Method and system for efficiently simulating grouting at different temperatures based on deep learning

Abstract

The invention relates to the technical field of grouting simulation, in particular to an efficient simulation method and system for grouting at different temperatures based on deep learning, and the method comprises the steps: constructing a relational expression between a rock-soil mass medium and the change of a grouting material under different temperature conditions; the rock-soil mass medium and grouting material change relational expressions under different temperature conditions comprise a relational expression of temperature and porosity, a relational expression of temperature and stress strain, and a relational expression of temperature and slurry viscosity; on the basis of actual stratum data of the grouting area, to-be-determined parameters in all the relational expressions are obtained through fitting; constructing an improved recurrent neural network; training the improved recurrent neural network to obtain a trained improved recurrent neural network; and inputting the temperature of the to-be-predicted grouting area into the trained improved recurrent neural network to obtain the slurry flow velocity and the slurry permeation and diffusion distance of the to-be-predicted grouting area. And efficient simulation of the whole grouting process at different temperatures is realized.

Description

Technical Field

[0001] The present invention relates to the technical field of grouting simulation, and in particular to an efficient simulation method and system for grouting at different temperatures based on deep learning. Background Art

[0002] At present, the construction of underground engineering projects is in a period of rapid development. More and more underground spaces and tunnel projects are moving towards extremely complex areas. The complex geological sections and frequent geological disasters encountered during the construction process pose great challenges to the construction of underground projects. Among them, the influence of ground temperature is one of the problems encountered in the construction of underground projects in recent years. Compared with the normal temperature environment, sudden geological disasters such as high-temperature water and high-temperature steam splashing in cracks are often more destructive due to the temperature effect. Grouting, as one of the effective means to control such sudden geological disasters, can often achieve good plugging and reinforcement effects. However, affected by the temperature effect, the physical parameter properties of grouting materials often change significantly, further affecting the plugging and reinforcement effects of grouting construction, making it difficult to effectively treat the geological disasters ahead.

[0003] The existing research methods for grouting mainly include physical experiments, theoretical analysis and numerical simulation. Among them, grouting numerical simulation can well visualize and analyze the whole process of grouting and has been widely used. However, the existing research on the grouting simulation method considering the temperature effect is not perfect. It does not consider the non-single function change of rheological properties caused by the heat interaction between the grout and the surrounding environment during the grouting process, and it is difficult to effectively simulate the plugging and reinforcement effects of the penetration and diffusion of grout considering the temperature effect. Summary of the Invention

[0004] In order to solve the deficiencies of the prior art, the present invention provides an efficient simulation method and system for grouting at different temperatures based on deep learning; considering the time-varying hydration reaction rate and viscosity of the grout caused by temperature changes, and at the same time considering the stress-strain changes of the geotechnical medium itself affected by temperature and the expansion and contraction properties generated with temperature changes, considering temperature as a learning parameter during model training, considering the heat interaction between the grout itself and the medium environment, performing mapping of similar temperature conditions at different time series to accelerate the simulation solution process, establishing the corresponding relationship between the physical parameter changes of the grout-geotechnical medium and the grout diffusion under different temperature conditions, and realizing the efficient simulation of the whole process of grouting at different temperatures.

[0005] On the one hand, an efficient simulation method for grouting at different temperatures based on deep learning is provided, including:

[0006] Construct the relationship expressions for the changes of rock and soil media and grouting materials under different temperature conditions; the relationship expressions for the changes of rock and soil media and grouting materials under different temperature conditions include: the relationship expression between temperature and porosity, the relationship expression between temperature and stress-strain, and the relationship expression between temperature and slurry viscosity; based on the actual formation data of the grouting area, fit the undetermined parameters in each of the above relationship expressions;

[0007] Construct an improved recurrent neural network; the improved recurrent neural network includes: a recurrent layer, and the recurrent layer includes: a first design layer, a second design layer, and a third design layer connected in sequence, wherein the output end of the second design layer is also connected to the input end of the first design layer; the relationship expression between temperature and porosity, the relationship expression between temperature and stress-strain, and the relationship expression between temperature and slurry viscosity are set in the third design layer;

[0008] Input the temperature of the grouting area to be predicted into the trained improved recurrent neural network to obtain the slurry flow velocity and the slurry penetration and diffusion distance of the grouting area to be predicted.

