Method for dynamically determining explosion safety distance based on deep learning model

Abstract

The invention relates to the technical field of risk assessment, and discloses a method for dynamically determining an explosion safety distance based on a deep learning model, and the method comprises the following steps: S1, obtaining multiple groups of data through an engineering simulation platform and CFD software, and constructing the multiple groups of data into a training data set; s2, training a deep learning model through the training data set, wherein the deep learning model obtains a relationship between input features and output features; s3, acquiring real-time data of pressure, temperature, gas concentration and flow velocity in the environment at the corresponding time point through a data acquisition module; s4, performing numerical simulation on the explosion process of the created corresponding explosion scene model through an engineering simulation platform and CFD software; s5, obtaining a prediction value through a deep learning model; s6, the explosion safety distance is calculated through an explosion safety distance calculation formula; and S7, repeating the steps S3-S6, and dynamically determining the explosion safety distance in real time. The method can be applied to different explosion scenes, and can accurately obtain the safety distance in the explosion scene in real time.

Description

Technical Field

[0001] The present invention relates to the technical field of risk assessment, and in particular to a method for dynamically determining explosion safety distances based on a deep learning model. Background Art

[0002] In industrial production, explosions pose significant risks to equipment, personnel, and the environment. Existing methods for determining safe distances in explosion situations often rely on empirical formulas, static simulations, or a single data source. However, relying solely on empirical formulas or static simulations to determine safe distances ignores the complex working conditions and ever-changing environmental conditions in real-world scenarios, resulting in low safety distance prediction accuracy. Furthermore, relying solely on a single data source, such as the diffusion path of chemicals, is only applicable within chemical industrial parks and has a limited scope of application. Summary of the Invention

[0003] In order to overcome the shortcomings of the existing technology, the purpose of the present invention is to provide a method for dynamically determining the explosion safety distance based on a deep learning model, which can be applied to different explosion scenarios and can accurately obtain the safety distance in the explosion scenario in real time, thereby assisting staff to make accurate safety response decisions.

[0004] The technical solution adopted by the present invention is as follows: a method for dynamically determining an explosion safety distance based on a deep learning model, comprising the following steps:

[0005] S1: Create different explosion scenario models using an engineering simulation platform. Use CFD software to numerically simulate the explosion process in different scenarios to obtain multiple sets of data. Each set of data includes spatial coordinates, time step, local pressure value, shock wave propagation speed, overpressure threshold at a specific location, total energy released by the explosion, and danger zone radius. These multiple sets of data are then constructed into training datasets.

[0006] S2: The deep learning model is trained using the training dataset. The deep learning model obtains the relationship between the input feature space coordinates, time step, local pressure value, shock wave propagation speed, and the output feature overpressure threshold at a specific location, the total energy released by the explosion, and the radius of the danger zone.

[0007] S3: Obtain real-time data of pressure, temperature, gas concentration, and flow rate in the environment at the corresponding time point through the data acquisition module;

[0008] S4: Using the engineering simulation platform, the collected real-time data of pressure, temperature, gas concentration, flow rate, and the geometric parameters of the explosive object are processed to create a corresponding explosion scene model. The explosion process of the corresponding explosion scene model is numerically simulated using CFD software to obtain the spatial coordinates, time step, local pressure value, and shock wave propagation speed;

[0009] S5: Input the obtained spatial coordinates, time step, local pressure value, and shock wave propagation velocity into the trained deep learning model, and use the deep learning model to obtain the predicted overpressure threshold at a specific location, the total energy released by the explosion, and the radius of the danger zone;

[0010] S6: Calculate the explosion safety distance at the corresponding time point using the obtained overpressure threshold, total energy released by the explosion, and the radius of the danger range through the explosion safety distance calculation formula;

[0011] S7: Repeat S3-S6. The data acquisition module transmits the collected information to the engineering simulation platform at a preset frequency to dynamically determine the explosion safety distance in real time.

[0012] As a preferred embodiment of the present invention, it is characterized in that: S1 includes the following steps:

[0013] S11: In the engineering simulation platform, define the solid boundary, opening boundary, and environmental boundary, set the initial pressure, temperature, gas concentration, and flow rate for the explosion area, define the geometric parameters of the explosion area, and create the corresponding explosion scene model;

[0014] S12: Use the Fluent Meshing tool in the CFD software to mesh the explosion scene model;

[0015] S13: Set up the turbulence model in the Fluent tool in the CFD software, simulate the spark ignition process using the form of component transfer, and set the time step and calculation duration;

[0016] S14: Start the Fluent simulation program and perform numerical simulation on the explosion process in the scenario;

