Explosion range prediction method and device and storage medium

Abstract

The invention provides an explosion range prediction method and device and a storage medium, and relates to the technical field of pipeline design, in order to solve the problem of how to rapidly and accurately predict the explosion range, the explosion range prediction method comprises the following steps: obtaining to-be-predicted data of a detection pipeline; the to-be-predicted data is used for predicting an explosion range when the detection pipeline explodes; inputting the to-be-predicted data into a pre-trained explosion range prediction model to obtain an explosion range prediction result of the detection pipeline; the explosion range prediction model is obtained by training according to a target loss function; the target loss function comprises an explosion range prediction loss, a physical residual loss determined by a related physical formula and an entropy loss. According to the method, the explosion range can be quickly and accurately predicted, and an explosion range prediction result conforming to physical constraints is obtained.

Description

Technical Field

[0001] This application relates to the technical field of pipeline design, and particularly to an explosion range prediction method, device and storage medium. Background Art

[0002] As a key link in the hydrogen energy industry chain, hydrogen energy storage and transportation mainly relies on the transportation of hydrogen through hydrogen pipelines at present. However, hydrogen is flammable and explosive, and the explosion of hydrogen may cause serious damage. Therefore, the safe design and operation of hydrogen pipelines are crucial.

[0003] Accurately predicting the explosion range of hydrogen pipelines can provide an important basis for the safe design of hydrogen pipelines. For example, accurately predicting the explosion range of hydrogen pipelines can not only provide accurate data support for the design of the safe distance between hydrogen pipelines and surrounding facilities, but also provide a reference for the evaluation and renovation of the safe distance of underground hydrogen pipelines built earlier.

[0004] Therefore, how to quickly and accurately predict the explosion range of hydrogen pipelines is a technical problem that needs to be solved urgently at present. Summary of the Invention

[0005] The purpose of this application is to provide an explosion range prediction method, device and storage medium, aiming to solve the problem of how to quickly and accurately predict the explosion range.

[0006] To achieve the above purpose, this application adopts the following technical solutions: In the first aspect, an explosion range prediction method is provided, including: obtaining the data to be predicted of the detection pipeline; the data to be predicted is used to predict the explosion range when the detection pipeline explodes; inputting the data to be predicted into a pre-trained explosion range prediction model to obtain the explosion range prediction result of the detection pipeline; the explosion range prediction model is trained according to the target loss function; the target loss function includes: a first loss function for determining the explosion range prediction loss, a second loss function for determining the physical residual loss, and a third loss function for determining the entropy loss; the explosion range prediction loss is used to represent the difference between the explosion range prediction result predicted by the explosion range prediction model and the real explosion range result; the physical residual loss is used to make the explosion range prediction result predicted by the explosion range prediction model conform to the physical constraints.

[0007] Optionally, the explosion range prediction model is trained in the following way: Obtaining a training sample set; the training sample set includes a plurality of training samples and the label of each training sample; the training sample includes the training data of the sample pipeline; the label of the training sample includes the real explosion range result of the sample pipeline.

[0008] Based on the training sample set, train the original explosion range prediction model to obtain a trained explosion range prediction model.

[0009] Optionally, based on the training sample set, training the original explosion range prediction model to obtain a trained explosion range prediction model includes: Input the training sample set into the explosion range prediction model to obtain the loss value of the current training iteration, and train the explosion range prediction model according to the loss value of the current training iteration until the loss value meets the convergence condition to obtain a trained explosion range prediction model; the loss value of the current training iteration is obtained by the following method: Input the training sample into the explosion range prediction model of the current training iteration to obtain the current explosion range prediction result corresponding to the training sample.

[0010] Determine the current explosion range prediction loss according to the current explosion range prediction result, the label of the training sample, and the first loss function.

[0011] Determine the current physical residual loss according to the training sample, the physical residual equation, and the second loss function.

[0012] Determine the current entropy loss according to the training sample, the entropy regularization term equation, and the third loss function.

[0013] Determine the loss value of the current training iteration according to the current explosion range prediction loss, the current physical residual loss, the current entropy loss, and the target loss function.

[0014] Optionally, determining the loss value of the current training iteration according to the current explosion range prediction loss, the current physical residual loss, the current entropy loss, and the target loss function includes: The target loss function satisfies the following formula:

[0015] Where is the loss value of the current training iteration, λ1 is the weight of the first loss function, W is the current explosion range prediction result, W Real is the true explosion range result of the sample pipeline, λ2 is the weight of the second loss function, f i is the i-th physical residual equation, λ3 is the weight of the third loss function, is the entropy regularization term equation.

[0016] Optionally, obtaining the training sample set includes: Obtain the original data corresponding to the training data of the sample pipeline.

[0017] Preprocess the original data to obtain the data features corresponding to the original data; the preprocessing includes at least one of outlier removal processing, standardization processing, and missing value filling processing.

[0018] Determine the degree of association between any two variable features in the data features.

[0019] Filter the data features according to the degree of association between any two variable features to determine the training sample set.

[0020] Optionally, determining the degree of association between any two variable features in the data features includes: The degree of association between any two variable features satisfies the following formula:

[0021] where r s is the degree of association between any two variable features, d i is the rank difference of the i-th ranked data value in any two variable features, and m is the number of training samples.

[0022] Optionally, filtering the data features according to the degree of association between any two variable features to determine the training sample set includes: When the degree of association between any two variable features is greater than or equal to the first preset threshold, remove the variable feature with the smaller variable feature value, and when the degree of association between any two variable features is less than the first preset threshold, retain any two variable features to obtain the training sample set.

[0023] Optionally, the original data includes at least one of the chemical property data of hydrogen gas transmitted in the sample pipeline, the physical property data of the sample pipeline, the mechanical property data of the soil in the area where the sample pipeline is located, and the energy release data generated when the sample pipeline explodes.

[0024] The data to be predicted includes at least one of the chemical property data of hydrogen gas transmitted in the detection pipeline, the physical property data of the detection pipeline, the mechanical property data of the soil in the area where the detection pipeline is located, and the energy release data generated when the detection pipeline explodes.

[0025] In a second aspect, there is provided an explosion range prediction device, including: a communication unit and a processing unit.

[0026] The communication unit is configured to obtain the data to be predicted of the detection pipeline; the data to be predicted is used to predict the explosion range when the detection pipeline explodes.

[0027] A processing unit for inputting data to be predicted into a pre-trained explosion range prediction model to obtain an explosion range prediction result of a detection pipeline; the explosion range prediction model is trained according to an objective loss function; the objective loss function includes: a first loss function for determining an explosion range prediction loss, a second loss function for determining a physical residual loss, and a third loss function for determining an entropy loss; the explosion range prediction loss is used to represent the difference between the explosion range prediction result predicted by the explosion range prediction model and the true explosion range result; the physical residual loss is used to make the explosion range prediction result predicted by the explosion range prediction model conform to physical constraints.

