Integrated intelligent ventilation and dust reduction method and device for spiral tunnel

Abstract

The embodiment of the invention provides an integrated intelligent ventilation and dust fall method and device for a spiral tunnel. Relates to the technical field of tunnel engineering. The method comprises the following steps: constructing a physical model according to a press-in ventilation spiral tunnel structure and determining a constraint code; a deep learning network model and a CFD model are adopted to simulate the spiral tunnel structure, the deviation of the deep learning network model on hydrodynamic force is obtained according to the simulation result of the CFD model and field observation data, and the deep learning network model is adjusted according to the deviation; constraint codes are added into the adjusted deep learning network model for sensitivity analysis, and key parameters influencing ventilation and dust falling of the spiral tunnel are obtained; and monitoring the key parameters in real time, adjusting the corresponding key parameters according to the dust concentration predicted value of each monitoring point output by the deep learning network model added with the constraint code, optimizing the ventilation rate and enabling the dust concentration of each monitoring point to meet the standard. Therefore, the adaptability of the model in a complex environment and the ventilation and dust reduction processing efficiency are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of tunnel engineering, and in particular to an integrated intelligent ventilation and dust reduction method and device for a spiral tunnel. Background Art

[0002] In recent years, spiral tunnels have gradually appeared in my country. This type of tunnel can reach a higher height in a shorter distance, greatly shortening the distance between two places, and effectively reducing the overall slope of the highway, improving the safety of vehicle driving. In spiral tunnel construction projects, especially those that pass through hard rock areas, the drilling and blasting method is widely used because of its cost-effectiveness. This traditional method involves drilling, charging and detonating. Although it has significant advantages in excavation efficiency, it is also accompanied by the generation of a large amount of harmful dust. These dusts not only pose a threat to the environment, but may also endanger the health of tunnel construction workers and increase the risk of pneumoconiosis. This problem is particularly significant in long-distance tunnels. The particularity of the tunnel construction environment, such as changing cross-sectional dimensions, diverse geological conditions and technical limitations, all pose challenges to the control of dust diffusion. However, dust control is key to ensuring the health of workers and maintaining the quality of the construction environment.

[0003] At present, the ventilation method after tunnel drilling and blasting mainly relies on the forced ventilation method, which is a method of mechanically forcing fresh air into underground caverns or mines to improve the working environment and protect the health of workers. This ventilation method uses ventilation machinery installed outside the tunnel to deliver fresh air to the working surface through ventilation ducts, and at the same time discharges dirty air to the ground, thereby achieving the purpose of diluting and removing harmful gases. Under the existing forced ventilation conditions, there is obvious dust retention in certain areas of the ventilation duct, especially at the top and corners, which makes it difficult to quickly and effectively reduce the dust concentration in the tunnel after blasting. This not only limits the efficiency of tunnel construction, but also endangers the occupational health of construction workers. In actual construction, the dust diffusion after the application of the forced ventilation method will be affected by many factors. Therefore, a method that can accurately predict dust diffusion is needed.

[0004] Existing methods for predicting tunnel dust diffusion include CFD simulation, but this method often requires complex parameter settings and grid division, and it is difficult to effectively integrate real-time monitoring data. In addition, in the process of model establishment and prediction, traditional deep learning methods do not fully consider the relevant physical laws, making it difficult to ensure prediction accuracy. Summary of the invention

[0005] The present invention provides a method and device for integrated intelligent ventilation and dust reduction in a spiral tunnel, equipment and storage medium.

[0006] According to a first aspect of the present disclosure, a spiral tunnel integrated intelligent ventilation and dust reduction method is provided. The method comprises:

[0007] Construct a physical model according to the spiral tunnel structure of forced ventilation, and determine the constraint encoding according to the physical model;

[0008] Use a deep learning network model and a CFD model to simulate the spiral tunnel structure of forced ventilation. According to the simulation results of the CFD model and the on-site observation data, obtain the deviation of the deep learning network model in terms of hydrodynamic force, and adjust the deep learning network model according to the deviation;

[0009] Add the constraint encoding to the adjusted deep learning network model for sensitivity analysis to obtain the key parameters affecting the ventilation and dust reduction of the spiral tunnel;

[0010] Monitor the key parameters in real time, and adjust the corresponding key parameters according to the predicted dust concentration values corresponding to each monitoring point output by the deep learning network model with the constraint encoding added, optimize the ventilation rate, and make the dust concentration corresponding to each monitoring point meet the preset standard.

[0011] In some realizable ways of the first aspect, the constructing a physical model according to the spiral tunnel structure of forced ventilation includes:

[0012] Construct a physical model according to the spiral tunnel structure of forced ventilation by using the Navier-Stokes equation, the turbulent kinetic energy equation, the turbulent energy dissipation rate equation, and the component mass conservation equation.