[0017] S15: Use the Fluent tool in the CFD software to obtain simulation results, including pressure distribution cloud diagram, shock wave propagation path diagram, and shock wave diffusion time-pressure curve;

[0018] S16: Export the pressure distribution cloud map and shock wave propagation path map, extract the values ​​of the pressure distribution cloud map and shock wave propagation path map through image analysis and program reading functions, obtain spatial coordinates based on the grid nodes of the explosion scene model, obtain local pressure values ​​based on the pressure distribution cloud map and spatial coordinates, obtain shock wave propagation velocity based on the shock wave propagation path map, and obtain the overpressure threshold, total energy released by the explosion, and danger range radius from the CFD software result report;

[0019] S17: Repeat S11-S16 to create different explosion scene models and obtain multiple sets of data. The spatial coordinates, local pressure value, shock wave propagation speed, time step, overpressure threshold, total energy released by the explosion, and danger range radius in each set of data are used as labels. Multiple sets of data together constitute a training data set.

[0020] As a preferred embodiment of the present invention, S2 includes the following steps:

[0021] S21: Input the training data set into the deep learning model. The first hidden layer, the second hidden layer, and the fully connected layer in the deep learning model automatically obtain corresponding weights and biases through the training data set. At the same time, the input features in the training data set are batch-inputted into the deep learning model in a single group. Feature extraction is performed through the first hidden layer, the second hidden layer, and the fully connected layer to obtain three feature tensors. Each feature tensor corresponds to a set of temporal or spatial features of the input data. Each feature tensor is a row-column matrix structure arranged according to the output features and feature dimensions.

[0022] S22: Process a single feature in the corresponding feature tensor through the softmax layer in the deep learning model, calculate the probability of a single feature in the feature tensor relative to the sum of the single features in the feature tensor, and select the single feature with the largest probability as the predicted value;

[0023] S23: Calculate the deviation between the predicted value and the output features in the training dataset using the mean square error loss function, and optimize the weights and biases corresponding to the first hidden layer, second hidden layer, and fully connected layer in the deep learning model;

[0024] S24: Repeat S21-S24, and automatically stop training when the maximum number of iterations is reached.

[0025] As a preferred embodiment of the present invention, S21 further includes the following steps:

[0026] S211: The deep learning model includes a first hidden layer, a second hidden layer, a fully connected layer, and a softmax layer. The first hidden layer includes 16 LSTM units, and the second hidden layer includes 32 LSTM units.

[0027] S212: The training data set is input into the deep learning model. The first hidden layer, the second hidden layer, and the fully connected layer in the deep learning model automatically obtain corresponding weights and biases through the training data set. At the same time, the input features in the training data set are batch-inputted into the deep learning model in the form of a single group. Feature extraction is performed through the first hidden layer, the second hidden layer, and the fully connected layer to obtain three feature tensors. Each feature tensor corresponds to the temporal or spatial features of a group of batch samples. Each feature tensor is a row and column matrix structure arranged according to the sample and feature dimensions.

[0028] As a preferred embodiment of the present invention, S5 further includes the following steps:

[0029] S51: The obtained spatial coordinates, time step, local pressure value, and shock wave propagation velocity are input into the trained deep learning model as input features, as shown in formula (1).

[0030] X t =[x,y,z,t,P,v] (1),

[0031] In formula (1), X t represents the input features, x, y, z are spatial coordinates, t is the time step, P is the local pressure value, and v is the shock wave propagation velocity;

[0032] S52: The feature tensor of the first hidden layer, the second hidden layer, and the fully connected layer in the deep learning model extracts the input features, as shown in formulas (2), (3), (4), (5), (6), (7), (8), and (9).

[0033] f t =σ(W f ·[h t-1 ,X t ]+b f ) (2),

[0034] i t =σ(W i ·[h t-1 ,X t ]+b i ) (3),

[0035]

[0036] o t =σ(W o ·[h t-1 ,X t ]+b o ) (6),

[0037]

[0038] z=W fc ·h t ′+b fc (9),

[0039] In formulas (2), (3), (4), (5), (6), (7), (8), and (9), X t is the input feature, h t 、h t′ are the output of the first hidden layer and the output of the second hidden layer, z is the output of the fully connected layer, h t-1 is the hidden state of the first layer, which is set to zero vector, f t It is the forget gate, i t is the input gate, is a candidate state, C t is the cell state, o t is the output gate, W f 、W i 、W C 、W o 、W fc They are weight matrix, b f 、b i 、b C 、b o 、b fc They are bias;

[0040] S53: The single feature in the extracted corresponding feature tensor is processed by the softmax layer in the deep learning model, and the probability of each single feature in the feature tensor relative to the sum of the single features in the feature tensor is calculated. The single feature with the largest probability is selected as the predicted value, as shown in formula (10).