[0009] On the other hand, a high-efficiency simulation system for grouting at different temperatures based on deep learning is provided, including:

[0010] A relationship expression construction module, which is configured to: construct the relationship expressions for the changes of rock and soil media and grouting materials under different temperature conditions; the relationship expressions for the changes of rock and soil media and grouting materials under different temperature conditions include: the relationship expression between temperature and porosity, the relationship expression between temperature and stress-strain, and the relationship expression between temperature and slurry viscosity; based on the actual formation data of the grouting area, fit the undetermined parameters in each of the above relationship expressions;

[0011] A network construction module, which is configured to: construct an improved recurrent neural network; the improved recurrent neural network includes: a recurrent layer, and the recurrent layer includes: a first design layer, a second design layer, and a third design layer connected in sequence, wherein the output end of the second design layer is also connected to the input end of the first design layer; the relationship expression between temperature and porosity, the relationship expression between temperature and stress-strain, and the relationship expression between temperature and slurry viscosity are set in the third design layer;

[0012] An output module, which is configured to: input the temperature of the grouting area to be predicted into the trained improved recurrent neural network to obtain the slurry flow velocity and the slurry penetration and diffusion distance of the grouting area to be predicted.

[0013] The above technical solutions have the following advantages or beneficial effects:

[0014] Establish a database of grouting parameters under different temperature conditions, collect grouting diffusion data under different temperature conditions to form a data set, and at the same time store the physical parameter change data of the geotechnical medium and the grout corresponding to the temperature change under limited temperature conditions, the divided three-dimensional spatial grid and its central point data.

[0015] Consider the physical parameter changes of the grout and the geotechnical medium caused by the temperature effect, and comprehensively determine the temperature change of the geotechnical medium by combining fluid-solid heat transfer and environmental heat transfer. When training the model, consider the temperature as a parameter to be learned, and consider the mapping of similar temperature conditions in different time series to accelerate the simulation solution process. In this process, the area with more mapping times is the trend type where the grout is more likely to spread, which helps to determine the overall grout diffusion effect.

[0016] At the same time, consider the degree of gain of different parameters on the fluidity of the grout and the actual penetration and diffusion distance, determine the actual flow velocity and penetration and diffusion distance of the grout through non-linear gain superposition, obtain the grout viscosity properties and penetration and diffusion forms at each spatial position in different time series in real time, analyze the diffusion effect of the grout in the geotechnical medium, judge the actual plugging and reinforcement situation of the grouting, and further optimize the grouting strategy for efficient treatment of grouting disasters under different temperature conditions. Brief Description of the Drawings

[0017] The specification drawings forming a part of the present invention are used to provide a further understanding of the present invention. The schematic embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute an improper limitation of the present invention.

[0018] Figure 1 It is the flowchart of the method for the first embodiment;

[0019] Figure 2 It is the internal structure diagram of the improved recurrent neural network for the first embodiment. Detailed Description of the Invention

[0020] It should be noted that the following detailed description is exemplary and is intended to provide further explanation of the present invention. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by those of ordinary skill in the technical field to which the present invention belongs.

[0021] Example 1, as Figure 1 shown, this embodiment provides an efficient simulation method for grouting at different temperatures based on deep learning, including:

[0022] S101: Establish the relational expressions for the changes of geotechnical media and grouting materials under different temperature conditions; the relational expressions for the changes of geotechnical media and grouting materials under different temperature conditions include: the relational expression between temperature and porosity, the relational expression between temperature and stress-strain, and the relational expression between temperature and slurry viscosity; based on the actual formation data of the grouting area, fit the undetermined parameters in each of the above relational expressions;

[0023] S102: Build an improved recurrent neural network; as Figure 2 shown, the improved recurrent neural network includes: a recurrent layer, and the recurrent layer includes: a first design layer, a second design layer, and a third design layer connected in sequence, wherein the output end of the second design layer is also connected to the input end of the first design layer; the relational expressions between temperature and porosity, the relational expression between temperature and stress-strain, and the relational expression between temperature and slurry viscosity are set in the third design layer;

[0024] S103: Input the temperature of the grouting area to be predicted into the trained improved recurrent neural network to obtain the slurry flow velocity and the slurry penetration and diffusion distance of the grouting area to be predicted.

