(1),…,X1(n),…Xk(1),…,Xk(n)} is selected as the input variable of the prediction model, and Y={y(n),y(n+1),…,y(n+m)} is selected as the output variable; In this model, k is the number of measurable parameter types; n is the time series; m is the mth moment of the predicted cable core temperature; k, n, and m are natural numbers; Establish a prediction model: Select ReLU as the activation function, use the Adam function for optimization training with a learning rate of 0.001, use the mean absolute error MAE and determination coefficient R2 as the model performance evaluation indicators, and establish a CNN-LSTM cable core temperature prediction model; Through the operation of its CNN-LSTM hybrid model, the desired cable core temperature prediction value is obtained in the output layer.

7. The method for predicting fire temperature of an urban underground cable tunnel network according to claim 6, characterized in that: The attention mechanism is introduced into the CNN-LSTM model. The correlation between time steps is calculated through the learning function, different weights are assigned, and the weights are normalized using the softmax function to obtain the attention weight of each time step.

8. Urban underground cable tunnel network fire temperature prediction system, characterized in that: include: Data acquisition module, used to collect environmental data in the cable tunnel and material data of the cable; The finite element model building module is used to build a three-dimensional finite element model including electromagnetic field, thermal field and fluid field based on environmental data and material data, simulate the temperature response of the cable under different conditions, and obtain sample data of the cable core temperature; The hybrid model building module is used to train the convolutional neural network (CNN) to extract features based on environmental data and material data, and then process the time series data through the long short-term memory network (LSTM) to build a CNN-LSTM hybrid model; The prediction output module is used to use the sample data of the cable core temperature obtained as the input of the CNN-LSTM hybrid model to obtain the predicted value of the cable core temperature.

9. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the steps of the urban underground cable tunnel network fire temperature prediction method as described in any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the steps of the urban underground cable tunnel network fire temperature prediction method as described in any one of claims 1 to 7 are implemented.

Images