[0041]

[0042] In formula (10), z j is the jth feature of the fully connected layer;

[0043] S54: Repeat S53 to calculate the probability of each single feature in the other two feature tensors relative to the sum of the single features in the feature tensor, and take the single feature with the largest probability as the corresponding prediction value.

[0044] As a preferred embodiment of the present invention, S6 includes the following steps:

[0045] S61: Calculate the explosion safety distance at the corresponding time point using the obtained overpressure threshold, the total energy released by the explosion, and the radius of the danger range through the explosion safety distance calculation formula, as shown in formula (11).

[0046]

[0047] In formula (11), d is the safety distance, R is the radius of the danger range, E is the total energy released by the explosion, P is the overpressure threshold, k is a constant related to the environmental conditions, and n is an index related to the explosion type.

[0048] Compared with the prior art, the present invention has the following beneficial effects:

[0049] 1) The present invention simulates different explosion scenarios through an engineering simulation platform and obtains a large number of sets of data through CFD simulation, that is, data under different explosion scenarios are obtained, and a large amount of data is used as a training data set for the deep learning model to learn. Therefore, in the actual application stage, the deep learning model can accurately output the corresponding input features, that is, with the help of the trained deep learning model, the overpressure threshold, the total energy released by the explosion, and the radius of the danger range under the current complex environment are quickly predicted, avoiding time-consuming CFD simulation and achieving real-time dynamic response. It can be applied to different explosion scenarios, that is, the corresponding overpressure threshold, the total energy released by the explosion, and the radius of the danger range can be directly obtained based on the data obtained from different explosion scenarios, thereby dynamically obtaining the safe distance in real time;

[0050] 2) The present invention can be widely used in high-risk industrial scenarios such as chemical plants, oil storage tank areas, gas stations, high-pressure gas transmission stations, power facilities, mining, and other dangerous operations, effectively reducing accident risks. Through real-time collected environmental data and deep learning models, it can achieve rapid prediction of overpressure thresholds, total energy released by explosions, and radius of dangerous areas in explosion events, as well as dynamic optimization of safety distances. It improves the accuracy and real-time performance of safety distance calculations, adapts to the dynamic changes of complex industrial scenarios, significantly improves the efficiency of safety warnings in explosion scenarios, enhances the safety assurance capabilities of industrial sites, and improves decision-making efficiency.

[0051] 3) Traditional explosion safety distance calculations mostly rely on static parameters (such as empirical formulas, fixed deceleration coefficients, etc.), while deep learning models can integrate multi-dimensional dynamic data (such as shock wave propagation speed, pressure changes, ambient temperature and humidity, etc.) in real time, and automatically adjust the prediction results through nonlinear modeling. For example, the present invention introduces a time series data processing module (LSTM structure) to capture the dynamic evolution of the explosion process. Through real-time data drive and deep learning model prediction, high-precision dynamic simulation of explosion-hazardous areas is achieved.

[0052] 4) Dynamic Determination of Explosion Safety Distance Based on a Deep Learning Model: Compared to traditional chemical reaction kinetics methods for determining explosion safety distance, this method integrates multi-dimensional dynamic data (shock wave propagation velocity, pressure changes) in real time, capturing the dynamic evolution of the explosion process through time series models such as LSTM, and is adaptable to complex scenarios. Furthermore, by automatically extracting the spatiotemporal characteristics of shock wave propagation and implementing nonlinear modeling, the deep learning model overcomes the limitations of traditional methods, which rely on static parameters and simplified assumptions, and reduces errors in complex scenarios such as turbulence effects and obstacle reflections. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] Figure 1 It is a flow chart of a method for dynamically determining explosion safety distance based on a deep learning model of the present invention;

[0054] Figure 2This is a structural diagram of a deep learning model in the method for dynamically determining an explosion safety distance based on a deep learning model of the present invention;

[0055] Figure 3 This is a structural diagram of an LSTM unit in the method for dynamically determining explosion safety distance based on a deep learning model of the present invention;

[0056] Figure 4 The geometric structure of the explosive object in the method of dynamically determining the explosion safety distance based on the deep learning model of the present invention;

[0057] Figure 5 It is a grid diagram of the explosion scene model in the method for dynamically determining the explosion safety distance based on the deep learning model of the present invention;

[0058] Figure 6 It is a spatial coordinate diagram in the method of dynamically determining explosion safety distance based on a deep learning model of the present invention;

