Restricted space fire blast consequence prediction method based on gas leakage process dynamic parameters

Abstract

The invention discloses a gas leakage process dynamic parameter-based limited space combustion explosion consequence prediction method. The method comprises the following steps of: acquiring concentration and speed segmentation time sequence data in every delta t time period from gas leakage in each working condition; forming a three-dimensional primary ignition source array, and simplifying the ignition source array according to a blasting numerical simulation result; obtaining a maximum temperature field data set, a static pressure field data set and a dynamic pressure field data set on a typical section in the burning and explosion process; mapping the concentration and speed segmented time sequence data set to a maximum temperature field, static pressure field and dynamic pressure field data set to obtain a combustible gas leakage explosion result prediction data set, preliminarily establishing a combustion explosion result prediction model, and training; and collecting gas leakage concentration and speed time sequence data in the limited space, and inputting the data into the combustion explosion consequence prediction model to obtain a temperature, static pressure and dynamic pressure maximum value field in the combustion explosion process. According to the invention, automatic prediction of the maximum overpressure, the dynamic pressure and the temperature field in the explosion process after combustible gas leakage in a real-time or super-real-time limited space is realized, and the prediction result is accurate.

Description

Technical Field

[0001] The invention relates to a method for predicting explosion consequences in a confined space based on dynamic parameters of a gas leakage process. Background Art

[0002] In confined spaces, especially those with significant confinement, such as large buildings and tunnels, the presence of natural gas storage, transportation, or application equipment presents a risk of gas leaks. Such incidents are often accompanied by a high degree of uncertainty regarding the leak source parameters and environmental conditions, making it difficult to accurately predict the consequences of explosions caused by gas leaks. When responding to gas leaks, being able to reasonably predict the gas explosion load is crucial not only for ensuring the safe evacuation and rescue of personnel within confined spaces, but also for providing critical information support for the effective handling of gas safety incidents.

[0003] Currently, there are three methods for calculating gas explosion pressure to predict the consequences of gas leak explosions in confined spaces: empirical formulas, numerical simulations, and intelligent algorithm-based calculations. However, these methods are based on predicting the consequences of explosions caused by uniformly mixed gases or empirical methods. They lack the ability to better predict the consequences of gas explosions caused by actual gas leaks and scenarios, resulting in low prediction accuracy. Summary of the Invention

[0004] The purpose of the present invention is to provide a method for predicting explosion consequences in a confined space based on dynamic parameters of a gas leakage process with high prediction accuracy.

[0005] The object of the present invention is achieved by the following technical measures: a method for predicting explosion consequences in a confined space based on dynamic parameters of a gas leakage process, characterized by comprising the following steps:

[0006] S1. Design a numerical simulation working condition for the flammable gas leakage and diffusion process in a confined space. Arrange concentration sensors and velocity sensors in the confined space to perform numerical simulation of gas leakage and diffusion. Collect segmented time series data of concentration and velocity from each sensor in each working condition within a time period of Δt from the start of the gas leakage. Normalize the segmented time series data of concentration and velocity to form segmented time series data of concentration and velocity.

[0007] S2. Ignition sources are arranged in a confined space at intervals of Δx along its length, Δy along its width, and Δz along its height to form a three-dimensional primary ignition source array. Different ignition source positions are set on the ignition source array. Using the combustible gas cloud obtained after a gas leak for a time period of ΔT under typical operating conditions in step S1, numerical simulations of combustion and explosion under the conditions of the combustible gas cloud are performed respectively. The temperature, static pressure, and dynamic pressure curves formed by the numerical simulation results are compared, and the ignition source array is simplified based on the results.

[0008] S3. Utilize the combustible gas clouds formed under different gas leakage durations in different working conditions in the confined space in step S1, and sequentially set different ignition source positions in the simplified ignition source array to conduct numerical simulations of gas combustion and explosion under different leakage time conditions, obtain the time series data of the two-dimensional temperature field, static pressure field, and dynamic pressure field on the typical cross-sections corresponding to different ignition source positions during the gas combustion and explosion process, and extract the maximum value in the time series data of the entire combustion and explosion process for each pixel point in the time series data to obtain the maximum temperature field, static pressure field, and dynamic pressure field data sets on the typical cross-sections during the combustion and explosion process;