[0013] In some realizable ways of the first aspect, the using a deep learning network model and a CFD model to simulate the spiral tunnel structure of forced ventilation includes:

[0014] Use a deep learning network model and a CFD model to simulate the hydrodynamic response and temperature distribution of the spiral tunnel of forced ventilation under different ventilation environments and emergency environments.

[0015] In some realizable ways of the first aspect, the ventilation environments include the ventilation environments at the initial construction stage and when the ventilation equipment is enabled, and the ventilation environment under complex geological conditions during the mid-construction stage;

[0016] The emergency environments include fire and failure of the ventilation and dust reduction equipment.

[0017] In some realizable ways of the first aspect, the adjusting the deep learning network model according to the deviation includes:

[0018] Optimize the structure, hyperparameters, and training strategy of the deep learning network model according to the deviation; wherein,

[0019] The optimizing the structure of the deep learning network model includes increasing or decreasing the number of network layers and adjusting the activation function;

[0020] The hyperparameters for optimizing the deep learning network model include adjusting the learning rate and the batch size;

[0021] The training strategy for optimizing the deep learning network model includes introducing a regularization algorithm and changing the weight allocation of the loss function.

[0022] In some realizable ways of the first aspect, the method further includes:

[0023] Changing the weight allocation of the loss function of the deep learning network model according to the hydrodynamic response and temperature distribution simulation results of the CFD model and the on-site observation data to optimize multiple objectives; wherein,

[0024] The multiple objectives include the ventilation rate, the temperature control strategy, and the dust concentration at each monitoring point.

[0025] In some realizable ways of the first aspect, adding a constraint encoding to the adjusted deep learning network model for sensitivity analysis includes:

[0026] Adding a constraint encoding to the adjusted deep learning network model, and using the forward propagation algorithm to predict the influence of the environmental parameters of the spiral tunnel on the output value of the adjusted deep learning network model;

[0027] Using the automatic differentiation method to sort the influence of the predicted environmental parameters of the spiral tunnel on the output value of the adjusted deep learning network model in descending order.

[0028] In some realizable ways of the first aspect, the key parameters include:

[0029] The position and number of ventilation openings of the spiral tunnel, the ventilation wind speed, the spatial layout of the construction area of the spiral tunnel, and the external climate conditions.

[0030] According to the second aspect of the present disclosure, there is provided an integrated intelligent ventilation and dust reduction device for a spiral tunnel. The device includes:

[0031] A constraint encoding acquisition module, configured to construct a physical model according to the structure of the spiral tunnel with forced ventilation, and determine the constraint encoding according to the physical model;

[0032] A deep learning network model adjustment module, configured to simulate the structure of the spiral tunnel with forced ventilation by using a deep learning network model and a CFD model, obtain the deviation of the deep learning network model in terms of hydrodynamic force according to the simulation results of the CFD model and the on-site observation data, and adjust the deep learning network model according to the deviation;

[0033] A key parameter acquisition module, which is used to perform sensitivity analysis by adding constraint encoding to the adjusted deep learning network model, and acquire the key parameters affecting the ventilation and dust reduction of the spiral tunnel;

[0034] A key parameter optimization module, which is used to monitor the key parameters in real time, and adjust the corresponding key parameters according to the predicted dust concentration values corresponding to each monitoring point output by the deep learning network model with constraint encoding added, optimize the ventilation rate, and make the dust concentration corresponding to each monitoring point meet the preset standard.

[0035] According to a third aspect of the present disclosure, an electronic device is provided. The electronic device includes: at least one processor; and a memory communicatively connected to the at least one processor; the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to execute the method as described above.

[0036] According to a fourth aspect of the present disclosure, a non-transitory computer-readable storage medium storing computer instructions is provided, and the computer instructions are used to cause a computer to execute the method as described above.

[0037] In the present disclosure, a physical model is constructed according to the spiral tunnel structure of forced ventilation and the constraint encoding is determined; the spiral tunnel structure is simulated by using a deep learning network model and a CFD model, the deviation of the deep learning network model in terms of hydrodynamic force is acquired according to the simulation results of the CFD model and the on-site observation data, and the deep learning network model is adjusted according to the deviation; sensitivity analysis is performed by adding constraint encoding to the adjusted deep learning network model, and the key parameters affecting the ventilation and dust reduction of the spiral tunnel are acquired; the key parameters are monitored in real time, and the corresponding key parameters are adjusted according to the predicted dust concentration values corresponding to each monitoring point output by the deep learning network model with constraint encoding added, optimize the ventilation rate, and make the dust concentration at each monitoring point meet the preset standard. In this way, the research on the tunnel dust diffusion problem by the deep learning network model (PINN model) is supplemented, the model establishment process is simplified, the adaptability of the model in a complex flow environment and the ventilation and dust reduction treatment efficiency are improved, and the model can accurately predict the tunnel dust concentration at a certain moment under limited data.