[0059] Figure 7 This is a geometric diagram of the explosive object when the method of the present invention for dynamically determining the explosion safety distance based on a deep learning model is applied to a natural gas pipeline leakage explosion scene;

[0060] Figure 8 This is a pressure distribution cloud diagram of a natural gas pipeline leakage and explosion scenario in the method of dynamically determining explosion safety distance based on a deep learning model of the present invention;

[0061] Figure 9 This is a diagram of the shock wave propagation path when the method of the present invention, which dynamically determines the explosion safety distance based on a deep learning model, is applied to a natural gas pipeline leakage explosion scenario. DETAILED DESCRIPTION

[0062] Typical embodiments that embody the features and advantages of the present invention are described in detail in the following description. It should be understood that the present invention is capable of various variations in different embodiments without departing from the scope of the present invention, and that the descriptions and illustrations are intended to be illustrative rather than limiting.

[0063] The present invention will be described in detail below with reference to the accompanying drawings and in conjunction with embodiments.

[0064] Methods for dynamically determining explosion safety distances based on deep learning models, such as Figure 1 As shown, the following steps are included:

[0065] S1: Create different explosion scenario models using an engineering simulation platform. Use CFD software to numerically simulate the explosion process in different scenarios to obtain multiple sets of data. Each set of data includes spatial coordinates, time step, local pressure value, shock wave propagation speed, overpressure threshold at a specific location, total energy released by the explosion, and danger zone radius. These multiple sets of data are then constructed into training datasets.

[0066] In this embodiment, the engineering simulation platform is ANSYS Workbench, and the specific location refers to the location in the explosion scene corresponding to the spatial coordinates of the key target area (such as key equipment, personnel activity area or boundary), which is used to evaluate the intensity of the explosion impact at that location.

[0067] Specifically, S1 includes the following steps:

[0068] S11: In the engineering simulation platform, define the solid boundary, opening boundary and environmental boundary, and set the initial pressure, temperature, gas concentration and flow rate for the explosion area, such as Figure 4 As shown, the geometric parameters of the explosion area are defined and the corresponding explosion scene model is created;

[0069] S12: Through the Fluent Meshing tool in CFD software, such as Figure 5 As shown, the explosion scene model is meshed;

[0070] S13: Set up the turbulence model in the Fluent tool in the CFD software, simulate the spark ignition process using the form of component transfer, and set the time step and calculation duration;

[0071] S14: starting the Fluent simulation program in the CFD software and performing a numerical simulation of the explosion process in the scenario;

[0072] S15: Use the Fluent tool in the CFD software to obtain simulation results, including pressure distribution cloud diagram, shock wave propagation path diagram, and shock wave diffusion time-pressure curve;

[0073] S16: Export the pressure distribution cloud map and shock wave propagation path map, and extract the values ​​of the pressure distribution cloud map and shock wave propagation path map through the image analysis and program reading function of the post-processing module in the CFD software, such as Figure 7 As shown in the figure, the spatial coordinates are obtained according to the grid nodes of the explosion scene model, the local pressure value is obtained according to the pressure distribution cloud map and the spatial coordinates, the shock wave propagation velocity is obtained according to the shock wave propagation path map, and the overpressure threshold, the total energy released by the explosion and the radius of the danger range are obtained through the CFD software result report;

[0074] S17: Repeat S11-S16 to create different explosion scene models and obtain multiple sets of data. The spatial coordinates, local pressure value, shock wave propagation speed, time step, overpressure threshold, total energy released by the explosion, and danger range radius in each set of data are used as labels. Multiple sets of data together constitute a training data set.

[0075] In this embodiment, the solid boundary, the opening boundary and the environment boundary are edited according to actual conditions in actual applications.

[0076] S2: The deep learning model is trained using the training dataset. The deep learning model obtains the relationship between the input feature space coordinates, time step, local pressure value, shock wave propagation speed, and the output feature overpressure threshold at a specific location, the total energy released by the explosion, and the radius of the danger zone.

[0077] Specifically, S2 includes the following steps:

[0078] S21: Input the training data set into the deep learning model. The first hidden layer, the second hidden layer, and the fully connected layer in the deep learning model automatically obtain corresponding weights and biases through the training data set. At the same time, the input features in the training data set are batch-inputted into the deep learning model in a single group. Feature extraction is performed through the first hidden layer, the second hidden layer, and the fully connected layer to obtain three feature tensors. Each feature tensor corresponds to a set of temporal or spatial features of the input data. Each feature tensor is a row-column matrix structure arranged according to the output features and feature dimensions.

