High-speed railway tunnel shotcrete rebound rate control method, terminal and storage medium

Abstract

The present application relates to the technical field of transportation facilities, and specifically discloses a high-speed railway tunnel shotcrete rebound rate control method, terminal and storage medium, which uses artificial intelligence technology based on deep learning to perform correlation analysis on various influencing factors affecting the rebound rate of concrete, optimizes the characteristic expression of various influencing factors based on the implicit correlation mode between various influencing factors, and uses the expected rebound rate as a conditional constraint. Based on the adaptive response relationship between the expected rebound rate and the influencing factors, the corresponding spraying parameters are intelligently predicted to achieve optimized control of the rebound rate of concrete. In this way, the rationality and accuracy of the rebound rate control can be effectively improved, thereby improving the efficiency and economic benefits of the entire construction process.

Description

Technical Field

[0001] The present application relates to the technical field of transportation facilities, and more specifically, to a method, terminal and storage medium for controlling the rebound rate of shotcrete in a high-speed railway tunnel. Background Art

[0002] Shotcrete is a construction material widely used in the field of foundation engineering. During the construction of shotcrete, the rebound rate is an important quality indicator. A high rebound rate means that the concrete has not been fully compacted, affecting the stability and durability of the construction results. A low rebound rate indicates that there is a problem of over-compaction during the concrete spraying process, which also affects the construction results.

[0003] In the process of controlling the rebound rate construction, the prior art will optimize the rebound rate of shotcrete based on data perception and data processing methods. For example, a shotcrete construction optimization system based on digital twins is proposed in Publication No. CN115859832A. This system optimizes the rebound rate of shotcrete through data collection and data simulation, and performs simulation deduction based on the digital twin system. However, in the actual construction process, not only the spraying parameters of the spraying equipment will affect the rebound rate, but also the specific construction environment, humidity, stone particle size and other influencing parameters will also affect the rebound rate. The influencing factors involved are relatively complex. Only controlling the rebound rate based on the spraying parameters such as the spraying path and the spraying pressure cannot achieve the optimal rebound rate control effect.

[0004] In response to this technical problem, China's authorized patent CN117851749B discloses a method and system for controlling the rebound rate of shotcrete. It constructs a predictive regression model by analyzing the relationship between various factors and the rebound rate, and regulates the shotcrete equipment according to the model output results to achieve effective control of the rebound rate.

[0005] In the process of implementing the above technical solution, it was found that the Chinese patent CN117851749B adopted the control variable method to study the relationship between various factors and the rebound rate and obtain the compensation coefficient based on correlation analysis. However, this data analysis method ignores the implicit correlation between various factors. Specifically, there may be local implicit correlations between influencing factors such as construction environment, humidity, stone particle size, injection path and injection air pressure. For example, there may be some implicit correlation between injection air pressure and humidity, and the two as a synergistic whole have a linear or nonlinear effect on the rebound rate. In the research method provided by the Chinese patent CN117851749B, it can only study the correlation between a single factor and the rebound rate through the control variable method, which will lead to the lack of robustness and low precision of the regression accuracy of the regression model.

[0006] At the same time, it is also challenging to accurately screen out the main influencing factors and error influencing factors from multiple influencing factor parameters, and an accurate algorithm is needed to evaluate the contribution and impact of each influencing factor. However, considering the implicit correlation between the various influencing factors, if the implicit correlation is ignored and only a single factor is relied on for contribution evaluation, this evaluation algorithm has essential defects.

[0007] Therefore, a more optimized rebound rate control scheme for shotcrete in high-speed railway tunnels is expected. Summary of the invention

[0008] In order to solve the above technical problems, the present application is proposed. The embodiments of the present application provide a high-speed railway tunnel shotcrete rebound rate control method, terminal and storage medium, which adopts artificial intelligence technology based on deep learning to perform correlation analysis on various influencing factors affecting the concrete rebound rate, optimizes the characteristic expression of each influencing factor based on the implicit correlation mode between each influencing factor, and uses the expected rebound rate as a conditional constraint. Based on the adaptive response relationship between the expected rebound rate and the influencing factors, the corresponding spraying parameters are intelligently predicted to achieve optimized control of the concrete rebound rate. In this way, the rationality and accuracy of the rebound rate control can be effectively improved, thereby improving the efficiency and economic benefits of the entire construction process.

[0009] Accordingly, according to one aspect of the present application, a method for controlling the rebound rate of shotcrete in a high-speed railway tunnel is provided, comprising:

[0010] Acquire a set of influencing factor parameters, wherein the set of influencing factor parameters includes concrete water-to-material ratio, gravel particle size, ambient temperature, and ambient humidity;

[0011] Performing a semantic association pattern analysis between parameters on each influencing factor parameter in the set of influencing factor parameters to obtain an influencing factor parameter semantic association topology matrix;

[0012] Set expected rebound rate;

[0013] One-hot encoding the expected rebound rate to obtain an expected rebound rate label constraint encoding vector;

[0014] Performing fine-grained feature query response analysis on the expected rebound rate label constraint encoding vector and the influencing factor parameter semantic association topology matrix to obtain a query response adaptive matching result representation matrix;

[0015] Based on the query response adaptive matching result representation matrix, an injection parameter decoding value is generated, wherein the injection parameter decoding value includes an initial acceleration value, an injection angle value, and an injection distance value.

[0016] In the above-mentioned high-speed railway tunnel shotcrete rebound rate control method, the semantic association pattern analysis between parameters of each influencing factor parameter in the set of influencing factor parameters is performed to obtain the influencing factor parameter semantic association topological matrix, including: embedding coding of each influencing factor parameter in the set of influencing factor parameters to obtain a set of influencing factor parameter embedded coding vectors; based on the semantic association between each influencing factor parameter embedded coding vector in the set of influencing factor parameter embedded coding vectors, the set of influencing factor parameter embedded coding vectors is associated topologically coded to obtain the influencing factor parameter semantic association topological matrix.