[0009] S4. Map the concentration and velocity segmented time series data sets to the maximum temperature field, static pressure field, and dynamic pressure field data sets after a×Δt time period to obtain the prediction data set for the combustion and explosion consequences of combustible gas leakage and explosion, and initially establish a prediction model for combustion and explosion consequences. When a = 0, real-time prediction of combustion and explosion consequences is achieved, and when a>0, ultra-real-time prediction of gas combustion and explosion consequences is achieved;

[0010] S5. Use the prediction data set for the combustion and explosion consequences of combustible gas leakage and explosion to train the prediction model for combustion and explosion consequences, and use the loss values of the maximum temperature field, static pressure field, and dynamic pressure field during the prediction process of combustion and explosion consequences as the loss function during the training process to conduct deep learning training to obtain a trained prediction model for combustion and explosion consequences;

[0011] S6. Collect the time series data of gas leakage concentration and velocity in the confined space and input them into the prediction model for combustion and explosion consequences, and calculate the maximum value fields of temperature, static pressure, and dynamic pressure during the combustion and explosion process based on the time series data of concentration and velocity through the prediction model for combustion and explosion consequences.

[0012] The present invention combines existing deep learning methods, utilizes the concentration and velocity data monitored during the leakage process of combustible gas in the confined space, and realizes the automatic prediction of the maximum overpressure, dynamic pressure, and temperature fields during the combustion and explosion process after the leakage of combustible gas in the confined space in real time or ultra-real time, and the prediction results are accurate.

[0013] During the process of simplifying the ignition source array in the present invention, if the difference between the corresponding gas temperature, static pressure, and dynamic pressure in the calculation results of the combustion and explosion conditions under the ignition source position conditions within a certain space range is within the set range, the ignition sources between this ignition source position spacing are ignored, and an ignition source is selected within this position area to replace the calculation results.

[0014] The set range in the present invention is less than 20%.

[0015] In step S1 of the present invention, numerical simulation or experimental methods are used to carry out the numerical simulation working condition design of the combustible gas leakage and diffusion process in the confined space.

[0016] In step S1 of the present invention, the concentration and velocity segmented time-series data are enhanced by adding noise, changing the relative position of the sensor and the leakage source, or data mirroring.

[0017] In step S2 of the present invention, in the typical working conditions, at least two levels are taken for each factor of the leakage source conditions and the confined space environmental conditions.

[0018] The typical cross-section of the present invention is the central cross-section of the confined space.

[0019] In step S2 of the present invention, the temperature, static pressure, and dynamic pressure curves are the data of the central cross-section in the width direction of the confined space.

[0020] Compared with the prior art, the present invention has the following remarkable effects:

[0021] (1) High automation: The present invention can predict the parameters of the combustion explosion consequence field formed after the gas combustion explosion through the time-series data obtained by the distributed gas concentration sensors and gas velocity sensors arranged in the confined space.

[0022] (2) High prediction accuracy: The present invention can predict the consequence characteristics of possible combustion explosion accidents through the dynamic parameters of the highly non-linear gas leakage process by means of deep learning methods, and obtain high prediction accuracy.

[0023] (3) High self-adaptability: The present invention can be automatically detected by the central integrated control system according to the collected sample information, and there is no time limit. As long as the relevant parameters are set, the detection can be carried out both during the day and at night. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] The present invention will be further described in detail below with reference to the drawings and specific embodiments.

[0025] Figure 1 is the flow chart of the combustible gas leakage data processing of the present invention;

[0026] Figure 2 is the flow chart of the ignition source array setting of the present invention;

[0027] Figure 3 is the flow chart of the explosion consequence data processing and prediction of the present invention;

[0028] Figure 4 is the flow chart of the deep learning prediction of the combustion explosion consequence of the example of the present invention;

[0029] Figure 5 is the schematic diagram of the deep learning prediction model of the example of the present invention;

[0030] Figure 6It is a schematic diagram of the prediction result of the combustion and explosion consequences of an example of the present invention. Specific implementation manners

[0031] The present invention will be described in detail below in conjunction with the embodiments and their accompanying drawings to help those skilled in the art better understand the inventive concept of the present invention. However, the protection scope of the claims of the present invention is not limited to the following embodiments. For those skilled in the art, all other embodiments obtained without creative labor on the premise of not departing from the inventive concept of the present invention belong to the protection scope of the present invention.