[0079] Among them, such as Figure 2 、 3 As shown, S21 further includes the following steps:

[0080] S211: The deep learning model includes a first hidden layer, a second hidden layer, a fully connected layer, and a softmax layer. The first hidden layer includes 16 LSTM units, and the second hidden layer includes 32 LSTM units.

[0081] S212: The training data set is input into the deep learning model. The first hidden layer, the second hidden layer, and the fully connected layer in the deep learning model automatically obtain corresponding weights and biases through the training data set. At the same time, the input features in the training data set are batch-inputted into the deep learning model in the form of a single group. Feature extraction is performed through the first hidden layer, the second hidden layer, and the fully connected layer to obtain three feature tensors. Each feature tensor corresponds to the temporal or spatial features of a group of batch samples. Each feature tensor is a row and column matrix structure arranged according to the sample and feature dimensions.

[0082] S22: Process a single feature in the corresponding feature tensor through the softmax layer in the deep learning model, calculate the probability of a single feature in the feature tensor relative to the sum of the single features in the feature tensor, and select the single feature with the largest probability as the predicted value;

[0083] S23: Calculate the deviation between the predicted value and the output features in the training dataset using the mean square error loss function, and optimize the weights and biases corresponding to the first hidden layer, second hidden layer, and fully connected layer in the deep learning model;

[0084] S24: Repeat S21-S24, and automatically stop training when the maximum number of iterations is reached.

[0085] In this embodiment, the feature tensor, that is, the population matrix, and the model training process can also adopt an early termination strategy. When there is no obvious deviation between the predicted value calculated by the mean square error loss function and the output feature in the training data set, the training is automatically stopped.

[0086] In this embodiment, the input features are input into the deep learning model in batches in the form of a single group, which means that the input features input into the deep learning model each time are in units of one group of data, but the corresponding multiple groups of data are continuously input into the deep learning model.

[0087] In this embodiment, samples (more than 1000 groups) extracted from CFD software working condition simulation are used to train the deep learning model. The loss function is MSE, and the error of the final prediction value is controlled within ±10%, as shown in formula (12):

[0088]

[0089] In formula (12), n is the number of samples, y i is the true value of the i-th sample (that is, the output feature in the training data set), is the model's predicted value for the i-th sample.

[0090] S3: Obtain real-time data of pressure, temperature, gas concentration, and flow rate in the environment at the corresponding time point through the data acquisition module;

[0091] S4: Using the engineering simulation platform, the collected real-time data of pressure, temperature, gas concentration, flow rate, and the geometric parameters of the explosive object are processed to create a corresponding explosion scene model. The explosion process of the corresponding explosion scene model is numerically simulated using CFD software to obtain the spatial coordinates, time step, local pressure value, and shock wave propagation speed;

[0092] In this embodiment, the specific steps of S4 refer to S11-S15.

[0093] S5: Input the obtained spatial coordinates, time step, local pressure value, and shock wave propagation velocity into the trained deep learning model, and use the deep learning model to obtain the predicted overpressure threshold at a specific location, the total energy released by the explosion, and the radius of the danger zone;

[0094] Specifically, S5 further includes the following steps:

[0095] S51: The obtained spatial coordinates, time step, local pressure value, and shock wave propagation velocity are input into the trained deep learning model as input features, as shown in formula (1).

[0096] X t =[x,y,z,t,P,v] (1),

[0097] In formula (1), X t represents the input features, x, y, z are spatial coordinates, t is the time step, P is the local pressure value, and v is the shock wave propagation velocity;

[0098] S52: The feature tensor of the first hidden layer, the second hidden layer, and the fully connected layer in the deep learning model extracts the input features, as shown in formulas (2), (3), (4), (5), (6), (7), (8), and (9).

[0099] f t =σ(W f ·[h t-1 ,X t ]+b f ) (2),

[0100] i t =σ(W i ·[h t-1 ,X t ]+b i ) (3),

[0101]

[0102] o t =σ(W o ·[h t-1 ,X t ]+b o ) (6),

[0103]

[0104] z=W fc ·h t ′+b fc (9),

[0105] In formulas (2), (3), (4), (5), (6), (7), (8), and (9), X t is the input feature, h t 、h t ′ are the output of the first hidden layer and the output of the second hidden layer, z is the output of the fully connected layer, h t-1 is the hidden state of the first layer, which is set to zero vector, f t It is the forget gate, i t is the input gate, is a candidate state, C t is the cell state, o t is the output gate, W f 、W i 、W C 、W o 、W fc They are weight matrix, b f 、b i 、b C 、b o 、b fc They are bias;

[0106] S53: The single feature in the extracted corresponding feature tensor is processed by the softmax layer in the deep learning model, and the probability of each single feature in the feature tensor relative to the sum of the single features in the feature tensor is calculated. The single feature with the largest probability is selected as the predicted value, as shown in formula (10).