[0025] Further, the relational expression between temperature and porosity has the following specific expression:

[0026] φ(T) = φ 0 (1 + α T (T - T 0 ));

[0027] wherein, T represents temperature, φ(T) represents the porosity varying with temperature, φ 0 is the porosity under the reference temperature T 0 condition, and α T is the volume expansion coefficient varying with temperature.

[0028] Further, the relational expression between temperature and stress-strain has the following specific expression:

[0029]

[0030] wherein, T represents temperature, σ(T) represents the stress value varying with temperature, E 0 is the elastic modulus under the reference temperature T 0 condition, β T is the exponential temperature decay coefficient of the elastic modulus, and ε is strain.

[0031] Further, the relational expression between temperature and slurry viscosity has the following specific expression:

[0032] μ(T,t) = μ 0 (T)·e (α(T)t) + μ 1(T);

[0033] Among them, μ(T, t) represents the slurry viscosity function that varies with temperature and time, T represents temperature, t represents time, μ 0 (T) is the coefficient that varies with temperature, α(T) is the viscosity change coefficient that varies with temperature, μ 1 (T) is the coefficient that varies with temperature.

[0034] It should be understood that the porosity refers to the percentage of the pores in the rock and soil mass to the total volume in the natural state.

[0035] Stress refers to the internal force generated by the interaction between parts of an object when the rock and soil mass is affected by external forces, its own gravity, or other factors such as humidity and temperature. Strain refers to the relative local deformation of an object under the action of external forces and non-uniform temperature fields and other factors.

[0036] The slurry viscosity refers to the rheological property that shows non-linear growth over time, which is reflected in the phase transformation of the slurry from a liquid-solid two-phase state.

[0037] It should be understood that first, consider the actual formation conditions in the grouting area, collect the original samples of the rock and soil mass in this area for analysis, keep its temperature constant through a constant temperature curing box, measure the porosity and stress-strain relationship of the medium in the original sample state through physical and mechanical tests, repeat this step, successively select the original samples in different temperature regions of this formation for analysis, obtain the porosity and stress-strain relationship data in different temperature regions through non-linear fitting, and fit the curve of the porosity change and stress-strain relationship of the rock and soil medium considering the temperature effect based on the obtained data.

[0038] Collect the basic properties of the grouting material. The basic properties of the grouting material include: the materials selected for the slurry and the slurry type selection; the materials selected for the grouting include: different types of cement, water glass; the slurry type selection includes: single-component cement slurry, cement-sodium silicate double liquid, chemical slurry, etc. Determine the size of the grouting cement particles selected, obtain the time-varying data of the slurry viscosity of each slurry type selection under different temperature conditions through rheological tests, fit the viscosity time-varying function through data, and simultaneously measure the strength characteristics after solidification.

[0039] Furthermore, based on the actual formation data in the grouting area, fit the undetermined parameters in the above relationship, and implement it by using polynomial curve fitting.

[0040] Furthermore, for the improved recurrent neural network, the basic framework of the recurrent neural network includes an input layer, a hidden layer, and an output layer. In the calculation of the hidden layer, considering the mutual influence on different time series, the value of the hidden layer at the current moment depends not only on the input at the current moment but also on the value of the previous hidden layer.

[0041] Further, the improved recurrent neural network includes an input layer, a hidden layer, and an output layer connected in sequence. The output end of the hidden layer is also connected to the input end of a recurrent layer, and the output end of the recurrent layer is connected to the input end of the hidden layer. In the calculation of the hidden layer of the improved recurrent neural network, a loop step for temperature judgment is added at each time series. Before considering the input received by the hidden layer at the current time series and the value of the hidden layer at the previous time series, the determination of the ambient temperature between the current time series and the previous time series is added. When the judgment effect is met, the calculation of the hidden layer of the current time series is directly skipped, and the temperature of the previous time series is directly mapped to the current time series through the temperature comparison relationship with the previous time series. At the same time, temperature data is stored at each time series for the determination and mapping of the ambient temperature of the next time series.