[0032] As Figures 1 to 6 shown, it is a method for predicting the combustion and explosion consequences in a confined space based on the dynamic parameters of the gas leakage process of the present invention, including the following steps:

[0033] S1. Refer to Figure 1 , carry out the numerical simulation condition design of the combustible gas leakage and diffusion process in the confined space. Arrange concentration sensors and velocity sensors in the confined space, conduct the numerical simulation of gas leakage and diffusion, collect the concentration and velocity segmented time series data of each sensor at intervals of Δt from the start of gas leakage in each condition, and perform normalization processing on the concentration and velocity segmented time series data to form the concentration and velocity segmented time series data sets; specifically as follows:

[0034] Use numerical simulation or experimental methods to carry out the numerical simulation condition design of the combustible gas leakage and diffusion process in the confined space. It is necessary to consider the leakage source conditions such as leakage source flow rate and orientation, as well as the confined space environmental conditions such as environmental wind, obstacles, and structures (various factors of the conditions) according to local conditions. The gas leakage and diffusion conditions are recorded as N groups.

[0035] Conduct numerical simulations of gas leakage and diffusion. The duration of each numerical simulation case is denoted as ΔT, and a series of gas concentration and air velocity sensor groups are set in the confined space. Denote the number of concentration sensors as M, and sequentially number each concentration sensor point (the serial numbers range from 1 - M); denote the number of velocity sensors as Q, and sequentially number each velocity sensor point (the serial numbers range from 1 - Q). The layout of concentration sensors and velocity sensors is based on the principle of convenience in sensor layout within the confined space and the ability to uniformly capture the flow field information during the leakage process of combustible gas in the confined space. And save the concentration parameters and velocity parameters of the sensor group in the confined space formed at every Δt time interval starting from the beginning of gas leakage during the numerical simulation process (at the 0 - Δt, Δt - 2Δt, 2Δt - 3Δt, …, (n - 1)Δt - nΔt moments during the leakage and diffusion numerical simulation process, where the value range of n is (1, ΔT / Δt)). Save the concentration and velocity segmented time - series data of each sensor in each case as tensor data in sequence as the input data of the prediction model. The rank of the velocity tensor data is (N×ΔT / Δt, Δt, M), and the rank of the velocity tensor data is (N×ΔT / Δt, Δt, Q). The first dimension of the concentration and data tensor data represents that there are N×ΔT / Δt samples in this tensor data, the second dimension represents that each sample contains Δt time series, and the third dimension represents that there are M or Q sensors. At the same time, methods such as adding noise, changing the relative position of the sensor group and the leakage source, or data mirroring can be considered to enhance the data to improve the subsequent data prediction performance.

[0036] Perform global normalization on the formed concentration and velocity tensor data sets respectively. Commonly used standard normalization or maximum - minimum normalization and other methods can be considered.

[0037] S2, Participate Figure 2 , Set ignition sources at intervals of Δx in the length direction, Δy in the width direction, and Δz in the height direction within the confined space to form a three - dimensional primary ignition source array. Set different ignition source positions on the ignition source array, and use the combustible gas cloud obtained after ΔT duration of gas leakage under the typical working conditions in step S1 to conduct gas combustion and explosion numerical simulations under the conditions of this combustible gas cloud respectively, and compare the temperature, static pressure, and dynamic pressure curves (maximum pressure dynamic pressure temperature curves) formed by the numerical simulation results, and simplify the ignition source array according to the results; specifically as follows:

[0038] Set a three - dimensional primary ignition source array in the confined space, where a series of ignition sources are set at intervals of Δx in the length direction, Δy in the width direction, and Δz in the height direction of the confined space to form a three - dimensional primary ignition source array.