[0038] It should be understood that the content described in the summary of the invention section is not intended to limit the key or important features of the embodiments of the present disclosure, nor to limit the scope of the present disclosure. Other features of the present disclosure will become easily understood through the following description. Brief Description of the Drawings

[0039] In conjunction with the accompanying drawings and with reference to the following detailed description, the above and other features, advantages, and aspects of the embodiments of the present disclosure will become more apparent. The accompanying drawings are used to better understand the solution and do not limit the present disclosure. In the drawings, the same or similar reference numerals denote the same or similar elements, where:

[0040] Figure 1 shows a flowchart of a spiral tunnel integrated intelligent ventilation and dust reduction method provided by an embodiment of the present disclosure;

[0041] Figure 2 shows a comparison chart of the predicted dust concentration value obtained by a deep learning network model at a certain monitoring point in a forced ventilation spiral tunnel provided by an embodiment of the present disclosure and the actual monitoring value at the corresponding moment;

[0042] Figure 3 shows a structural diagram of a spiral tunnel integrated intelligent ventilation and dust reduction device provided by an embodiment of the present disclosure;

[0043] Figure 4 shows a structural diagram of an exemplary electronic device capable of implementing the embodiments of the present disclosure. Detailed Embodiments

[0044] To make the objectives, technical solutions, and advantages of the embodiments of the present disclosure clearer, the technical solutions in the embodiments of the present disclosure will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present disclosure. Apparently, the described embodiments are some, but not all, of the embodiments of the present disclosure. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present disclosure without creative efforts shall fall within the protection scope of the present disclosure.

[0045] In addition, the term "and / or" in this document merely describes an association relationship between associated objects and indicates that three relationships may exist. For example, A and / or B may represent: A exists alone, both A and B exist simultaneously, and B exists alone. In addition, the character " / " in this document generally represents an "or" relationship between the preceding and following associated objects.

[0046] In view of the problems in the background art, the embodiments of the present disclosure provide a spiral tunnel integrated intelligent ventilation and dust reduction method and device. Specifically, a physical model is constructed according to the spiral tunnel structure of forced ventilation and the constraint coding is determined; a deep learning network model and a CFD model are used to simulate the spiral tunnel structure. According to the simulation results of the CFD model and the on-site observation data, the deviation of the deep learning network model in terms of hydrodynamic force is obtained, and the deep learning network model is adjusted according to the deviation; the constraint coding is added to the adjusted deep learning network model for sensitivity analysis to obtain the key parameters affecting the ventilation and dust reduction of the spiral tunnel; the key parameters are monitored in real time, and according to the predicted dust concentration values corresponding to each monitoring point output by the deep learning network model with the constraint coding added, the corresponding key parameters are adjusted to optimize the ventilation rate and make the dust concentration at each monitoring point meet the preset standard. In this way, the research on the tunnel dust diffusion problem by the deep learning network model (PINN model) is supplemented, the model establishment process is simplified, the adaptability of the model in a complex flow environment and the ventilation and dust reduction treatment efficiency are improved, and the model can accurately predict the tunnel dust concentration at a certain moment with limited data.

[0047] The following combines the drawings and details the spiral tunnel integrated intelligent ventilation and dust reduction method provided by the embodiments of the present disclosure through specific embodiments.

[0048] Figure 1 The flowchart of a spiral tunnel integrated intelligent ventilation and dust reduction method provided by the embodiments of the present disclosure is shown. Method 100 includes the following steps:

[0049] S110, construct a physical model according to the spiral tunnel structure of forced ventilation, and determine the constraint coding according to the physical model.

[0050] In some embodiments, the constructing a physical model according to the spiral tunnel structure of forced ventilation includes:

[0051] According to the spiral tunnel structure of forced ventilation, use the Navier-Stokes equation, the turbulent kinetic energy equation, the turbulent energy dissipation rate equation and the component mass conservation equation to construct a physical model;

[0052] Further, the Navier-Stokes equation is expressed as:

[0053]

[0054]

[0055] where u represents the velocity vector of the air in the spiral tunnel, ρ represents the density of the air in the spiral tunnel, t represents time, p represents the pressure in the spiral tunnel, and μ is the dynamic viscosity of the air in the spiral tunnel;

[0056] The turbulent kinetic energy equation is expressed as:

[0057]

[0058] where k and ε represent the turbulent kinetic energy and the turbulent energy dissipation rate respectively, G k , μ t , σ k represent the turbulent kinetic energy generation term caused by the mean velocity gradient, the turbulent viscosity, and the turbulent Prandtl number of the turbulent kinetic energy equation respectively. is the Hamiltonian operator;

[0059] The turbulent energy dissipation rate equation is expressed as:

[0060]

[0061] where σ ε , S, and v represent the turbulent Prandtl number of the turbulent energy dissipation rate equation, the strain rate modulus, and the kinematic viscosity of the fluid respectively, C 1 represents an empirical constant. Specifically,

[0062]

[0063] where the parameter

[0064] C 2 is a constant. According to experience, C 2 is generally 1.9, σ k is 1, and σ ε is 1.2;

[0065] The component mass conservation equation is expressed as:

[0066]

[0067] where C, S1, and D represent the mass concentration of dust, the generation or disappearance rate of dust, and the diffusion coefficient of dust respectively.