[0107]

[0108] In formula (10), z j is the jth feature of the fully connected layer;

[0109] S54: Repeat S53 to calculate the probability of each single feature in the other two feature tensors relative to the sum of the single features in the feature tensor, and take the single feature with the largest probability as the corresponding prediction value.

[0110] S6: Calculate the explosion safety distance at the corresponding time point using the obtained overpressure threshold, total energy released by the explosion, and the radius of the danger range through the explosion safety distance calculation formula;

[0111] S6 includes the following steps:

[0112] S61: Calculate the explosion safety distance at the corresponding time point using the obtained overpressure threshold, the total energy released by the explosion, and the radius of the danger range through the explosion safety distance calculation formula, as shown in formula (11).

[0113]

[0114] In formula (11), d is the safety distance, R is the radius of the danger range, E is the total energy released by the explosion, P is the overpressure threshold, k is a constant related to the environmental conditions, and n is an index related to the explosion type.

[0115] In this embodiment, the specific value of k is selected according to China's "Safety Regulations for Blasting" (GB 6722) and the American Petroleum Institute (API) standard, and n is usually 2 to 3.

[0116] S7: Repeat S3-S6. The data acquisition module transmits the collected information to the engineering simulation platform at a preset frequency to dynamically determine the explosion safety distance in real time.

[0117] Because the steps for obtaining spatial coordinates, local pressure values, shock wave diffusion paths, overpressure thresholds, total energy released by the explosion, and the radius of the danger zone differ in CFD software for numerical simulation of explosion models, obtaining these overpressure thresholds, total energy released by the explosion, and the radius of the danger zone requires more post-processing steps and analytical processing;

[0118] Specifically, in CFD post-processing software (such as CFD-Post), the spatial coordinate information of a certain point in the model can be obtained through the probe tool in the toolbar; the local pressure value can be directly obtained through the processing tool in CFD; during the simulation of the explosion, the propagation of the shock wave in space can be intuitively seen through the CFD software, and the shock wave diffusion speed can be directly obtained; while the acquisition of the overpressure threshold, the total energy released by the explosion and the radius of the danger range requires additional steps, which require data extraction (setting monitoring points in the model and recording the pressure changes of these points during the simulation process), data analysis (smoothing or filtering the data to eliminate the influence of numerical noise), energy calculation (integrating the internal energy, kinetic energy and other energy items in the flow field), and multiple analyses and adjustments of multiple simulation results before the result report of the CFD software can be obtained.

[0119] In this embodiment, the specific values ​​obtained by the data acquisition module are used to obtain the required six values ​​through CFD software. The overpressure threshold, total energy released by the explosion, and the radius of the danger range at the last three specific locations are predicted through deep learning. In the actual application stage, the predicted data can be quickly obtained through the deep learning model to achieve real-time dynamic response. There is no need to use offline CFD software simulation every time the dynamic safety distance is obtained. Although offline CFD software simulation has high accuracy, it consumes a lot of computing resources and cannot output timely predictions in real time. The three predicted values ​​can be quickly obtained through the deep learning model.

[0120] The following uses a natural gas pipeline leakage and explosion scenario as an example to explain in detail:

[0121] S3: Deploy multiple sensor nodes at key locations on the natural gas pipeline to collect the following data:

[0122] Initial pressure P0 = 1050000 Pa;

[0123] Initial temperature T0 = 298K;

[0124] Gas volume fraction CCH4 = 9.5% (i.e. gas concentration);

[0125] Flow velocity V0 = 10 m / s;

[0126] S4: Using the engineering simulation platform, the real-time data of pressure, temperature, gas concentration, and flow rate collected in the environment, as well as the geometric parameters of the explosive object, are processed to create a corresponding explosion scene model. The explosion scene model is divided into unstructured networks. The EDC turbulent reaction model in the CFD software (the EDC model is a detailed chemical reaction model used to simulate turbulent combustion) is used to simulate spark ignition. A 10ms time step is set, and the Fluent simulation program in the CFD software is started. The explosion process in this scenario is numerically simulated, and a pressure distribution cloud map and a shock wave propagation path map are obtained. The values ​​in the pressure distribution cloud map and the shock wave propagation path map are extracted through the image analysis and program reading functions of the post-processing module in the CFD software.