[0028] In a third aspect, there is provided an explosion range prediction device, including a memory and a processor; the memory is used to store computer execution instructions, and the processor is connected to the memory through a bus; when the explosion range prediction device runs, the processor executes the computer execution instructions stored in the memory, so that the explosion range prediction device executes the explosion range prediction method of the first aspect.

[0029] The explosion range prediction device may be an electronic device or a part of a device in an electronic device, such as a chip system in an electronic device. The chip system is used to support the electronic device to implement the functions involved in the first aspect and any possible implementation manner thereof. For example, it acquires, determines, and sends the data and / or information involved in the above explosion range prediction method. The chip system includes a chip and may also include other discrete devices or circuit structures.

[0030] In a fourth aspect, there is provided a computer-readable storage medium, which includes computer execution instructions. When the computer execution instructions run on a computer, the computer is made to execute the explosion range prediction method of the first aspect.

[0031] In a fifth aspect, there is also provided a computer program product, which includes computer instructions. When the computer instructions run on an explosion range prediction device, the explosion range prediction device is made to execute the explosion range prediction method as described in the first aspect above.

[0032] It should be noted that the above computer instructions may be stored in whole or in part on a computer-readable storage medium. Among them, the computer-readable storage medium may be packaged together with the processor of the explosion range prediction device or separately packaged from the processor of the explosion range prediction device. The embodiments of the present application do not limit this.

[0033] The descriptions of the second aspect, the third aspect, the fourth aspect, and the fifth aspect in this application may refer to the detailed description of the first aspect.

[0034] In the embodiments of the present application, the name of the above explosion range prediction device does not constitute a limitation to the device or functional module itself. In actual implementation, these devices or functional modules may appear under other names. For example, the communication unit may also be referred to as a communication module, etc. As long as the functions of each device or functional module are similar to those of the present application and fall within the scope of the claims of the present application and their equivalent technologies.

[0035] The technical solutions provided by the present application at least bring the following beneficial effects: Based on any of the above aspects, an embodiment of the present application provides an explosion range prediction method, including: First, it is possible to obtain the data to be predicted of the detection pipeline (the data to be predicted is used to predict the explosion range when the detection pipeline explodes). Then, input the data to be predicted into a pre-trained explosion range prediction model to obtain the explosion range prediction result of the detection pipeline. Among them, the pre-trained explosion range prediction model is trained according to a target loss function; the target loss function includes: a first loss function for determining the explosion range prediction loss, a second loss function for determining the physical residual loss, and a third loss function for determining the entropy loss.

[0036] As can be seen from the above, first, the present application can input the data to be predicted into a pre-trained explosion range prediction model to obtain the explosion range prediction result of the detection pipeline, thereby improving the prediction efficiency.

[0037] Second, since the above explosion range prediction model introduces the first loss function, it can determine the difference between the explosion range prediction result predicted by the explosion range prediction model and the true explosion range result, thereby improving the prediction accuracy of the explosion range prediction model.

[0038] Then, since the above explosion range prediction model introduces the second loss function, the explosion range prediction result predicted by the explosion range prediction model conforms to the physical constraints, further improving the prediction accuracy of the explosion range prediction model. Again, since the above explosion range prediction model introduces the third loss function, therefore, by determining the entropy loss, the generalization ability of the explosion range prediction model can be improved and the problem of overfitting of the explosion range prediction model can be prevented.

[0039] For the beneficial effects of the first aspect, second aspect, third aspect, fourth aspect, and fifth aspect in the present application, reference can be made to the analysis of the above beneficial effects, and details are not repeated here. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] To more clearly illustrate the technical solutions of the embodiments of the present application, the following will briefly introduce the drawings required for the description of the embodiments. Obviously, the drawings in the following description are only some embodiments of the present application. For those of ordinary skill in the art, without creative efforts, other drawings can be obtained based on these drawings.

[0041] Figure 1 It is a schematic structural diagram of an explosion range prediction system provided by an embodiment of the present application; Figure 2 It is a schematic hardware structure diagram of an explosion range prediction device provided by an embodiment of the present application; Figure 3 It is a schematic flowchart of an explosion range prediction method provided by an embodiment of the present application; Figure 4 It is a schematic flowchart of another explosion range prediction method provided by an embodiment of the present application; Figure 5 It is a schematic flowchart of another explosion range prediction method provided by an embodiment of the present application; Figure 6 It is a schematic flowchart of another explosion range prediction method provided by an embodiment of the present application; Figure 7 It is a schematic structural diagram of an explosion range prediction model provided by an embodiment of the present application; Figure 8 It is a schematic structural diagram of an explosion range prediction device provided by an embodiment of the present application. Detailed implementation manners

[0042] The following will clearly and completely describe the technical solutions in the embodiments of the present application in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative efforts belong to the scope of protection of the present application.

[0043] In the description of the present application, it should be understood that the orientation or positional relationship indicated by the terms "upper", "lower", "left", "right", "front", "rear", "inner", "outer", etc. is based on the orientation or relative positional relationship shown in the drawings, and is only for the convenience of describing the present application and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore should not be construed as a limitation to the present application. Without special instructions, in the case of satisfying the relative positional relationship shown in the drawings, the above orientation descriptions can be flexibly set during the actual application process.

[0044] The terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the quantity of the indicated technical features. Thus, features defined with "first" and "second" may explicitly or implicitly include one or more of such features. In the description of this application, unless otherwise specified, the meaning of "a plurality of" is two or more.

[0045] In the embodiments of this application, the terms "comprise", "include" or any other variant thereof are intended to cover non-exclusive inclusion, such that a process, article or device comprising a series of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, article or device. Without further limitation, an element defined by the phrase "comprising an..." does not exclude the presence of additional identical elements in the process, article or device comprising such element.

[0046] In the embodiments of this application, words such as "exemplary" or "for example" are used to mean serving as an example, illustration or demonstration. Any embodiment or design described as "exemplary" or "for example" in the embodiments of this application should not be construed as being more preferred or having more advantages than other embodiments or designs. Rather, the use of words such as "exemplary" or "for example" is intended to present the relevant concepts in a specific manner.

[0047] In the description of this specification, specific features, structures, materials or characteristics may be combined in a suitable manner in any one or more embodiments or examples.