[0068] In some embodiments, the Navier-Stokes equation can be used to capture the velocity field and pressure field characteristics of the airflow in the spiral tunnel, providing an accurate background flow field for dust diffusion;

[0069] The turbulent kinetic energy equation (k equation) and the turbulent energy dissipation rate equation (ε equation) can be used to describe the significant turbulent phenomena existing in the spiral tunnel. Compared with the existing standard k-ε equation, the turbulent kinetic energy equation and the turbulent energy dissipation rate equation proposed in this embodiment are more adaptable in complex flow environments. Especially when dealing with complex situations such as strong rotation and separated flow, they can more accurately simulate the turbulent characteristics;

[0070] The diffusion and transportation process of dust in the air can be described by using the component mass conservation equation, which includes the convection and diffusion processes. The spatio-temporal distribution of dust concentration can be accurately described by the component mass conservation equation, and this equation takes into account the influence of turbulent diffusion on dust propagation. Therefore, using this equation can improve the reliability of the physical model.

[0071] S120, use a deep learning network model and a CFD model to simulate the spiral tunnel structure of forced ventilation. According to the simulation results of the CFD model and the on-site observation data, obtain the deviation of the deep learning network model in terms of hydrodynamic force, and adjust the deep learning network model according to the deviation.

[0072] In some embodiments, the use of a deep learning network model and a CFD model to simulate the spiral tunnel structure of forced ventilation includes:

[0073] Use a deep learning network model and a CFD model to simulate the hydrodynamic response and temperature distribution of the spiral tunnel of forced ventilation under different ventilation environments and emergency environments.

[0074] In some embodiments, the deep learning network model is a physics-informed neural network model (PINN model).

[0075] In some embodiments, the ventilation environments include the ventilation environments at the initial construction stage and when the ventilation equipment is enabled, and the ventilation environment under complex geological conditions during the mid-construction stage;

[0076] The emergency environments include fire and failure of ventilation and dust reduction equipment;

[0077] Furthermore, the simulation of the hydrodynamic response and temperature distribution of the spiral tunnel of forced ventilation under different ventilation environments and emergency environments specifically includes:

[0078] Simulate the airflow response of the spiral tunnel of forced ventilation at the initial construction stage and when the ventilation equipment is enabled;

[0079] As the tunnel extends, simulate the airflow response under complex geology and corresponding construction conditions during the mid-construction stage;

[0080] Simulate the emergency ventilation response under high-risk events such as fire and failure of ventilation and dust reduction equipment.

[0081] In some embodiments, the adjustment of the deep learning network model according to the deviation includes:

[0082] Optimize the structure, hyperparameters, and training strategy of the deep learning network model according to the deviation; where

[0083] The structure of the optimized deep learning network model includes increasing or decreasing the number of network layers and adjusting the activation function to improve the performance of the deep learning network model in complex scenarios;

[0084] The hyperparameters of the optimized deep learning network model include adjusting the learning rate and the batch size to optimize the training process of the deep learning network model and improve the convergence speed;

[0085] The training strategy of the optimized deep learning network model includes introducing a regularization algorithm and changing the weight distribution of the loss function to enhance the sensitivity and accuracy of the deep learning network model to boundary conditions and sudden changes;

[0086] By adjusting the deep learning network model with the above methods, the dust concentration prediction accuracy and applicability of the deep learning network model can be improved, ensuring the effectiveness and reliability of the model in practical applications.

[0087] In some embodiments, method 100 further includes:

[0088] According to the hydrodynamic response and temperature distribution simulation results of the CFD model and the on-site observation data, change the weight distribution of the loss function of the deep learning network model to optimize multiple objectives, where

[0089] the multiple objectives include the ventilation rate, the temperature control strategy, and the dust concentration at each monitoring point;

[0090] Furthermore, the deep learning network model can utilize the gradient information of the spiral tunnel and the genetic algorithm to achieve multi-objective optimization, which can significantly improve the efficiency of finding the optimal ventilation rate, the optimal temperature control strategy, and the lowest dust concentration corresponding to each monitoring point, thereby maximizing the engineering benefits and environmental safety.

[0091] In some embodiments, the loss function in the deep learning network model is a composite loss function, which includes a ventilation rate loss function, a temperature control loss function, and a dust concentration loss function.

[0092] S130, add constraint encoding to the adjusted deep learning network model for sensitivity analysis to obtain the key parameters affecting the ventilation and dust reduction of the spiral tunnel.