[0042] Further, the first design layer includes:

[0043] If T j,t = nβ j,k,t T j,k,t-1 holds, then execute T r,t = nT r,k,t-1 ;

[0044] where j represents different regions around the perimeter of the calculation region k of the previous time series in the region to be calculated first, T j,t represents the temperature in different j regions at time t, n represents an integer multiple relationship, β j,k,t represents the similarity coefficient under the condition of n times the relationship between different j regions around the perimeter of region k at time series t and the current calculation region; T j,k,t-1 represents the temperature in different j regions at time t - 1, T r,k,t-1 represents the temperature value of this calculation region at the previous time step, T r,t is the temperature data used in the actual calculation process, k represents the perimeter calculation region of the previous time series of the current calculation region, and β j,k is the similarity coefficient under the condition of n times the relationship between different j regions around the perimeter of region k and the current calculation region. By combining the time representation of trigonometric functions, a numerical limit extremely close to 1 is set for different loop iteration steps, and the cycle and discontinuous period of itself are set to approximate the temperature change value of the actual calculation.

[0045] It should be understood that the temperature mapping of adjacent sequences is considered in the current time series: the temperature mapping relationship of adjacent time series is added to the self-loop part of the hidden layer. Before solving the temperature change of the target area at the current time step, the adjacent areas of the target area are searched. For areas with similar peripheral temperature conditions in adjacent time dimensions, that is, the areas of the next time series of the current time series, the temperature conditions of the previous time series that have been calculated are directly used for temperature mapping to accelerate the calculation and prediction of the temperature conditions of each area at different time steps.

[0046] Furthermore, the second design layer includes:

[0047]

[0048] where, T i,t is the temperature change after the slurry interacts with heat in this area, T K,t is the temperature data in the surrounding three-dimensional directions, K represents the surrounding calculation areas of the current time series of the current calculation area, N represents the number of surrounding calculation areas of the current time series of the current calculation area, ε K,t is the temperature difference reduction coefficient, which is smaller in the slurry injection flow direction and larger in the other two directions. T r,t-1 represents the temperature data of this area at time t - 1. The temperature difference reduction coefficient refers to the temperature attenuation coefficient existing in the environmental heat interaction and can be set as different constants according to actual situations.

[0049] It should be understood that the temperature effect of the current time series: In the calculation of temperature data on different grids of the first time series and time series without mapping relationship, the influence of fluid-solid heat transfer and environmental heat interaction during the grouting process under the influence of temperature effect is considered simultaneously. That is, during the process of slurry penetration and diffusion, heat interaction will occur between the slurry and the surrounding geotechnical media, resulting in temperature changes. At the same time, the surrounding geotechnical media will also be affected by the surrounding environment and produce temperature changes. During the process of temperature change caused by this heat exchange, the slurry viscosity first follows the time-varying function relationship at the corresponding initial temperature in the database. As time progresses, the change in the temperature of the slurry itself causes it to obey different viscosity time-variations at each time sequence, and the viscosity time-variation of the slurry under different temperature conditions at each time sequence is obtained.

[0050] For the change in slurry temperature, the fluid-solid coupling heat transfer method is used for calculation, and the calculated temperature change is the change in the temperature of the slurry itself.

[0051] For the geotechnical medium in this area, the heat transfer method of fluid-solid coupling is combined with the heat interaction of the surrounding environment for calculation. Considering the heat transfer from the surrounding environment to this area, with the direction of the grouting slurry flow as the gradient descent direction of temperature change, different three-dimensional space points and temperature data are used as basic data. Based on the temperature value after the heat exchange between the slurry and the current grid area, the temperature of the geotechnical medium in this area still needs to be added with the difference between the temperature of the surrounding environment multiplied by the coefficient value, and the actual temperature value of the geotechnical medium in this area updated to the next time series is comprehensively determined.