[0048] The above is only the specific implementation manner of this application, but the protection scope of this application is not limited thereto. Any person skilled in the art can easily think of changes or substitutions within the technical scope disclosed in this application, and all of them should be covered by the protection scope of this application. Therefore, the protection scope of this application shall be subject to the protection scope of the claims.

[0049] Before introducing the explosion range prediction method of this application in detail, the application scenarios and implementation environments designed in this application will be briefly introduced first.

[0050] First, the application scenarios designed in this application will be briefly introduced.

[0051] As described in the background art, hydrogen energy occupies an important position in the global energy. Therefore, the storage and transportation of hydrogen energy are very important. At present, the storage and transportation of hydrogen energy mainly transport hydrogen through hydrogen pipelines. However, due to the characteristics of hydrogen, once hydrogen explodes, it may cause serious damage. Therefore, the safety design and operation of hydrogen pipelines are crucial. According to the prediction of the explosion range of the pipeline, it can provide an important basis for the safety design of hydrogen pipelines. Therefore, how to quickly and accurately predict the explosion range of hydrogen pipelines is a technical problem that needs to be solved at present.

[0052] In view of the above problems, the embodiment of the present application provides an explosion range prediction method, including: First, the to-be-predicted data of the detection pipeline can be obtained (the to-be-predicted data is used to predict the explosion range when the detection pipeline explodes). Then, the to-be-predicted data is input into a pre-trained explosion range prediction model to obtain the explosion range prediction result of the detection pipeline. Among them, the pre-trained explosion range prediction model is trained according to the target loss function; the target loss function includes: a first loss function for determining the explosion range prediction loss, a second loss function for determining the physical residual loss, and a third loss function for determining the entropy loss.

[0053] As can be seen from the above, first, the present application can input the to-be-predicted data into a pre-trained explosion range prediction model to obtain the explosion range prediction result of the detection pipeline, improving the prediction efficiency.

[0054] Secondly, since the above explosion range prediction model introduces the first loss function, it can determine the difference between the explosion range prediction result predicted by the explosion range prediction model and the true explosion range result, improving the prediction accuracy of the explosion range prediction model.

[0055] Then, since the above explosion range prediction model introduces the second loss function, the explosion range prediction result predicted by the explosion range prediction model conforms to the physical constraints, further improving the prediction accuracy of the explosion range prediction model. Again, since the above explosion range prediction model introduces the third loss function, therefore, by determining the entropy loss, the generalization ability of the explosion range prediction model can be improved and the problem of overfitting of the explosion range prediction model can be prevented.

[0056] The implementation environment of the above explosion range prediction method can be the explosion range prediction system provided by the embodiment of the present application.

[0057] Figure 1 The structural schematic diagram of an explosion range prediction system provided by the embodiment of the present application is shown. As Figure 1 shown, the explosion range prediction system includes: an explosion range prediction device 101 and a data providing device 102.

[0058] Among them, the explosion range prediction device 101 and the data providing device 102 are communicatively connected.

[0059] In practical applications, the explosion range prediction device 101 can be connected to any number of data providing devices 102. For the sake of easy understanding, Figure 1 a case where one explosion range prediction device 101 is connected to one data providing device 102 is taken as an example for illustration.

[0060] In the application embodiment, the data providing device 102 is used to provide the explosion range prediction device 101 with the data to be predicted of the detection pipeline, so that the explosion range prediction device 101 predicts the explosion range when the detection pipeline explodes according to the data to be predicted of the detection pipeline.

[0061] Optionally, the data providing device 102 can also be a data acquisition device, which is used to collect data in aspects such as hydrogen, pipelines, soil, and explosions. The sources of the collected data include but are not limited to experimental site monitoring data, historical failure database data, and data such as combustion and explosion consequences obtained by simulation.

[0062] Optionally, the explosion range prediction device 101 can be a server, a terminal, or other types of electronic devices, and the application embodiment does not limit this.

[0063] Optionally, the above terminal can be a device that provides voice and / or data connectivity to users, a handheld device with a wireless connection function, or other processing devices connected to a wireless modem. The wireless terminal can communicate with one or more core networks via a radio access network (RAN). The wireless terminal can be a mobile terminal, such as a mobile phone (or a "cellular" phone) and a computer with a mobile terminal, or it can also be a portable, pocket-sized, handheld, computer-integrated, or vehicle-mounted mobile device, which exchanges language and / or data with the wireless access network. For example, mobile phones, tablet computers, laptop computers, netbooks, and personal digital assistants (PDAs).

[0064] Optionally, the above server can be a server in a server cluster (composed of multiple servers), a chip in the server, a system-on-chip in the server, or can also be implemented by a virtual machine (VM) deployed on a physical machine, and the application embodiment does not limit this.

[0065] Optionally, the explosion range prediction device 101 and the data providing device 102 can be two independently provided devices, or can be integrated in the same device. When the explosion range prediction device 101 and the data providing device 102 are integrated in the same device, the data providing device 102 can be a storage module (such as a database, etc.) of the explosion range prediction device 101.

[0066] It is easy to understand that when the explosion range prediction device 101 and the data providing device 102 are integrated in the same device, the communication method between the explosion range prediction device 101 and the data providing device 102 is the communication between internal modules of this device. In this case, the communication process between the two is the same as the communication process between the explosion range prediction device 101 and the data providing device 102 when they are independent of each other.

[0067] For ease of understanding, this application takes the explosion range prediction device 101 and the data providing device 102 being independent of each other as an example for description.

[0068] The explosion range prediction device 101 in the explosion range prediction system includes as Figure 2 the included components. The following takes Figure 2 as an example to introduce the hardware structure of the explosion range prediction device 101.

[0069] Figure 2 FIG. is a schematic diagram of the hardware structure of an explosion range prediction device provided by an embodiment of this application. The explosion range prediction device includes a processor 21, a memory 22, a communication interface 23, and a bus 24. The processor 21, the memory 22, and the communication interface 23 can be connected through the bus 24.

[0070] The processor 21 is the control center of the explosion range prediction device, and can be a single processor or a collective term for multiple processing elements. For example, the processor 21 can be a general-purpose central processing unit (CPU), or other general-purpose processors, etc. Among them, the general-purpose processor can be a microprocessor or any conventional processor, etc.

[0071] As an embodiment, the processor 21 can include one or more CPUs, such as Figure 2 the CPU0 and CPU1 shown in

[0072] The memory 22 can be a read-only memory (ROM) or other types of static storage devices that can store static information and instructions, a random access memory (RAM) or other types of dynamic storage devices that can store information and instructions, or can also be an electrically erasable programmable read-only memory (EEPROM), a magnetic disk storage medium, or other magnetic storage devices, or any other medium that can be used to carry or store the desired program code in the form of instructions or data structures and can be accessed by a computer, but is not limited thereto.