[0093] In some embodiments, adding constraint encoding to the adjusted deep learning network model for sensitivity analysis includes:

[0094] Add constraint encoding to the adjusted deep learning network model and use the forward propagation algorithm to predict the influence of the environmental parameters of the spiral tunnel on the output value of the adjusted deep learning network model;

[0095] Using the automatic differentiation method, sort the influence of the predicted environmental parameters of the spiral tunnel on the output value of the adjusted deep learning network model in descending order.

[0096] In some embodiments, take the first few environmental parameters in the sorting as key parameters, so as to subsequently adjust the ventilation and dust reduction equipment according to the key parameters to optimize the ventilation rate and reduce the dust concentration.

[0097] In some embodiments, the key parameters include:

[0098] The position and number of ventilation openings of the spiral tunnel, the ventilation wind speed, the spatial layout of the construction area of the spiral tunnel, and the external climate conditions.

[0099] S140, monitor the key parameters in real time, and adjust the corresponding key parameters according to the predicted dust concentration values corresponding to each monitoring point output by the deep learning network model with constraint encoding, optimize the ventilation rate, and make the dust concentration corresponding to each monitoring point meet the preset standard.

[0100] In some embodiments, according to the predicted dust concentration values corresponding to each monitoring point output by the deep learning network model with constraint encoding and the real-time monitored key parameter situation, adjust the corresponding key parameters, which not only improves the performance of the ventilation and dust reduction equipment, but also significantly reduces the energy consumption, providing a more economical and environmentally friendly solution for tunnel construction.

[0101] The above is the introduction of the method embodiments. The following are specific embodiments adopting this method to further illustrate the solution of the present disclosure.

[0102] Construct a physical model according to the spiral tunnel structure when the left tunnel of Xiaopotou Tunnel is excavated to the III-class surrounding rock stage and determine the constraint encoding;

[0103] Taking the PINN model as the deep learning network model, use the PINN model and the CFD model to simulate the tunnel structure respectively. To verify the accuracy and reliability of the PINN model in predicting dust diffusion, select a representative monitoring point for actual observation, so as to compare the on-site observation data of this monitoring point and the simulation data of the CFD model at this monitoring point with the simulation data of the PINN model at this monitoring point, obtain the deviation of the PINN model in terms of hydrodynamic force, and adjust the PINN model according to this deviation; among them, the selection of this monitoring point should consider the dynamic change characteristics of the dust concentration to avoid the situation of too high dust concentration caused by being close to the dust source. In this example, the selected monitoring point is located 25m away from the tunnel face and close to the tunnel center line;

[0104] Figure 2Shows the comparison curve between the dust concentration predicted by the PINN model at this monitoring point within 5 - 10 minutes and the actual monitoring value. It can be seen from the figure that the dust concentration predicted by PINN basically coincides with the monitoring data at the same moment, but both are far higher than the allowable dust concentration. According to the limit value of the total dust concentration in the air of the workplace stipulated in the "Technical Specification for Highway Tunnel Construction" (JTG / T 3660—2020), the allowable dust concentration is 8mg / m 3 , calculate the root mean square error RMSE between the dust concentration predicted by PINN within 5 - 10 minutes and the monitoring data. The RMSE is 3.55mg / m 3 , which indicates that the PINN model has a high prediction accuracy;

[0105] In the adjusted PINN model, add constraint encoding for sensitivity analysis to obtain the key parameters affecting the ventilation and dust reduction of this spiral tunnel;

[0106] Real - time monitor these key parameters. According to the predicted dust concentration value of this monitoring point output by the PINN model with constraint encoding, adjust the corresponding key parameters to optimize the ventilation rate and reduce the dust concentration at this monitoring point to the allowable dust concentration.

[0107] According to the embodiments of the present disclosure, construct a physical model based on the structure of the forced - ventilation spiral tunnel and determine the constraint encoding; use a deep - learning network model and a CFD model to simulate the structure of the spiral tunnel. According to the simulation results of the CFD model and the on - site observation data, obtain the deviation of the deep - learning network model in terms of hydrodynamic force, and adjust the deep - learning network model according to the deviation; add constraint encoding in the adjusted deep - learning network model for sensitivity analysis to obtain the key parameters affecting the ventilation and dust reduction of this spiral tunnel; conduct real - time monitoring of the key parameters. According to the predicted dust concentration values corresponding to each monitoring point output by the deep - learning network model with constraint encoding, adjust the corresponding key parameters, optimize the ventilation rate and make the dust concentration at each monitoring point meet the preset standard. In this way, it makes up for the research on the tunnel dust diffusion problem by the deep - learning network model (PINN model), simplifies the model - building process, improves the adaptability of the model in a complex flow environment and the ventilation and dust - reduction treatment efficiency, and the model can also accurately predict the tunnel dust concentration at a certain moment with limited data.