[0127] Get the spatial coordinates (1.0, 0.5, 0) according to the explosion scene model grid node, such as Figure 8 、 9 As shown in the figure, the local pressure value obtained according to the pressure distribution cloud map and spatial coordinates is 4.715×10 5 pa, the shock wave propagation velocity is 325 m / s obtained according to the shock wave propagation path diagram;

[0128] S5: The obtained spatial coordinates (1.0, 0.5, 0), time step 10ms, local pressure value 4.715×10 5 pa, shock wave propagation speed 325m / s as input feature vector X t =[x,y,z,t,P,v] is input into the trained deep learning model, and the features are output through the first hidden layer of 16 LSTM units, the second hidden layer of 32 LSTM units, the fully connected layer, and the softmax layer. Specifically, the overpressure threshold is 7000Pa, the total energy released by the explosion is 6.5×106J, and the radius of the danger zone is 24.1m;

[0129] S6: The obtained overpressure threshold, total energy released by the explosion, and the radius of the dangerous range are calculated using the explosion safety distance calculation formula to obtain the explosion safety distance at the corresponding time point, where k = 1.2, n = 2, as shown in formula (11).

[0130]

[0131] The predicted value of d is 28.5m.

[0132] The above embodiments are only preferred embodiments of the present invention and cannot be used to limit the scope of protection of the present invention. Any non-substantial changes and replacements made by technicians in this field on the basis of the present invention fall within the scope of protection required by the present invention.

Claims

1. A method for dynamically determining explosion safety distance based on a deep learning model, characterized by: The following steps are involved: S1: Create different explosion scenario models using an engineering simulation platform. Use CFD software to numerically simulate the explosion process in different scenarios to obtain multiple sets of data. Each set of data includes spatial coordinates, time step, local pressure value, shock wave propagation speed, overpressure threshold at a specific location, total energy released by the explosion, and danger zone radius. These multiple sets of data are then constructed into training datasets. S2: The deep learning model is trained using the training dataset. The deep learning model obtains the relationship between the input feature space coordinates, time step, local pressure value, shock wave propagation speed, and the output feature overpressure threshold at a specific location, the total energy released by the explosion, and the radius of the danger zone. S3: Obtain real-time data of pressure, temperature, gas concentration, and flow rate in the environment at the corresponding time point through the data acquisition module; S4: Using the engineering simulation platform, the collected real-time data of pressure, temperature, gas concentration, flow rate, and the geometric parameters of the explosive object are processed to create a corresponding explosion scene model. The explosion process of the corresponding explosion scene model is numerically simulated using CFD software to obtain the spatial coordinates, time step, local pressure value, and shock wave propagation speed; S5: Input the obtained spatial coordinates, time step, local pressure value, and shock wave propagation velocity into the trained deep learning model, and use the deep learning model to obtain the predicted overpressure threshold at a specific location, the total energy released by the explosion, and the radius of the danger zone; S6: Calculate the explosion safety distance at the corresponding time point using the obtained overpressure threshold, total energy released by the explosion, and the radius of the danger range through the explosion safety distance calculation formula; S7: Repeat S3-S6. The data acquisition module transmits the collected information to the engineering simulation platform at a preset frequency to dynamically determine the explosion safety distance in real time.

2. The method for dynamically determining explosion safety distance based on a deep learning model according to claim 1, characterized in that: S1 includes the following steps: S11: In the engineering simulation platform, define the solid boundary, opening boundary, and environmental boundary, set the initial pressure, temperature, gas concentration, and flow rate for the explosion area, define the geometric parameters of the explosion area, and create the corresponding explosion scene model; S12: Use the Fluent Meshing tool in the CFD software to mesh the explosion scene model; S13: Set up the turbulence model in the Fluent tool in the CFD software, simulate the spark ignition process using the form of component transfer, and set the time step and calculation duration; S14: Start the Fluent simulation program and perform numerical simulation on the explosion process in the scenario; S15: Use the Fluent tool in the CFD software to obtain simulation results, including pressure distribution cloud maps and shock wave propagation path maps; S16: Export the pressure distribution cloud map and shock wave propagation path map, extract the values ​​of the pressure distribution cloud map and shock wave propagation path map through image analysis and program reading functions, obtain spatial coordinates based on the grid nodes of the explosion scene model, obtain local pressure values ​​based on the pressure distribution cloud map and spatial coordinates, obtain shock wave propagation velocity based on the shock wave propagation path map, and obtain the overpressure threshold, total energy released by the explosion, and danger range radius from the CFD software result report; S17: Repeat S11-S16 to create different explosion scene models and obtain multiple sets of data. The spatial coordinates, local pressure value, shock wave propagation speed, time step, overpressure threshold, total energy released by the explosion, and danger range radius in each set of data are used as labels. Multiple sets of data together constitute a training data set.