[0073] In a possible implementation, the memory 22 can exist independently of the processor 21. The memory 22 can be connected to the processor 21 through the bus 24 and is used to store instructions or program code. When the processor 21 calls and executes the instructions or program code stored in the memory 22, the explosion range prediction method provided in the following embodiments of the present application can be implemented.

[0074] In the embodiments of the present application, for the explosion range prediction device 101 and the data providing device 102, the software programs stored in the memory 22 are different, so the functions implemented by the explosion range prediction device 101 and the data providing device 102 are different. The functions performed by each device will be described in conjunction with the following flowcharts.

[0075] In another possible implementation, the memory 22 can also be integrated with the processor 21.

[0076] The communication interface 23 is used for the explosion range prediction device to connect to other devices through a communication network. The communication network can be an Ethernet, a wireless access network, a wireless local area network (WLAN), etc. The communication interface 23 can include a receiving unit for receiving data and a sending unit for sending data.

[0077] The bus 24 can be an industry standard architecture (ISA) bus, a peripheral component interconnect (PCI) bus, or an extended industry standard architecture (EISA) bus, etc. This bus can be divided into an address bus, a data bus, a control bus, etc. For ease of representation, Figure 2It is represented by only a thick line, but it does not mean that there is only one bus or one type of bus.

[0078] It should be noted that Figure 2 the structure shown in Figure 2 does not constitute a limitation on the explosion range prediction device. Except for the components shown, the explosion range prediction device may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0079] The explosion range prediction method provided by the embodiments of the present application will be introduced in detail below with reference to the accompanying drawings.

[0080] The explosion range prediction method provided by the embodiments of the present application is applied to Figure 1 the explosion range prediction device 101 in the explosion range prediction system shown in Figure 3 As shown, the explosion range prediction method provided by the embodiments of the present application includes: S301. Obtain the data to be predicted of the detection pipeline.

[0081] Among them, the data to be predicted is used to predict the explosion range when the detection pipeline explodes.

[0082] In some embodiments, the data to be predicted of the detection pipeline may include at least one of: chemical property data of hydrogen transmitted in the detection pipeline, physical property data of the detection pipeline, mechanical property data of the soil in the area where the detection pipeline is located, and energy release data generated when the detection pipeline explodes.

[0083] Optionally, the chemical property data of hydrogen transmitted in the detection pipeline may include: heat of combustion of hydrogen , diffusion coefficient D of hydrogen in air H , mass concentration C of hydrogen H , etc.

[0084] Optionally, the physical property data of the detection pipeline may include: pipeline pressure , pipe diameter , pipeline burial depth , etc.

[0085] Optionally, the mechanical property data of the soil in the area where the detection pipeline is located may include: soil density ρ soil , Poisson's ratio , etc.

[0086] Optionally, the energy release data generated when the detection pipeline explodes may include: temperature and the true result of the explosion range, etc.

[0087] Optionally, the explosion range when the detection pipeline explodes may be explosion-related range data such as the width of the explosion pit.

[0088] S302. Input the data to be predicted into the pre-trained explosion range prediction model to obtain the explosion range prediction result of the detection pipeline.

[0089] Specifically, the pre-trained explosion range prediction model can predict the explosion range prediction result of the detection pipeline based on the data to be predicted of the detection pipeline. Therefore, inputting the data to be predicted into the pre-trained explosion range prediction model can obtain the explosion range prediction result of the detection pipeline.

[0090] Among them, the explosion range prediction model is trained according to the target loss function. The target loss function includes: the first loss function for determining the explosion range prediction loss, the second loss function for determining the physical residual loss, and the third loss function for determining the entropy loss. The explosion range prediction loss is used to represent the difference between the explosion range prediction result predicted by the explosion range prediction model and the true explosion range result; the physical residual loss is used to make the explosion range prediction result predicted by the explosion range prediction model conform to the physical constraints.

[0091] Optionally, the entropy loss can be the entropy loss obtained through the entropy regularization term. By introducing the entropy regularization term, the generalization ability of the model can be improved and overfitting can be prevented.

[0092] Optionally, the target loss function can be obtained by weighted summation of the above first loss function, second loss function, and third loss function.

[0093] Optionally, the physical residual loss can be determined by the physical equations related to the explosion. In some embodiments, as Figure 4 shown, the explosion range prediction model is trained in the following manner: S401. Obtain the training sample set.

[0094] Among them, the training sample set includes multiple training samples and the label of each training sample. The training sample includes the training data of the sample pipeline; the label of the training sample includes the true explosion range result of the sample pipeline.

[0095] Specifically, the explosion range prediction device trains the explosion range prediction model. First, it requires the training sample set required for the training of the explosion range prediction model as the training data.

[0096] Optionally, the training sample set is a set of model training data.

[0097] S402. Based on the training sample set, train the original explosion range prediction model to obtain the trained explosion range prediction model.

[0098] In some embodiments, in combination with Figure 4 , such asFigure 5 As shown in Figure 5 , in step S402 above, the method for the explosion range prediction device to train the original explosion range prediction model based on the training sample set to obtain a trained explosion range prediction model specifically includes: S501. Input the training sample set into the explosion range prediction model to obtain the loss value of the current training iteration, and train the explosion range prediction model according to the loss value of the current training iteration until the loss value meets the convergence condition to obtain a trained explosion range prediction model.

[0099] Among them, the loss value of the current training iteration is obtained through the following method: Input the training sample into the explosion range prediction model of the current training iteration to obtain the current explosion range prediction result corresponding to the training sample.

[0100] Determine the current explosion range prediction loss according to the current explosion range prediction result, the label of the training sample, and the first loss function.

[0101] Determine the current physical residual loss according to the training sample, the physical residual equation, and the second loss function.

[0102] Determine the current entropy loss according to the training sample, the entropy regularization term equation, and the third loss function.

[0103] Determine the loss value of the current training iteration according to the current explosion range prediction loss, the current physical residual loss, the current entropy loss, and the target loss function.

[0104] Optionally, the current explosion range prediction loss can be obtained by weighting the squared difference between the current explosion range prediction result and the true explosion range result of the sample pipeline, which is the label of the training sample. The current physical residual loss can be obtained by performing corresponding weight calculation after solving the physical residual equation with the training sample data. The current entropy loss is obtained by weighting the result after entropy regularization. The result of the target loss function is the sum of the current explosion range prediction loss, the current physical residual loss, and the current entropy loss.

[0105] Optionally, the loss value of the current training iteration is the result of the target loss function for this training.

[0106] Optionally, substitute the physical equation into the explosion range prediction model, and calculate the residual through the automatic differentiation technique to determine the physical residual equation. The detailed determination process can refer to the relevant description of the general technology and will not be elaborated here.