[0108] It should be noted that for the foregoing method embodiments, for the sake of simple description, they are all expressed as a series of action combinations. However, those skilled in the art should know that the present disclosure is not limited by the described action sequence, because according to the present disclosure, certain steps can be performed in other sequences or simultaneously. Secondly, those skilled in the art should also know that the embodiments described in the specification are all optional embodiments, and the actions and modules involved are not necessarily essential to the present disclosure.

[0109] The above is an introduction to the method embodiments. The following will further illustrate the solution of the present disclosure through device embodiments.

[0110] Figure 3 Fig. shows the structural diagram of an integrated intelligent ventilation and dust reduction device for a spiral tunnel provided by an embodiment of the present disclosure. The device 300 includes:

[0111] A constraint code acquisition module 310, configured to construct a physical model according to the spiral tunnel structure of forced ventilation, and determine a constraint code according to the physical model.

[0112] In some embodiments, the constraint code acquisition module 310 is specifically configured to:

[0113] The constructing a physical model according to the spiral tunnel structure of forced ventilation includes:

[0114] According to the spiral tunnel structure of forced ventilation, using the Navier-Stokes equation, the turbulent kinetic energy equation, the turbulent energy dissipation rate equation, and the component mass conservation equation, construct a physical model.

[0115] A deep learning network model adjustment module 320, configured to simulate the spiral tunnel structure of forced ventilation by using a deep learning network model and a CFD model, obtain the deviation of the deep learning network model in terms of hydrodynamic force according to the simulation result of the CFD model and the on-site observation data, and adjust the deep learning network model according to the deviation.

[0116] In some embodiments, the deep learning network model adjustment module 320 is specifically configured to:

[0117] The simulating the spiral tunnel structure of forced ventilation by using a deep learning network model and a CFD model includes:

[0118] Use a deep learning network model and a CFD model to simulate the hydrodynamic response and temperature distribution of the spiral tunnel of forced ventilation under different ventilation environments and emergency environments.

[0119] In some embodiments, the deep learning network model adjustment module 320 is specifically further configured to:

[0120] The ventilation environment includes the ventilation environment during the initial construction stage and when the ventilation equipment is enabled, and the ventilation environment under complex geological conditions during the mid-construction stage;

[0121] The emergency environment includes fire and failure of ventilation and dust reduction equipment.

[0122] In some embodiments, the deep learning network model adjustment module 320 is specifically further configured to:

[0123] Adjusting the deep learning network model according to the deviation includes:

[0124] Optimizing the structure, hyperparameters, and training strategy of the deep learning network model according to the deviation; wherein,

[0125] Optimizing the structure of the deep learning network model includes increasing or decreasing the number of network layers and adjusting the activation function;

[0126] Optimizing the hyperparameters of the deep learning network model includes adjusting the learning rate and the batch size;

[0127] Optimizing the training strategy of the deep learning network model includes introducing a regularization algorithm and changing the weight allocation of the loss function.

[0128] In some embodiments, the apparatus 300 is further specifically configured to:

[0129] Changing the weight allocation of the loss function of the deep learning network model according to the hydrodynamic response and temperature distribution simulation results of the CFD model and the on-site observation data to optimize multiple objectives; wherein,

[0130] The multiple objectives include the ventilation rate, the temperature control strategy, and the dust concentration at each monitoring point.

[0131] The key parameter acquisition module 330 is configured to perform sensitivity analysis by adding constraint encoding to the adjusted deep learning network model to obtain the key parameters affecting the ventilation and dust reduction of the spiral tunnel.

[0132] In some embodiments, the key parameter acquisition module 330 is specifically configured to:

[0133] Performing sensitivity analysis by adding constraint encoding to the adjusted deep learning network model includes:

[0134] Adding constraint encoding to the adjusted deep learning network model and using the forward propagation algorithm to predict the influence of the environmental parameters of the spiral tunnel on the output value of the adjusted deep learning network model;

[0135] Using the automatic differentiation method to sort the influence of the predicted environmental parameters of the spiral tunnel on the output value of the adjusted deep learning network model in descending order.

[0136] In some embodiments, the key parameter acquisition module 330 is further specifically configured to:

[0137] The key parameters include:

[0138] The position and number of ventilation openings of the spiral tunnel, the ventilation air velocity, the spatial layout of the construction area of the spiral tunnel, and the external climate conditions.

[0139] The key parameter optimization module 340 is configured to monitor the key parameters in real time, and adjust the corresponding key parameters according to the predicted dust concentration values corresponding to each monitoring point output by the deep learning network model with constraint encoding added, so as to optimize the ventilation rate and make the dust concentration corresponding to each monitoring point meet the preset standard.

[0140] It can be understood that Figure 3 Each module / unit in the device 300 shown has the functions of implementing the steps in the method 100 provided in the embodiments of the present disclosure, and can achieve their corresponding technical effects. For the sake of brevity, they will not be described in detail here.