3. The method for dynamically determining explosion safety distance based on a deep learning model according to claim 1, characterized in that: S2 includes the following steps: S21: Input the training data set into the deep learning model. The first hidden layer, the second hidden layer, and the fully connected layer in the deep learning model automatically obtain corresponding weights and biases through the training data set. At the same time, the input features in the training data set are batch-inputted into the deep learning model in a single group. Feature extraction is performed through the first hidden layer, the second hidden layer, and the fully connected layer to obtain three feature tensors. Each feature tensor corresponds to a set of temporal or spatial features of the input data. Each feature tensor is a row-column matrix structure arranged according to the output features and feature dimensions. S22: Process a single feature in the corresponding feature tensor through the softmax layer in the deep learning model, calculate the probability of a single feature in the feature tensor relative to the sum of the single features in the feature tensor, and select the single feature with the largest probability as the predicted value; S23: Calculate the deviation between the predicted value and the output features in the training dataset using the mean square error loss function, and optimize the weights and biases corresponding to the first hidden layer, second hidden layer, and fully connected layer in the deep learning model; S24: Repeat S21-S24, and automatically stop training when the maximum number of iterations is reached.

4. The method for dynamically determining explosion safety distance based on a deep learning model according to claim 1, characterized in that: S21 further includes the following steps: S211: The deep learning model includes a first hidden layer, a second hidden layer, a fully connected layer, and a softmax layer. The first hidden layer includes 16 LSTM units, and the second hidden layer includes 32 LSTM units. S212: The training data set is input into the deep learning model. The first hidden layer, the second hidden layer, and the fully connected layer in the deep learning model automatically obtain corresponding weights and biases through the training data set. At the same time, the input features in the training data set are batch-inputted into the deep learning model in the form of a single group. Feature extraction is performed through the first hidden layer, the second hidden layer, and the fully connected layer to obtain three feature tensors. Each feature tensor corresponds to the temporal or spatial features of a group of batch samples. Each feature tensor is a row and column matrix structure arranged according to the sample and feature dimensions.

5. The method for dynamically determining explosion safety distance based on a deep learning model according to claim 1, characterized in that: S5 also includes the following steps: S51: The obtained spatial coordinates, time step, local pressure value, and shock wave propagation velocity are input into the trained deep learning model as input features, as shown in formula (1). X t =[x,y,z,t,P,v] (1), In formula (1), X t represents the input features, x, y, z are spatial coordinates, t is the time step, P is the local pressure value, and v is the shock wave propagation velocity; S52: The feature tensor of the first hidden layer, the second hidden layer, and the fully connected layer in the deep learning model extracts the input features, as shown in formulas (2), (3), (4), (5), (6), (7), (8), and (9). f t =σ(W f ·[h t-1 ,X t ]+b f ) (2), i t =σ(W i ·[h t-1 ,X t ]+b i ) (3), the t =σ(W o ·[h t-1 ,X t ]+b o ) (6), z=W fc ·h t ′+b fc (9), In formulas (2), (3), (4), (5), (6), (7), (8), and (9), X t is the input feature, h t 、h t ′ are the output of the first hidden layer and the output of the second hidden layer, z is the output of the fully connected layer, h t-1 is the hidden state of the first layer, which is set to zero vector, f t It is the forget gate, i t is the input gate, is a candidate state, C t is the cell state, o t is the output gate, W f 、W i 、W C 、W o 、W fc They are weight matrix, b f 、b i 、b C 、b o 、b fc They are bias; S53: The single feature in the extracted corresponding feature tensor is processed by the softmax layer in the deep learning model, and the probability of each single feature in the feature tensor relative to the sum of the single features in the feature tensor is calculated. The single feature with the largest probability is selected as the predicted value, as shown in formula (10). In formula (10), z j is the jth feature of the fully connected layer; S54: Repeat S53 to calculate the probability of each single feature in the other two feature tensors relative to the sum of the single features in the feature tensor, and take the single feature with the largest probability as the corresponding prediction value.

6. The method for dynamically determining explosion safety distance based on a deep learning model according to claim 1, characterized in that: S6 includes the following steps: S61: Calculate the explosion safety distance at the corresponding time point using the obtained overpressure threshold, total energy released by the explosion, and the radius of the danger range through the explosion safety distance calculation formula, as shown in formula (11). In formula (11), d is the safety distance, R is the radius of the danger range, E is the total energy released by the explosion, P is the overpressure threshold, k is a constant related to the environmental conditions, and n is an index related to the explosion type.

Images