[0107] Exemplarily, the physical residual equation mainly includes: Total mass conservation equation of hydrogen-air mixture:

[0108] Hydrogen component conservation equation during combustion:

[0109] Combustion momentum conservation equation:

[0110] Energy conservation equation:

[0111] Soil motion equation:

[0112] Soil constitutive equation:

[0113] Among them, the soil constitutive equation is realized on the premise of a linear elastic model.

[0114] Geometric constraint equation for explosion range result:

[0115] Among them, is the density of the hydrogen-air mixture; is the divergence operator; is the velocity vector of the mixture; C H is the mass concentration of hydrogen; D H is the diffusion coefficient of hydrogen in air, is the thermal conductivity, is the mass consumption rate of hydrogen combustion, is the calorific value of hydrogen combustion; is the pressure of the mixture; is the dynamic viscosity of the mixture; is the acceleration due to gravity; is the time; C P is the specific heat capacity at constant pressure of the mixture; is the temperature of the mixture; ρ soil is the soil density; is the Cauchy stress; F pipe Pipe impact load; is the shear modulus; is the strain tensor, , is the transpose of; is the soil Lame constant, related to the Poisson's ratio ; is the unit tensor; is the predicted result of the explosion range; is the displacement component of the radial coordinate in polar coordinates; is the velocity component along the r-axis; is the radial coordinate; 0 indicates that the coordinate along the z-axis is 0; is the end time of the explosion process.

[0116] In some embodiments, the explosion range prediction device determines the loss value for the current training iteration based on the current explosion range prediction loss, the current physical residual loss, the current entropy loss, and the target loss function, including: The target loss function satisfies the following formula:

[0117] where, is the loss value for the current training iteration, λ1 is the weight of the first loss function, W is the current explosion range prediction result, W Real is the true explosion range result of the sample pipeline, λ2 is the weight of the second loss function, f i is the i-th physical residual equation, λ3 is the weight of the third loss function, is the entropy regularization term equation.

[0118] Specifically, the explosion range prediction device determines whether the loss value of the current explosion range prediction model meets the preset conditions by judging the loss value of the current training iteration. Therefore, the explosion range prediction device needs to determine the loss value of the current training iteration.

[0119] Exemplarily, assuming the threshold is and the current loss value is less than , stop training to obtain the trained explosion range prediction model.

[0120] Optionally, adding entropy loss when calculating the loss value can improve the generalization ability of the model and prevent overfitting. The entropy loss result is determined by the entropy regularization term, and the expression is as follows: Entropy regularization term:

[0121] Entropy production rate:

[0122] where is the entropy regularization term; N is the number of training samples; i is the i-th training sample; is the entropy production rate; is the dynamic viscosity of the mixture; is the velocity vector of the mixture; is the transpose of; is the thermal conductivity; is the temperature; is the gradient operator.

[0123] In some embodiments, in combination with Figure 5 , such as Figure 6 shown, in the above S401, the method for the explosion range prediction device to obtain the training sample set specifically includes: S601. Obtain the original data corresponding to the training data of the sample pipeline.

[0124] In some embodiments, the original data includes at least one of the chemical property data of hydrogen transmitted in the sample pipeline, the physical property data of the sample pipeline, the mechanical property data of the soil in the area where the sample pipeline is located, and the energy release data generated during the explosion of the sample pipeline.

[0125] Optionally, the chemical property data of hydrogen transmitted in the sample pipeline may include the heat of combustion of hydrogen , the diffusion coefficient D of hydrogen in the air H , the mass concentration C of hydrogen H , etc.

[0126] Optionally, the physical property data of the sample pipeline may include the pipeline pressure , the pipe diameter , the pipeline burial depth , etc.

[0127] Optionally, the mechanical property data of the soil in the area where the sample pipeline is located may include the soil density ρ soil , the Poisson's ratio , etc.

[0128] Optionally, the energy release data generated during the explosion of the sample pipeline may include the temperature , the true result of the explosion range, etc.

[0129] S602. Preprocess the original data to obtain the data features corresponding to the original data.

[0130] Among them, the preprocessing includes at least one of outlier removal processing, standardization processing, and missing value filling processing.

[0131] Specifically, before training the explosion range prediction model, the explosion range prediction device needs to preprocess the obtained original data so that the data can adapt to the requirements of the model. Therefore, it is necessary to preprocess the obtained original data. Secondly, the data features are the input data of the explosion range prediction model. Therefore, it is necessary to obtain the data features corresponding to the original data.

[0132] Exemplarily, preprocessing the original data includes at least one of outlier removal processing, standardization processing, and missing value filling processing.

[0133] Exemplarily, outlier removal processing is performed on the original data to obtain the data after removing outliers, including: The box plot method is used to identify outliers in the original data, and its specific identification rules are as follows:

[0134] Among them, x is the original data, Q1 is the upper quartile, Q3 is the lower quartile, and IQR is the interquartile range.

[0135] As can be seen from the above formula, when the original data exceeds the outlier range determined by the upper quartile Q1, the lower quartile Q3, and the interquartile range IQR (less than the value Q1 - 1.5 IQR or greater than the value Q3 + 1.5 IQR), the original data will be identified as an outlier; otherwise, it will be identified as a normal value.

[0136] Exemplarily, normalization processing is performed on the data after removing outliers, including: Select data without missing values from the data after removing outliers. The Z-transformation score (Z-score) method is used to normalize the data, and the formula is as follows:

[0137] Among them, X is the data after removing outliers, and X ＇ is the data after normalization processing, μ is the mean, and σ is the standard deviation.

[0138] Exemplarily, missing value filling processing is performed on the normalized data, including: For the data containing missing values, the K-nearest neighbor algorithm is used to calculate the distance between the sample points with missing values and other sample points in the complete dataset. The formula is as follows:

[0139] Select the K nearest sample points, and then use the mean of the missing feature values corresponding to the selected K nearest sample points to fill the missing values in the data. The formula is as follows:

[0140] Among them, is the distance between the sample point with missing values and other sample points in the complete dataset, is the value of sample 1 of feature x, is the value of sample 2 of feature x, is the value of sample 1 of feature y, is the value of sample 2 of feature y, is the missing value, is the total number of the selected most adjacent sample points, is the value of the i-th adjacent sample point, where i is the i-th most adjacent sample point.

[0141] It can be seen that the explosion range prediction device performs missing value filling on the standardized data, can handle the blank values in the data, and ensures the integrity of the data.