[0141] Figure 4 The structural diagram of an exemplary electronic device capable of implementing the embodiments of the present disclosure is shown. The electronic device 400 is intended to represent various forms of digital computers, such as, laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device 400 can also represent various forms of mobile devices, such as, personal digital processors, cellular phones, smart phones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the present disclosure described and / or claimed herein.

[0142] As Figure 4 As shown, the electronic device 400 includes a computing unit 401, which can perform various appropriate actions and processes according to the computer program stored in the read-only memory (ROM) 402 or the computer program loaded from the storage unit 408 into the random access memory (RAM) 403. In the RAM 403, various programs and data required for the operation of the electronic device 400 can also be stored. The computing unit 401, the ROM 402, and the RAM 403 are connected to each other through a bus 404. The I / O interface 405 is also connected to the bus 404.

[0143] A plurality of components in the electronic device 400 are connected to the I / O interface 405, including: an input unit 406, such as a keyboard, a mouse, etc.; an output unit 407, such as various types of displays, speakers, etc.; a storage unit 408, such as a magnetic disk, an optical disk, etc.; and a communication unit 409, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 409 allows the electronic device 400 to exchange information / data with other devices through a computer network such as the Internet and / or various telecommunication networks.

[0144] The computing unit 401 can be various general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of the computing unit 401 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various dedicated artificial intelligence (AI) computing chips, various computing units running machine learning model algorithms, a digital signal processor (DSP), and any suitable processor, controller, microcontroller, etc. The computing unit 401 executes the various methods and processes described above, such as method 100. For example, in some embodiments, method 100 can be implemented as a computer software program tangibly embodied in a machine-readable medium, such as the storage unit 408. In some embodiments, part or all of the computer program can be loaded and / or installed onto the electronic device 400 via the ROM 402 and / or the communication unit 409. When the computer program is loaded into the RAM 403 and executed by the computing unit 401, one or more steps of the method 100 described above can be executed. Alternatively, in other embodiments, the computing unit 401 can be configured to execute method 100 in any other suitable manner (e.g., by means of firmware).

[0145] The various embodiments of the systems and techniques described above in this document can be implemented in digital electronic circuitry, integrated circuit systems, field-programmable gate arrays (FPGA), application-specific integrated circuits (ASIC), application-specific standard products (ASSP), system-on-a-chip (SOC), complex programmable logic devices (CPLD), computer hardware, firmware, software, and / or combinations thereof. These various embodiments can include: implemented in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which can be a special-purpose or general-purpose programmable processor, and can receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit the data and instructions to the storage system, the at least one input device, and the at least one output device.

[0146] The program code for implementing the methods of the present disclosure can be written in any combination of one or more programming languages. These program codes can be provided to a processor or controller of a general-purpose computer, a special-purpose computer, or other programmable data processing device, such that when the program codes are executed by the processor or controller, the functions / operations specified in the flowcharts and / or block diagrams are implemented. The program code can be executed entirely on the machine, partially on the machine, as a stand-alone software package partially on the machine and partially on a remote machine, or entirely on a remote machine or server.

[0147] In the context of the present disclosure, a machine-readable medium can be a tangible medium that can contain or store a program for use by or in connection with an instruction execution system, apparatus, or device. The machine-readable medium can be a machine-readable signal medium or a machine-readable storage medium. The machine-readable medium can include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination of the foregoing. More specific examples of the machine-readable storage medium would include an electrical connection based on one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or Flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0148] It should be noted that the present disclosure also provides a non-transitory computer-readable storage medium storing computer instructions, wherein the computer instructions are used to cause a computer to execute method 100 and achieve the corresponding technical effects achieved by the method of the embodiments of the present disclosure. For the sake of concise description, details are not repeated herein.

[0149] To provide interaction with a user, the systems and techniques described herein can be implemented on a computer having: a display device for displaying information to the user; and a keyboard and a pointing device (e.g., a mouse or a trackball) through which the user can provide input to the computer. Other kinds of devices can also be used to provide interaction with the user; for example, the feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including acoustic input, voice input, or tactile input).

[0150] The systems and techniques described herein can be implemented in a computing system including backend components (e.g., as a data server), or a computing system including middleware components (e.g., an application server), or a computing system including frontend components (e.g., a user computer having a graphical user interface or a web browser through which the user can interact with an implementation of the systems and techniques described herein), or a computing system including any combination of such backend components, middleware components, or frontend components. The components of the system can be interconnected to each other by any form or medium of digital data communication (e.g., a communication network). Examples of the communication network include: a local area network (LAN), a wide area network (WAN), and the Internet.

[0151] A computer system may include a client and a server. The client and the server are generally far from each other and usually interact via a communication network. The relationship between the client and the server is generated by computer programs running on the respective computers and having a client-server relationship with each other. The server may be a cloud server, a server of a distributed system, or a server incorporating a blockchain.