[0039] Using the formed three-dimensional primary ignition source array, set different ignition source positions on the ignition source array, and use the combustible gas cloud obtained after the gas leakage ΔT duration under typical working conditions (at least 2 levels for each factor) in step S1 to conduct numerical simulations of gas combustion and explosion under the conditions of this combustible gas cloud respectively. And compare the representative temperature, static pressure, and dynamic pressure curves formed by the numerical simulation results. According to the results, the ignition source array can be simplified. The simplification principle is: if the difference between the corresponding gas temperature, static pressure, and dynamic pressure in the calculation results of the combustion and explosion working conditions under the ignition source position conditions within a certain space range is within an acceptable range, the acceptable range of error generally refers to a reasonable range where the difference is less than 20%. Then ignore the ignition sources between this ignition source position spacing, and select one ignition source within this position area to replace the calculation results. The simplified ignition source array is named the simplified version of the ignition source array, and the number of ignition sources in the simplified version of the ignition source array is denoted as O, and the ignition sources are sorted in sequence.

[0040] S3. Using the combustible gas clouds formed under different gas leakage diffusion numerical simulations with different gas leakage durations under different working conditions in the confined space in step S1, sequentially set different ignition source positions in the simplified ignition source array formed in step S2 to conduct numerical simulations of gas combustion and explosion under different leakage moments, and obtain the time series data of the two-dimensional temperature field, static pressure field, and dynamic pressure field on the typical cross-section (the central cross-section of the confined space) corresponding to different ignition source positions during the gas combustion and explosion process. And for each pixel point in the time series data, extract the maximum value in the time series data of the entire combustion and explosion process respectively to obtain the maximum temperature field, static pressure field, and dynamic pressure field data sets on the typical cross-section during the combustion and explosion process; specifically as follows:

[0041] Using the combustible gas clouds formed under different gas leakage diffusion numerical simulations with different gas leakage durations under different working conditions in the confined space in step 1 (i.e., the combustible gas clouds formed at Δt, 2Δt, 3Δt,..., nΔt moments in the leakage diffusion numerical simulation, where the value range of n is (1, ΔT / Δt)), sequentially set different ignition source positions in the simplified version of the ignition source array formed in step S2 to conduct numerical simulations of gas combustion and explosion under different leakage moments. And save the time series data of the two-dimensional temperature field, static pressure field, and dynamic pressure field corresponding to the typical cross-section during the gas combustion and explosion process. The number of steps of the saved time series data is Te. Denote the pixel points in the length direction of the two-dimensional temperature field, static pressure field, and dynamic pressure field as X, and the number of pixel points in the width direction as Y. Since there are O ignition source positions corresponding to each leakage sample, there are a total of N×ΔT / Δt×O groups of combustion and explosion working conditions. Save the time series data of the temperature field, static pressure field, and dynamic pressure field in the combustion and explosion consequences as tensors of rank (N×ΔT / Δt×O, Te, X, Y) respectively.

[0042] For each pixel point in the time-series data of the two-dimensional temperature field, static pressure field, and dynamic pressure field on the typical cross-section in the obtained combustion and explosion consequences, the maximum value in the time-series data of the entire combustion and explosion process is extracted respectively, to obtain the maximum temperature field, static pressure field, and dynamic pressure field datasets on the typical cross-section during the combustion and explosion process, and the rank of the dataset tensor is (N×ΔT / Δt×O, X, Y).

[0043] S4. Map the segmented time-series datasets of concentration and velocity to the maximum temperature field, static pressure field, and dynamic pressure field datasets after a×Δt time period, to obtain the prediction dataset of the combustion and explosion consequences of combustible gas leakage explosion, and initially establish a prediction model for combustion and explosion consequences; specifically as follows:

[0044] During the prediction process, map the input concentration and velocity datasets to the combustion and explosion consequence datasets after a×Δt time period. The mapping process is to map the concentration and velocity time-series data in each (e×Δt, (e + 1)×Δt) time period range in each leakage and diffusion condition to the maximum temperature field, static pressure field, and dynamic pressure field data formed by ignition and combustion at ((e + 1)×Δt + a×Δt) moment. When a is 0, real-time prediction of combustion and explosion consequences is realized, and when a is greater than 0, ultra-real-time prediction of gas combustion and explosion consequences is realized. The input data of the prediction model includes concentration with a rank of (N×(ΔT / Δt - a), Δt, M) and velocity data with a rank of (N×(ΔT / Δt - a), Δt, Q). The output data of the prediction model are three groups of explosion overpressure, dynamic pressure, and temperature data with a rank of (N×(ΔT / Δt - a)×O, X, Y).