[0052] Further, the third design layer includes:

[0053] Y(l p ) t = W q,t α q,t m q,t - W q,t-1 α q,t-1 m q,t-1 + b(T r,t - T r,t-1 )

[0054] Where α q,t represents the vector gain value of different m q to l p at time series t, and W q,t represents the weight coefficient of different vector gain values at time series t; m q,t represents the physical parameter values of the slurry and the geotechnical medium at time series t, including the viscosity μ(T r,t , t) of the slurry, the porosity φ(T r,t ) t of the geotechnical medium, and the stress-strain σ(T r,t ) t ; T r,t-1 represents the temperature data of this area at time series t - 1; α q,t-1 represents the vector gain value of different m q to l p at time series t - 1, and W q,t-1 represents the weight coefficient of different vector gain values at time series t - 1; m q,t-1 represents the physical parameter values of the slurry and the geotechnical medium at time series t - 1, including the viscosity μ(T r,t-1 , t - 1) of the slurry, the porosity φ(T r,t-1 ) t-1 of the geotechnical medium, and the stress-strain σ(T r,t-1 ) t-1 ; T r,t-1 represents the temperature data of this area at time series t - 1; Y represents the output data of each time series, l pIndicates the vector growth of the vector slurry flow velocity and the penetration and diffusion distance under the current time series conditions, m q Indicates the physical parameter values of the normalized slurry and the geotechnical medium under different temperature conditions, α q Indicates at different m q For l p The vector gain value of, b represents the bias under different time series conditions, which varies accordingly according to different temperatures. The physical parameters include: the viscosity of the slurry, the porosity of the geotechnical medium, and the stress and strain.

[0055] It should be understood that for the slurry diffusion in different time series: after updating the temperature data in each time series, update the slurry diffusion data, consider the gain degree of each parameter of the slurry and the geotechnical medium on the slurry fluidity and the actual penetration and diffusion distance, and determine the actual flow velocity and penetration and diffusion distance of the slurry through the non-linear gain superposition value, which is used as the initial data for the spatio-temporal distribution of the slurry in the next time series.

[0056] Furthermore, the S103: training the improved recurrent neural network includes:

[0057] Constructing a first training set; the first training set is the known slurry flow velocity and slurry penetration and diffusion distance at different temperatures; inputting the first training set into the improved recurrent neural network, where the input value of the improved recurrent neural network is the temperature and the output values are the slurry flow velocity and the slurry penetration and diffusion distance, and training it. When the loss function value of the recurrent neural network no longer decreases, stop training to obtain the trained recurrent neural network. The loss function uses the mean squared error loss function.

[0058] It should be understood that the slurry flow velocity refers to the flow velocity of the slurry in the injected medium at each time during the grouting process; the slurry penetration and diffusion distance refers to the penetration and diffusion distance of the slurry in the injected medium at each time during the grouting process.

[0059] During the process of training the model, compare the predicted data calculated by the gradient transfer at each set time with the true value, and use the optimizer to update the gain value α q And the weight values corresponding to different temperature conditions.

[0060] In each training cycle, divide the data into multiple batches and train them sequentially. Use forward propagation to calculate the predicted values of the model, calculate the loss value in the loss function, and then perform backpropagation and use the optimizer to update the gain value and weight value of the model until the loss value is within the threshold range to complete the model training.

[0061] Embodiment 2

[0062] This embodiment provides a high-efficiency simulation system for grouting at different temperatures based on deep learning, including:

[0063] A relational expression construction module, which is configured to: construct relational expressions of the changes of the geotechnical medium and the grouting material under different temperature conditions; the relational expressions of the changes of the geotechnical medium and the grouting material under different temperature conditions include: the relational expression of temperature and porosity, the relational expression of temperature and stress-strain, and the relational expression of temperature and slurry viscosity; based on the actual formation data of the grouting area, fit the undetermined parameters in each of the above relational expressions.

[0064] A network construction module, which is configured to: construct an improved recurrent neural network; the improved recurrent neural network includes: a recurrent layer, and the recurrent layer includes: a first design layer, a second design layer, and a third design layer that are connected in sequence, wherein the output end of the second design layer is also connected to the input end of the first design layer; the relational expression of temperature and porosity, the relational expression of temperature and stress-strain, and the relational expression of temperature and slurry viscosity are set in the third design layer.

[0065] An output module, which is configured to: input the temperature of the grouting area to be predicted into the trained improved recurrent neural network to obtain the slurry flow velocity and the slurry penetration and diffusion distance of the grouting area to be predicted.