[0142] Secondly, when constructing the explosion range prediction model, the explosion range prediction device embeds the physical equations related to the explosion into the explosion range prediction model as physical residual equations, which can use the prior physical laws to fill the blanks in the sparse area of the experimental data and reduce the dependence on the high-cost and high-risk explosion experimental data.

[0143] S603. Determine the degree of association between any two variable features in the data features.

[0144] Specifically, there are multiple variable features in the data features. The explosion range prediction device needs to determine whether to remove the variable features by the degree of association between any two variable features, so as to realize the screening of the data features. Therefore, the explosion range prediction device needs to determine the degree of association between the variable features.

[0145] Optionally, the degree of association between variable features can also be called correlation.

[0146] Optionally, the data features may include: geometric parameters, experimental working condition parameters, soil characteristics near the pipeline, medium characteristics, and the true results of explosion range prediction, etc.

[0147] Optionally, the geometric parameters may include variable features such as time t, cylindrical coordinates (r, z), pipe diameter D, pipeline burial depth H, etc.

[0148] Optionally, the experimental working condition parameters may include variable features such as pipeline pressure P.

[0149] Optionally, the soil characteristics near the pipeline may include variable features such as soil density ρ soil and Poisson's ratio ν, etc.

[0150] Optionally, the medium characteristics may include hydrogen combustion heat , diffusion coefficient D of hydrogen in air H , mass concentration C of hydrogen H , etc. variable features.

[0151] S604. Screen the data features according to the degree of association between any two variable features, and determine the training sample set.

[0152] In some embodiments, the explosion range prediction device determines the degree of association between any two variable features in the data features, including: The degree of association between any two variable features satisfies the following formula:

[0153] where r s is the degree of association between the any two variable features, d i is the rank difference of the i-th rank data value among the any two variable features, and m is the number of training samples.

[0154] Optionally, the rank difference of the -th rank data value represents the difference between the i-th rank values after sorting the two variable feature values. The specific sorting process can refer to the relevant descriptions in the prior art and will not be elaborated here.

[0155] Exemplarily, assume that any two variable features are variable feature and variable feature , then the rank difference of the i-th rank data value is the difference between the i-th rank value of variable feature and the i-th rank value of variable feature .

[0156] In some embodiments, the explosion range prediction device screens the data features according to the degree of association between any two variable features to determine a training sample set, including: In the case where the degree of association between any two variable features is greater than or equal to a first preset threshold, the variable feature with a smaller variable feature value is removed, and in the case where the degree of association between any two variable features is less than the first preset threshold, any two variable features are retained to obtain a training sample set.

[0157] It should be noted that when the degree of association between any two variable features is less than the first preset threshold, specifically including: when the degree of association between any two variable features is greater than or equal to a second preset threshold and less than the first preset threshold, a prompt message can be output and manual screening can be performed based on experience; when the degree of association between any two variable features is less than the second preset threshold, any two variable features are retained. Wherein, the first preset threshold is greater than the second preset threshold.

[0158] Optionally, the above explosion range prediction model can be physics-informed neural networks (PINNs). Physics-informed neural networks are an innovative deep neural network architecture that can solve supervised learning tasks. Their core advantage lies in the ability to seamlessly integrate physical prior knowledge into the network structure, thereby imposing strict physical constraints on the network during training. With the help of automatic differentiation technology, PINNs can accurately incorporate partial differential equations and their definite solution conditions into the loss function, ensuring that the model output is strictly consistent with physical laws. Compared with traditional numerical methods, the numerical solutions obtained by PINNs not only have higher accuracy but also are closer to the real physical process, can better capture the essential characteristics of physical phenomena, and have unique advantages in dealing with complex geometric shapes and multi-physics field coupling problems.

[0159] It should be noted that the prediction of the explosion range is essentially a non-linear dynamics problem of strong coupling of multiple physical fields, involving the interaction mechanism of four dimensions: the chemical properties of hydrogen, the physical properties of the pipeline, the mechanical properties of the soil, and the energy release of the explosion, with a very complex non-linear mapping relationship, making it difficult to accurately predict with traditional deterministic mathematical models. In the general technology for predicting the explosion range results, numerical simulation and machine learning are relatively common methods. Among them, the numerical simulation method mainly relies on dynamic software such as flame acceleration simulation (FLACS) and computational fluid dynamics (CFD) to establish a detailed physical model to simulate the explosion process, thereby predicting the explosion range results. This method has high theoretical accuracy, but requires a large amount of computing resources and time, and has high requirements for the accuracy of model parameters. The machine learning method can automatically learn the patterns and laws in the data by analyzing a large amount of historical data, and has great advantages in dealing with complex non-linear relationships. However, most machine learning models require a large amount of labeled data. In the case of scarce data or high data annotation costs, the performance of the model may be severely affected, and key physical mechanisms may be ignored.

[0160] It can be seen that the explosion range prediction model constructed by the explosion range prediction device can also be called a physics-informed neural network model.

[0161] First, the explosion range prediction device can predict the explosion range result by constructing an explosion range prediction model and training the model based on data collected from aspects such as hydrogen, pipelines, soil, and explosions. This solves the problem that it is difficult for mathematical models to predict explosion range results and avoids the cumbersome grid processing process in traditional numerical methods in the general technology, improving the efficiency of calculating the explosion range prediction result and the accuracy of the explosion range prediction result.

[0162] Second, the explosion range prediction device embeds physical equations related to explosions into the explosion range prediction model through automatic differentiation technology, and fills the data blank areas through physical regularization terms (such as mass conservation residuals and energy constraint terms), greatly reducing the dependence on experimental data, further improving the accuracy of the explosion range prediction result, and also being able to efficiently solve complex physical processes, solving the problems in the general technology that machine learning methods have a high dependence on data and may ignore physical mechanisms.

[0163] Exemplarily, Figure 7 FIG. Figure 7 shows a schematic structural diagram of an explosion range prediction model provided by an embodiment of the present application. First, a multi-layer perceptron (MLP) is selected as the basic architecture to construct a fully connected neural network. The fully connected neural network includes an input layer, a hidden layer, and an output layer.

[0164] The input layer is used to input a training sample set, including several neurons for data such as geometric parameters, working condition parameters, soil characteristics, and hydrogen energy characteristics (for example Figure 7 in ).

[0165] The hidden layer (which can also be called a multi-layer hidden layer) includes weight parameters, bias parameters, and non-linear activation functions, etc.

[0166] The output layer is used to output the calculation result of the fully connected neural network, that is, the explosion range prediction result.

[0167] Then, the present application can also embed physical equations related to pipeline explosions into the fully connected neural network through automatic differentiation technology as physical residual equations, thereby constructing an original explosion range prediction model.