[0045] Use deep learning methods such as long short-term memory method, convolutional neural network or its variants to establish a hybrid deep learning model for predicting the maximum overpressure, temperature, and dynamic pressure fields of combustible gas leakage combustion and explosion based on the time-series data of gas leakage concentration and velocity as input data.

[0046] S5. Use the prediction dataset of combustible gas leakage explosion combustion and explosion consequences to train the combustion and explosion consequence prediction model, and use the loss values of the maximum temperature field, static pressure field, and dynamic pressure field during the combustion and explosion consequence prediction process as the loss function during the training process, and carry out deep learning training to obtain a trained combustion and explosion consequence prediction model; specifically as follows:

[0047] Use the loss values of the maximum temperature field, static pressure field, and dynamic pressure field during the combustion and explosion consequence prediction process as the loss function during the training process. Carry out deep learning training and save the trained combustion and explosion consequence prediction model.

[0048] The above steps S3 - S5 refer to Figure 3 .

[0049] S6. Collect the sequential data of gas leakage concentration and velocity in the confined space and input it into the combustion explosion consequence prediction model. Calculate the maximum value fields of temperature, static pressure, and dynamic pressure during the combustion explosion process based on the sequential data of concentration and velocity through the combustion explosion consequence prediction model.

[0050] During the actual application process, arrange sensors at the same positions of concentration and velocity sensors as in step S1 in the confined space, and number each sensor in the same sorting method as in step S1. Run the sensor group and record the sequential data of the concentration and velocity sensor groups in the confined space in real time. And slice the data in the same way as cutting the sequential data of concentration and velocity sensors in step S1.

[0051] Utilize the trained combustion explosion consequence prediction model. And input the processed sequential data of gas leakage concentration and velocity in the confined space into the combustion explosion consequence prediction model. Calculate the maximum value fields of temperature, static pressure, and dynamic pressure during the combustion explosion process based on the sequential data of concentration and velocity through the combustion explosion consequence prediction model.

[0052] Example

[0053] See Figures 4 to 6 , use the numerical simulation algorithm to carry out three-dimensional numerical simulation of combustible gas leakage under 17 leakage amounts and 9 environmental wind speed conditions in a long and narrow confined space. There are a total of 153 working conditions. The underground long and narrow space is 750m long, 15m wide, and 7.5m high. The calculation simulation duration for each working condition is 600s. During the numerical simulation process, save the sequential data of the distributed combustible gas concentration sensors and air flow velocity sensors at the vault of the confined space during the leakage and diffusion process in the confined space. Set the gas concentration and velocity sensors evenly at the same sensor spacing. The sensor spacing is set to 20m. There are 37 concentration and velocity sensors respectively. The time step of the sensor data is 0.5s, and the sensors are numbered in sequence.

[0054] For the sequential data of concentration and velocity during the 153 groups of 600s combustible gas leakage and diffusion processes, perform data cutting and data processing. Cut the 600s combustible gas leakage data into 30 segments at 20s time intervals respectively. And perform overall normalization on the sequential data of concentration and velocity for each 20s segment.

[0055] Two typical leakage amounts are set in the confined space, and the gas clouds formed after 600 s of leakage diffusion under two environmental wind conditions are numerically simulated for ignition of the uniformly distributed combustible gas clouds by setting a three-dimensional initial ignition source array. The spacing of the ignition source array in the length direction of the confined space is 50 m, the spacing in the height direction of the ignition source array in the confined space is 1 m, and the spacing in the width direction of the ignition source array in the confined space is 1 m. After data processing, the combustion and explosion consequences under different ignition source position conditions are obtained. After data processing, the relative difference between the explosion overpressures formed between all ignition source positions in the length direction of the confined space is greater than 20%. The relative differences between the ignition source positions in the height and width directions of the confined space are both less than 10%. Therefore, the differences between the ignition source positions in the width and height directions of the confined space are ignored. A series of ignition source positions are only set in the length direction of the confined space to simplify the entire ignition source array. The spacing between the ignition source positions is 50 m, and there are a total of 15 ignition source positions.