[0066] The above are only the preferred embodiments of the present invention and are not used to limit the present invention. For those skilled in the art, the present invention can have various changes and modifications. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included within the protection scope of the present invention.

Claims

1. An efficient simulation method for grouting at different temperatures based on deep learning, characterized by: include: Construct the relationship between the changes of rock and soil medium and grouting materials under different temperature conditions; The relationship between the rock and soil medium and the grouting material under different temperature conditions includes: the relationship between temperature and porosity, the relationship between temperature and stress-strain, and the relationship between temperature and slurry viscosity; based on the actual formation data of the grouting area, fitting the parameters to be determined in the above-mentioned relationship formulas; Constructing an improved recurrent neural network; the improved recurrent neural network comprises: a recurrent layer, the recurrent layer comprises: a first design layer, a second design layer and a third design layer connected in sequence, wherein the output end of the second design layer is also connected to the input end of the first design layer; the third design layer is provided with the relationship between the temperature and the porosity, the relationship between the temperature and the stress-strain, and the relationship between the temperature and the slurry viscosity; The temperature of the grouting area to be predicted is input into the trained improved recurrent neural network to obtain the slurry flow velocity and slurry penetration and diffusion distance of the grouting area to be predicted.

2. The high-efficiency simulation method for grouting at different temperatures based on deep learning as claimed in claim 1 is characterized in that: The relationship between temperature and porosity is specifically expressed as follows: φ(T)=φ0(1+α T (T-T0)); Where T represents temperature, φ(T) represents the porosity that changes with temperature, φ0 represents the porosity at the reference temperature T0, and α T is the coefficient of volume expansion that varies with temperature.

3. The high-efficiency simulation method for grouting at different temperatures based on deep learning as claimed in claim 1 is characterized in that: The relationship between temperature and stress-strain is specifically expressed as follows: Where T represents temperature, σ(T) represents the stress value that changes with temperature, E0 is the elastic modulus under the reference temperature T0, β T is the exponential temperature attenuation coefficient of the elastic modulus, and ε is the strain.

4. The high-efficiency simulation method for grouting at different temperatures based on deep learning as claimed in claim 1 is characterized in that: The relationship between the temperature and the slurry viscosity is specifically expressed as follows: μ(T,t)=μ0(T)·e (α(T)t) +μ1(T); Among them, μ(T,t) represents the viscosity function of the slurry that changes with temperature and time, T represents temperature, t represents time, μ0(T) is the coefficient that changes with temperature, α(T) is the coefficient of viscosity change with temperature, and μ1(T) is the coefficient that changes with temperature.

5. The high-efficiency simulation method for grouting at different temperatures based on deep learning as claimed in claim 1 is characterized in that: The improved recurrent neural network comprises: an input layer, a hidden layer and an output layer connected in sequence, wherein the output end of the hidden layer is also connected to the input end of the recurrent layer, and the output end of the recurrent layer is connected to the input end of the hidden layer; in the hidden layer calculation of the improved recurrent neural network, a cyclic step for temperature judgment is added to each time series, and before considering that the hidden layer of the current time series receives the input and the hidden layer value of the previous time series, the surrounding temperature judgment of the current time series and the previous time series is added, and when the judgment effect is met, the hidden layer calculation of the current time series is directly skipped, and the temperature of the previous time series is directly mapped to the current time series through the temperature comparison relationship with the previous time series, and the temperature data is stored in each time series for the surrounding temperature judgment and mapping of the next time series.

6. The high-efficiency simulation method for grouting at different temperatures based on deep learning as claimed in claim 1 is characterized in that: The first design layer includes: If T j,t =nβ j,k,t T j,k,t-1 If established, execute T r,t =nT r,k,t-1 ; Among them, j represents the surrounding different areas of the previous time series of the first calculation area k, T j,t represents the temperature of different j regions at time t, n represents the integer multiple relationship, β j,k,t It represents the similarity coefficient of the time series t in different j regions around region k under the condition of n times relationship with the current calculation region; T j,k,t-1 represents the temperature of different j regions at time t-1, T r,k,t-1 Indicates the temperature value of the calculation area in the previous time step, T r,t is the temperature data used in the actual calculation process, k represents the surrounding calculation area of ​​the previous time series of the current calculation area, β j,k That is, it is the similarity coefficient under the condition of n times relationship between different j regions around k region and the current calculation region. By combining the time representation of trigonometric functions, setting the numerical limit very close to 1 under different cycle iteration steps, and setting its own cycle and interruption period to be close to the actual calculated temperature change value.