[0168] During the training process of the original explosion range prediction model, the data prediction variance can be determined through the explosion range prediction result output by the fully connected neural network, thereby constructing a first loss function of the target loss function.

[0169] Secondly, the variance of the partial differential equation determined by the physical residual equation can also be used to construct the second loss function of the target loss function.

[0170] Thirdly, the third loss function of the target loss function can also be constructed by the entropy regularization term equation.

[0171] Subsequently, the target loss function can be constructed by the first loss function, the second loss function, and the third loss function.

[0172] Then, it can be determined whether the loss value converges according to the loss value output by the target loss function.

[0173] If so, a trained explosion range prediction model is obtained.

[0174] Optionally, the trained explosion range prediction model can be the optimal model parameters of the trained explosion range prediction model.

[0175] If not, the explosion range prediction model continues to be trained until the loss value meets the convergence condition.

[0176] In some embodiments, the explosion range prediction device can also test the explosion range prediction model and evaluate the performance of the explosion range prediction model to obtain the test result of the explosion range prediction model. In this case, the explosion range prediction device can use the test set to test the trained explosion range prediction model and evaluate the performance of the model.

[0177] Among them, the test set S contains , N test sample data.

[0178] Optionally, the training set and the test set can be obtained by dividing the collected original data according to a ratio of 7:3.

[0179] The explosion range prediction device can calculate the prediction error on the test set and evaluate the accuracy and generalization ability of the model through the comparison result between the prediction result of the explosion range prediction model and the real explosion range result.

[0180] Optionally, the explosion range prediction device can test and evaluate the performance of the model through methods such as the receiver operating characteristic (ROC) curve and recall rate. And evaluate the accuracy and generalization ability of the model by calculating aspects such as the precision and accuracy of the model.

[0181] The above mainly introduced the solution provided by the embodiments of the present application from the perspective of methods. To implement the above functions, it includes the corresponding hardware structures and / or software modules for executing each function. Those skilled in the art should easily realize that, combining the units and algorithm steps of each example described in the embodiments disclosed herein, the present application can be implemented in the form of hardware or a combination of hardware and computer software. Whether a certain function is executed in the way of hardware or computer software driving hardware depends on the specific application and design constraints of the technical solution. Professional technicians can use different methods to implement the described functions for each specific application, but such implementation should not be considered to exceed the scope of the present application.

[0182] The embodiments of the present application can divide the function modules of the explosion range prediction device according to the above method examples. For example, each function module can be divided corresponding to each function, or two or more functions can be integrated into one processing module. The above integrated modules can be implemented in the form of hardware or in the form of software function modules. Optionally, the division of modules in the embodiments of the present application is illustrative, only a logical function division, and there can be other division methods in actual implementation.

[0183] As Figure 8 shown, it is a schematic structural diagram of an explosion range prediction device provided by an embodiment of the present application. Figure 8 The explosion range prediction device shown includes: a communication unit 801 and a processing unit 802.

[0184] The communication unit 801 is used to obtain the data to be predicted of the detection pipeline.

[0185] Among them, the data to be predicted is used to predict the explosion range when the detection pipeline explodes.

[0186] The processing unit 802 is used to input the data to be predicted into a pre-trained explosion range prediction model to obtain the explosion range prediction result of the detection pipeline.

[0187] Among them, the explosion range prediction model is trained according to the target loss function; the target loss function includes: a first loss function for determining the explosion range prediction loss, a second loss function for determining the physical residual loss, and a third loss function for determining the entropy loss; the explosion range prediction loss is used to represent the difference between the explosion range prediction result predicted by the explosion range prediction model and the real explosion range result; the physical residual loss is used to make the explosion range prediction result predicted by the explosion range prediction model conform to the physical constraints.

[0188] Optionally, the explosion range prediction model is trained in the following manner: The processing unit 802 is further used to obtain a training sample set.

[0189] Among them, the training sample set includes multiple training samples and the labels of each training sample. The training samples include the training data of the sample pipeline; the labels of the training samples include the true explosion range results of the sample pipeline.

[0190] The processing unit 802 is further configured to train the original explosion range prediction model based on the training sample set to obtain a trained explosion range prediction model.

[0191] Optionally, the processing unit 802 is specifically configured to: Input the training sample set into the explosion range prediction model to obtain the loss value of the current training iteration, and train the explosion range prediction model according to the loss value of the current training iteration until the loss value meets the convergence condition to obtain a trained explosion range prediction model; the loss value of the current training iteration is obtained in the following manner: Input the training sample into the explosion range prediction model of the current training iteration to obtain the current explosion range prediction result corresponding to the training sample.

[0192] Determine the current explosion range prediction loss according to the current explosion range prediction result, the label of the training sample, and the first loss function; determine the current physical residual loss according to the training sample, the physical residual equation, and the second loss function; determine the current entropy loss according to the training sample, the entropy regularization term equation, and the third loss function; determine the loss value of the current training iteration according to the current explosion range prediction loss, the current physical residual loss, the current entropy loss, and the target loss function.

[0193] Optionally, the processing unit 802 is specifically configured to: The target loss function satisfies the following formula:

[0194] Among them, is the loss value of the current training iteration, λ1 is the weight of the first loss function, W is the current explosion range prediction result, W Real is the true explosion range result of the sample pipeline, λ2 is the weight of the second loss function, f i is the i-th physical residual equation, λ3 is the weight of the third loss function, is the entropy regularization term equation.

[0195] Optionally, the processing unit 802 is specifically configured to: Obtain the original data corresponding to the training data of the sample pipeline.

[0196] Preprocess the original data to obtain the data features corresponding to the original data; the preprocessing includes at least one of outlier removal processing, normalization processing, and missing value filling processing.

[0197] Determine the degree of association between any two variable features among the data features.

[0198] Filter the data features according to the degree of association between any two variable features, and determine the training sample set.

[0199] Optionally, the processing unit 802 is specifically configured to determine the degree of association between any two variable features among the data features, including: The degree of association between any two variable features satisfies the following formula:

[0200] where r s is the degree of association between any two variable features, d i is the rank difference of the i-th ranked data value among any two variable features, and m is the number of training samples.

[0201] Optionally, the processing unit 802 is specifically configured to: When the degree of association between any two variable features is greater than or equal to the first preset threshold, eliminate the variable feature with a smaller variable feature value, and when the degree of association between any two variable features is less than the first preset threshold, retain any two variable features to obtain the training sample set.

[0202] Optionally, the original data includes at least one of the chemical property data of hydrogen gas transmitted in the sample pipeline, the physical property data of the sample pipeline, the mechanical property data of the soil in the area where the sample pipeline is located, and the energy release data generated when the sample pipeline explodes.