[0152] It should be understood that various forms of the processes shown above may be used, steps may be reordered, added or deleted. For example, the steps described in this disclosure may be executed in parallel, sequentially, or in a different order, as long as the desired results of the technical solutions disclosed in this disclosure can be achieved, and this is not limited herein.

[0153] The above specific embodiments do not constitute a limitation on the protection scope of this disclosure. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principle of this disclosure shall be included within the protection scope of this disclosure.

Claims

1. A spiral tunnel integrated intelligent ventilation and dust reduction method, characterized in that: include: A physical model is constructed according to the spiral tunnel structure of pressure ventilation, and constraint coding is determined according to the physical model; The deep learning network model and the CFD model are used to simulate the spiral tunnel structure of the pressure-type ventilation, and the deviation of the deep learning network model in fluid dynamics is obtained according to the simulation results of the CFD model and the field observation data, and the deep learning network model is adjusted according to the deviation; Constraint coding is added to the adjusted deep learning network model to perform sensitivity analysis to obtain key parameters that affect ventilation and dust reduction in the spiral tunnel; The key parameters are monitored in real time, and according to the dust concentration prediction value corresponding to each monitoring point output by the deep learning network model with constraint coding, the corresponding key parameters are adjusted to optimize the ventilation rate and make the dust concentration corresponding to each monitoring point meet the preset standard.

2. The method according to claim 1, characterized in that The physical model is constructed according to the spiral tunnel structure of the forced-in ventilation, including: According to the spiral tunnel structure of pressure ventilation, a physical model was constructed using the Navier-Stokes equations, turbulent kinetic energy equations, turbulent energy dissipation rate equations and component mass conservation equations.

3. The method according to claim 1, characterized in that The simulation of the spiral tunnel structure of the pressure-type ventilation using the deep learning network model and the CFD model includes: The deep learning network model and CFD model are used to simulate the fluid dynamic response and temperature distribution of the pressure-ventilated spiral tunnel under different ventilation environments and emergency environments.

4. The method according to claim 3, characterized in that The ventilation environment includes the ventilation environment in the initial construction stage and when the ventilation equipment is put into use, and the ventilation environment under complex geological conditions in the mid-term construction stage; The emergency environment includes fire and failure of ventilation and dust reduction equipment.

5. The method according to claim 3, characterized in that: The step of adjusting the deep learning network model according to the deviation comprises: The structure, hyperparameters and training strategy of the deep learning network model are optimized according to the deviation; wherein, Optimizing the structure of the deep learning network model includes increasing or decreasing the network level and adjusting the activation function; The hyperparameters of the deep learning network model optimization include adjusting the learning rate and the batch processing quantity; The training strategy for optimizing the deep learning network model includes introducing a regularization algorithm and changing the weight distribution of the loss function.

6. The method according to claim 5, characterized in that The method further comprises: According to the fluid dynamics response and temperature distribution simulation results of the CFD model and the field observation data, the weight distribution of the loss function of the deep learning network model is changed to optimize multiple objectives; among them, The multiple objectives include ventilation rate, temperature control strategy and dust concentration at each monitoring point.

7. The method according to claim 1, characterized in that The sensitivity analysis of adding constraint coding to the adjusted deep learning network model includes: Adding constraint coding to the adjusted deep learning network model, and using a forward propagation algorithm to predict the influence of the environmental parameters of the spiral tunnel on the output value of the adjusted deep learning network model; The influence of the predicted environmental parameters of the spiral tunnel on the output value of the adjusted deep learning network model is sorted in descending order using the automatic differentiation method.

8. The method according to claim 1, characterized in that The key parameters include: The location and number of spiral tunnel ventilation openings, ventilation wind speed, spatial layout of the spiral tunnel construction area, and external climate conditions.

9. An integrated intelligent ventilation and dust reduction device for a spiral tunnel, characterized in that: include: A constraint code acquisition module is used to construct a physical model according to the spiral tunnel structure of the forced-in ventilation, and determine the constraint code according to the physical model; A deep learning network model adjustment module is used to simulate the spiral tunnel structure of the pressure-type ventilation using the deep learning network model and the CFD model, obtain the deviation of the deep learning network model in fluid dynamics according to the simulation results of the CFD model and the field observation data, and adjust the deep learning network model according to the deviation; A key parameter acquisition module is used to add constraint coding to the adjusted deep learning network model to perform sensitivity analysis and obtain key parameters that affect the ventilation and dust reduction of the spiral tunnel; The key parameter optimization module is used to monitor the key parameters in real time, and adjust the corresponding key parameters according to the dust concentration prediction value corresponding to each monitoring point output by the deep learning network model with constraint coding, optimize the ventilation rate and make the dust concentration corresponding to each monitoring point meet the preset standard.

10. An electronic device, characterized in that: include: at least one processor; as well as a memory communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the method of any one of claims 1 to 8.

Images