7. The high-efficiency simulation method for grouting at different temperatures based on deep learning as claimed in claim 1 is characterized in that: The second design layer comprises: Among them, T i,t is the temperature change after the slurry interacts with the area to generate heat, T K,t is the temperature data in the surrounding three-dimensional direction, K represents the surrounding calculation area of ​​the current time series of the current calculation area, N represents the number of surrounding calculation areas of the current time series of the current calculation area, ε K,t is the temperature difference reduction coefficient. It is smaller in the direction of grouting slurry flow and larger in the other two directions. r,t-1 Represents the temperature data of the area at time t-1.

8. The high-efficiency simulation method for grouting at different temperatures based on deep learning as claimed in claim 1 is characterized in that: The third design layer comprises: Y(l p ) t =W q,t α q,t m q,t -W q,t-1 α q,t-1 m q,t-1 +b(T r,t -T r,t-1 ); Among them, α q,t Indicates that the time series t is different from m q Yes p The vector gain value, W q,t Represents the weight coefficient of different vector gain values ​​of time series t; m q,t Represents the physical parameter values ​​of the slurry and rock medium in the time series t, including the viscosity μ(T r,t ,t), the porosity of the rock medium φ(T r,t ) t , stress-strain σ(T r,t ) t ; T r,t-1 Represents the temperature data of the area in the t-1 time series; α q,t-1 Indicates that the t-1 time series is different from m q Yes p The vector gain value, W q,t-1 Represents the weight coefficient of different vector gain values ​​of the t-1 time series; m q,t-1 represents the physical parameter values ​​of the slurry and rock medium in the time series t-1, including the viscosity μ(T r,t-1 ,t-1), the porosity of the rock medium φ(T r,t-1 ) t-1 , stress-strain σ(T r,t-1 ) t-1 ; T r,t-1 represents the temperature data of the area in the t-1 time series; Y represents the output data of each time series, l p Represents the vector growth of the vector slurry flow velocity and the permeation diffusion distance under the current time series conditions, m q represents the physical parameter values ​​of the slurry and rock medium normalized under different temperature conditions, α q Indicated in different m q Yes p The vector gain value of b represents the bias under different time series conditions, which changes accordingly according to the temperature.

9. The high-efficiency simulation method for grouting at different temperatures based on deep learning as claimed in claim 1, characterized in that: Improved recurrent neural network after training. The training process includes: Construct a first training set; the first training set is the slurry flow velocity and slurry penetration diffusion distance at different known temperatures; input the first training set into an improved recurrent neural network, the input value of the improved recurrent neural network is temperature, and the output value is the slurry flow velocity and slurry penetration diffusion distance, and train it. When the loss function value of the recurrent neural network no longer decreases, stop training, and obtain the trained recurrent neural network, and the loss function adopts the mean square error loss function.

10. An efficient simulation system for grouting at different temperatures based on deep learning, characterized by: include: A relational formula building module is configured to: build a relational formula for changes in rock and soil medium and grouting materials under different temperature conditions; The relationship between the rock and soil medium and the grouting material under different temperature conditions includes: the relationship between temperature and porosity, the relationship between temperature and stress-strain, and the relationship between temperature and slurry viscosity; based on the actual formation data of the grouting area, fitting the parameters to be determined in the above-mentioned relationship formulas; A network construction module is configured to: construct an improved recurrent neural network; the improved recurrent neural network comprises: a recurrent layer, the recurrent layer comprises: a first design layer, a second design layer and a third design layer connected in sequence, wherein the output end of the second design layer is also connected to the input end of the first design layer; the third design layer is provided with the relationship between the temperature and the porosity, the relationship between the temperature and the stress-strain, and the relationship between the temperature and the slurry viscosity; The output module is configured to: input the temperature of the grouting area to be predicted into the trained improved recurrent neural network to obtain the slurry flow velocity and slurry penetration and diffusion distance of the grouting area to be predicted.

Images