[0056] For the combustible gas distribution obtained at every 20 s time interval during the leakage and diffusion process of the combustible gas under each working condition, the ignition source positions in the simplified ignition source array are respectively set to conduct numerical simulations of ignition and explosion of the gas cloud under each time interval condition of each working condition. And the temperature field, static pressure field, and dynamic pressure field data during the explosion process are saved. The temperature field, static pressure field, and dynamic pressure field data under each combustion and explosion working condition are the time series data of the representative slices during the explosion process. Here, the representative slice is selected as the cross-section at the center in the width direction of the confined space. The shape of the time series data of the ignition and explosion numerical simulations of the corresponding gas clouds in each leakage process passing through the simplified ignition source array is (15, 500, 50, 15), where the first axis 15 represents different ignition source positions in the simplified ignition source array, the second axis 500 is the number of time steps during the combustion and explosion process, the third axis 50 is the number of pixel points selected in the length direction of the representative slice, and the fourth axis 15 is the number of pixel points selected in the width direction of the representative slice.

[0057] The temperature field, overpressure field, and dynamic pressure field data under each group of combustion and explosion working conditions are processed. For each group of temperature field, overpressure field, and dynamic pressure field time series data with the shape of (15, 500, 50, 15), the maximum values at 500 time steps of the temperature field, overpressure field, and dynamic pressure field of each pixel point on the central cross-section during the combustion and explosion process are selected to obtain the maximum value fields of temperature, static pressure, and dynamic pressure. The shape of the maximum value field data of temperature, static pressure, and dynamic pressure under each combustion and explosion working condition is (15, 50, 15).

[0058] Global data normalization processing is respectively carried out on the data of the maximum value fields of temperature, static pressure, and dynamic pressure obtained in all combustion and explosion working conditions.

[0059] The concentration and velocity segmented data during the leakage and diffusion processes under different working conditions are used as the model input data, and are respectively mapped to the maximum temperature, static pressure, and dynamic pressure field data during the ignition and explosion processes corresponding to different ignition source positions at the end of this time period, with a taken as 0. Along the length direction of the confined space, the input data and output data are respectively mirrored and combined with the original data to achieve data augmentation. In addition, data augmentation is carried out by adding noise. A prediction dataset for the consequences of combustible gas leakage and explosion is obtained.

[0060] Taking the concentration and velocity time-series data segments during the leakage and diffusion processes of combustible gas as the input data together, and taking the maximum overpressure field, temperature field, and dynamic pressure field data during the combustible gas explosion formed during the ignition and explosion processes after the leakage of combustible gas as the output data. Deep learning explosion consequence prediction models are jointly established using algorithms such as long short-term memory, convolutional neural network, and recurrent neural network. See Figure 5 。

[0061] The prediction dataset for the consequences of combustible gas leakage and explosion is divided into a training set, a validation set, and a test set, with the division ratios being 90%, 5%, and 5%. And the training set data is put into the deep learning explosion consequence prediction model for training. During the training process, the average value of the mean square errors of the predicted results of the explosion static pressure field, temperature field, and dynamic pressure field in the test set is used as the evaluation index. It can be considered that the training is completed when the evaluation index during the training process reaches below 0.01, and the trained model is saved. After the training is completed, the predicted values of the maximum temperature field, static pressure field, and dynamic pressure field during the explosion at the ignition source position numbered 8 are compared with the actual values. See Figure 6 。

[0062] During the actual application process, sensors are arranged at the same concentration and velocity sensor point positions as in step S1 in the confined space, and the sensors are numbered in the same sorting method as in step S1. The sensor group is run, and the time-series data of the concentration and velocity sensors in the confined space are recorded in real time. And the data is sliced in the same way as cutting the time-series data of the concentration and velocity sensors in step S1.

[0063] Using the trained explosion consequence prediction model. And the processed time-series data of the gas leakage concentration and velocity in the confined space are input into the explosion consequence prediction model. Through the explosion consequence prediction model, the maximum values of the temperature, static pressure, and dynamic pressure during the explosion based on the time-series data of the concentration and velocity are calculated.