[0203] The data to be predicted includes at least one of the chemical property data of hydrogen gas transmitted in the detection pipeline, the physical property data of the detection pipeline, the mechanical property data of the soil in the area where the detection pipeline is located, and the energy release data generated when the detection pipeline explodes.

[0204] The embodiment of the present application further provides a computer-readable storage medium, and the computer-readable storage medium includes computer-executable instructions. When the computer-executable instructions run on a computer, the computer is caused to execute the explosion range prediction method provided in the above embodiment.

[0205] The embodiments of the present application also provide a computer program product. This computer program product can be directly loaded into a memory and contains software code. After being loaded and executed by a computer, this computer program product can implement the explosion range prediction method provided in the above embodiments. Finally, it should be noted that: the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit them. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that they can still modify or equivalently replace the technical solutions of the present invention, and these modifications or equivalent replacements do not make the modified technical solutions deviate from the spirit and scope of the technical solutions of the present invention.

[0206] For the system provided in the above embodiments, only the division of the above functional modules is used as an example for illustration. In actual applications, the above functions can be assigned to different functional modules according to needs, that is, the modules or steps in the embodiments of the present invention can be further decomposed or combined. For example, the modules in the above embodiments can be combined into one module, or further split into multiple sub-modules to complete all or part of the functions described above. For the names of the modules and steps involved in the embodiments of the present invention, they are only used to distinguish each module or step and are not regarded as an improper limitation of the present invention.

[0207] Those skilled in the art should be able to realize that the modules and method steps of each example described in combination with the embodiments disclosed herein can be implemented by electronic hardware, computer software, or a combination of the two. The programs corresponding to the software modules and method steps can be placed in a random access memory, internal memory, read-only memory, electrically erasable programmable read-only memory, register, hard disk, removable disk, compact disc read-only memory (CD-ROM), or any other form of storage medium well-known in the technical field. To clearly illustrate the interchangeability of electronic hardware and software, the composition and steps of each example have been generally described according to functions in the above description. Whether these functions are executed in the form of electronic hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered to exceed the scope of the present invention.

Claims

1. An explosion range prediction method, characterized in that, Including: Obtain the data to be predicted for the detection pipeline; The data to be predicted is used to predict the explosion range when the detection pipeline explodes; Input the data to be predicted into a pre-trained explosion range prediction model to obtain the explosion range prediction result of the detection pipeline; The explosion range prediction model is trained according to a target loss function; the target loss function includes: a first loss function for determining the explosion range prediction loss, a second loss function for determining the physical residual loss, and a third loss function for determining the entropy loss; the explosion range prediction loss is used to represent the difference between the explosion range prediction result predicted by the explosion range prediction model and the true explosion range result; the physical residual loss is used to make the explosion range prediction result predicted by the explosion range prediction model conform to physical constraints.

2. The method according to claim 1, wherein The explosion range prediction model is trained through the following method: Obtain a training sample set; the training sample set includes multiple training samples and the label of each training sample; the training sample includes the training data of the sample pipeline; the label of the training sample includes the true explosion range result of the sample pipeline; Based on the training sample set, train the original explosion range prediction model to obtain the trained explosion range prediction model.

3. The method according to claim 2, wherein The training the original explosion range prediction model based on the training sample set to obtain the trained explosion range prediction model includes: Input the training sample set into the explosion range prediction model to obtain the loss value of the current training iteration, and train the explosion range prediction model according to the loss value of the current training iteration until the loss value meets the convergence condition to obtain the trained explosion range prediction model; the loss value of the current training iteration is obtained through the following method: Input the training sample into the explosion range prediction model of the current training iteration to obtain the current explosion range prediction result corresponding to the training sample; Determine the current explosion range prediction loss according to the current explosion range prediction result, the label of the training sample, and the first loss function; Determine the current physical residual loss according to the training sample, the physical residual equation, and the second loss function; Determine the current entropy loss according to the training sample, the entropy regularization term equation, and the third loss function; Determine the loss value of the current training iteration according to the current explosion range prediction loss, the current physical residual loss, the current entropy loss, and the target loss function.

4. The method according to claim 3, characterized in that, The determining the loss value of the current training iteration according to the current explosion range prediction loss, the current physical residual loss, the current entropy loss, and the target loss function includes: The target loss function satisfies the following formula: Among them, is the loss value of the current training iteration, λ1 is the weight of the first loss function, W is the predicted explosion range result of the current, and W Real is the true explosion range result of the sample pipeline, λ2 is the weight of the second loss function, and f i is the i-th physical residual equation, λ3 is the weight of the third loss function, is the entropy regularization term equation.

5. The method according to claim 2, wherein The obtaining the training sample set includes: Obtain the original data corresponding to the training data of the sample pipeline; Preprocess the original data to obtain the data features corresponding to the original data; the preprocessing includes at least one of outlier removal processing, standardization processing, and missing value filling processing; Determine the correlation degree between any two variable features in the data features; Screen the data features according to the degree of association between any two variable features, and determine the training sample set.

6. The method according to claim 5, wherein Determining the degree of association between any two variable features among the data features includes: The degree of association between any two variable features satisfies the following formula: where r s is the degree of association between any two variable features, d i is the rank difference of the i-th ranked data value among any two variable features, and m is the number of the training samples.

7. The method according to claim 5, characterized in that Screening the data features according to the degree of association between any two variable features to determine the training sample set includes: When the degree of association between any two variable features is greater than or equal to a first preset threshold, eliminate the variable feature with a smaller variable feature value, and when the degree of association between any two variable features is less than the first preset threshold, retain the any two variable features to obtain the training sample set.

8. The method according to claim 5, characterized in that, The original data includes at least one of the chemical property data of hydrogen transmitted in the sample pipeline, the physical property data of the sample pipeline, the mechanical property data of the soil in the area where the sample pipeline is located, and the energy release data generated when the sample pipeline explodes; The data to be predicted includes at least one of the chemical property data of hydrogen transmitted in the detection pipeline, the physical property data of the detection pipeline, the mechanical property data of the soil in the area where the detection pipeline is located, and the energy release data generated when the detection pipeline explodes.

9. An explosion range prediction device, characterized in that, Includes: A processor and a memory; wherein, the memory is used to store one or more programs, and the one or more programs include computer execution instructions. When the explosion range prediction device runs, the processor executes the computer execution instructions stored in the memory, so that the explosion range prediction device executes the method according to any one of claims 1 to 8.

10. A computer-readable storage medium, characterized in that, When the computer execution instructions stored in the computer-readable storage medium are executed by the processor of the explosion range prediction device, the explosion range prediction device can execute the method according to any one of claims 1 to 8.

Images