Claims

1. A method for predicting the consequences of confined space deflagration based on the dynamic parameters of the gas leakage process, characterized in that It includes the following steps: S1. Conduct numerical simulation condition design for the leakage and diffusion process of combustible gas in a confined space. Arrange concentration sensors and velocity sensors in the confined space, conduct numerical simulation of gas leakage and diffusion, collect the concentration and velocity segmented time-series data of each sensor at intervals of Δt from the start of gas leakage in each condition, and perform normalization processing on the concentration and velocity segmented time-series data to form concentration and velocity segmented time-series data sets; S2. Set ignition sources at intervals of Δx in the length direction, Δy in the width direction, and Δz in the height direction in the confined space to form a three-dimensional primary ignition source array. Set different ignition source positions on the ignition source array. Use the combustible gas cloud obtained after ΔT duration of gas leakage under typical conditions in step S1 to conduct numerical simulation of gas combustion and explosion under the condition of this combustible gas cloud respectively, and compare the temperature, static pressure, and dynamic pressure curves formed by the numerical simulation results. Simplify the ignition source array according to the results; S3. Use the combustible gas clouds formed under different gas leakage duration conditions in the numerical simulation of gas leakage and diffusion in the confined space under different conditions in step S1. Set different ignition source positions in the simplified ignition source array in sequence to conduct numerical simulation of gas combustion and explosion under different leakage time conditions, and obtain the two-dimensional temperature field, static pressure field, and dynamic pressure field time-series data on the typical cross-section corresponding to different ignition source positions during the gas combustion and explosion process. Extract the maximum value in the time-series data of the entire combustion and explosion process for each pixel point in the time-series data to obtain the maximum temperature field, static pressure field, and dynamic pressure field data sets on the typical cross-section during the combustion and explosion process; S4. Map the concentration and velocity segmented time-series data sets to the maximum temperature field, static pressure field, and dynamic pressure field data sets after a×Δt time period to obtain a combustible gas leakage explosion combustion and explosion consequence prediction data set, and initially establish a combustion and explosion consequence prediction model. When a = 0, realize real-time prediction of combustion and explosion consequences. When a>0, realize ultra-real-time prediction of gas combustion and explosion consequences; S5. Use the combustible gas leakage explosion combustion and explosion consequence prediction data set to train the combustion and explosion consequence prediction model, and use the loss values of the maximum temperature field, static pressure field, and dynamic pressure field during the combustion and explosion consequence prediction process as the loss function during the training process to conduct deep learning training to obtain a trained combustion and explosion consequence prediction model; S6. Collect the gas leakage concentration and velocity time-series data in the confined space and input them into the combustion and explosion consequence prediction model, and calculate the maximum value fields of temperature, static pressure, and dynamic pressure during the combustion and explosion process based on the concentration and velocity time-series data through the combustion and explosion consequence prediction model.

2. The method for predicting the combustion explosion consequence in a confined space based on the dynamic parameters of the gas leakage process according to claim 1, wherein: During the process of simplifying the ignition source array, if the difference between the corresponding gas temperature, static pressure, and dynamic pressure in the calculation results of the combustion and explosion conditions under the ignition source position conditions within a certain spatial range is within the set range, ignore the ignition sources between this ignition source position interval, and select one ignition source in this position area to replace the calculation results.

3. The method for predicting the confined space combustion and explosion consequences based on the dynamic parameters of the gas leakage process according to claim 2, wherein: The set range is less than 20%.

4. The method for predicting the confined space combustion explosion consequence based on the dynamic parameters of the gas leakage process according to claim 3, wherein: In step S1, use numerical simulation or experimental methods to conduct numerical simulation condition design for the leakage and diffusion process of combustible gas in a confined space.

5. The method for predicting the combustion explosion consequence in a confined space based on the dynamic parameters during the gas leakage process according to claim 4, wherein: In the step S1, the concentration and velocity segmented time-series data are enhanced by means of adding noise, changing the relative position between the sensor and the leakage source, or data mirroring.

6. The method for predicting the combustion explosion consequences in a confined space based on the dynamic parameters during the gas leakage process according to claim 5, wherein: In the step S2, for the typical working conditions, at least two levels are taken for each factor of the leakage source conditions and the confined space environmental conditions.

7. The method for predicting the confined space deflagration consequence based on the dynamic parameters of the gas leakage process according to claim 6, wherein: The typical cross-section is the central cross-section of the confined space.

8. The method for predicting the combustion explosion consequence in a confined space based on the dynamic parameters of the gas leakage process according to claim 7, wherein: In the step S2, the temperature, static pressure, and dynamic pressure curves are the data of the central cross-section in the width direction of the confined space.

Images