[0017] In the above-mentioned high-speed railway tunnel shotcrete rebound rate control method, each influencing factor parameter in the set of influencing factor parameters is embedded and encoded to obtain a set of influencing factor parameter embedded coding vectors, including: semantically embedding and encoding the text items of each influencing factor parameter in the set of influencing factor parameters to obtain a set of influencing factor text embedded coding vectors; and adding the numerical items of each influencing factor parameter in the set of influencing factor parameters to the tail of the corresponding influencing factor text embedded coding vector in the set of influencing factor text embedded coding vectors to obtain the set of influencing factor parameter embedded coding vectors.

[0018] In the above-mentioned high-speed railway tunnel shotcrete rebound rate control method, based on the semantic association between each influencing factor parameter embedded coding vector in the set of influencing factor parameter embedded coding vectors, the set of influencing factor parameter embedded coding vectors is subjected to association topological coding to obtain the influencing factor parameter semantic association topological matrix, including: calculating the semantic association between any two influencing factor parameter embedded coding vectors in the set of influencing factor parameter embedded coding vectors to obtain a parameter association topological matrix composed of multiple semantic associations; using the set of influencing factor parameter embedded coding vectors as nodes, and the parameter association topological matrix as edges between nodes, inputting the set of influencing factor parameter embedded coding vectors and the parameter association topological matrix into a parameter association pattern feature extractor based on a graph convolutional neural network model to obtain the influencing factor parameter semantic association topological matrix.

[0019] In the above-mentioned high-speed railway tunnel shotcrete rebound rate control method, the expected rebound rate label constraint coding vector and the influencing factor parameter semantic association topology matrix are subjected to fine-grained feature query response analysis to obtain a query response adaptive matching result representation matrix, including: using the expected rebound rate label constraint coding vector as a query feature vector, and using the set of row vectors of the influencing factor parameter semantic association topology matrix as a set of library feature vectors, and inputting the query feature vector and the set of library feature vectors into a query response dynamic encoder based on feature decoupling to obtain the query response adaptive matching result representation matrix.

[0020] In the above-mentioned high-speed railway tunnel shotcrete rebound rate control method, the query feature vector and the set of library feature vectors are input into the query response dynamic encoder based on feature decoupling to obtain the query response adaptive matching result representation matrix, including: respectively performing local semantic association enhancement on each library feature vector in the set of library feature vectors to obtain a set of library feature vector fine-grained decoupled local association feature vectors; calculating the covariance matrix between the query feature vector and each library feature vector fine-grained decoupled local association feature vector in the set of library feature vector fine-grained decoupled local association feature vectors to obtain a set of query response covariance matrices; calculating the matrix trace of each query response covariance matrix in the set of query response covariance matrices to obtain a set of query response trace measurement factors; inputting the set of query response trace measurement factors into a Sigmoid activation function to obtain a set of query response trace attention weights; using each query response trace attention weight in the set of query response trace attention weights as a weight, calculating the weighted sum of the set of query response covariance matrices to obtain the query response adaptive matching result representation matrix.

[0021] In the above-mentioned high-speed railway tunnel shotcrete rebound rate control method, each library feature vector in the set of library feature vectors is locally semantically enhanced to obtain a set of library feature vector fine-grained decoupled local associated feature vectors, including: fine-grained decoupling of the library feature vectors to obtain a subsequence of the library feature vector; the subsequence of the library feature vector is input into a subsequence semantic enhancement encoder based on a Bi-LSTM model to obtain the library feature vector fine-grained decoupled local associated feature vector.

[0022] In the above-mentioned high-speed railway tunnel shotcrete rebound rate control method, the injection parameter decoding value is generated based on the query response adaptive matching result representation matrix, including: inputting the query response adaptive matching result representation matrix into the decoder-based injection parameter generator to obtain the injection parameter decoding value.

[0023] According to another aspect of the present application, a terminal device is provided, which includes: a memory for storing instructions; a processor coupled to the memory, and the processor is configured to execute the high-speed railway tunnel shotcrete rebound rate control method as described above based on the instructions stored in the memory.

[0024] According to another aspect of the present application, a storage medium is provided, on which a high-speed railway tunnel shotcrete rebound rate control program is stored. When the high-speed railway tunnel shotcrete rebound rate control program is executed by a processor, the high-speed railway tunnel shotcrete rebound rate control method as described above is implemented.

[0025] Compared with the prior art, the high-speed railway tunnel shotcrete rebound rate control method, terminal and storage medium provided by the present application use artificial intelligence technology based on deep learning to perform correlation analysis on various influencing factors affecting the concrete rebound rate, optimize the characteristic expression of various influencing factors based on the implicit correlation mode between various influencing factors, and use the expected rebound rate as a conditional constraint. Based on the adaptive response relationship between the expected rebound rate and the influencing factors, the corresponding spraying parameters are intelligently predicted to achieve optimal control of the concrete rebound rate. In this way, the rationality and accuracy of the rebound rate control can be effectively improved, thereby improving the efficiency and economic benefits of the entire construction process. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] By describing the embodiments of the present application in more detail in conjunction with the accompanying drawings, the above and other purposes, features and advantages of the present application will become more apparent. The accompanying drawings are used to provide a further understanding of the embodiments of the present application and constitute a part of the specification. Together with the embodiments of the present application, they are used to explain the present application and do not constitute a limitation of the present application. In the accompanying drawings, the same reference numerals generally represent the same components or steps.

[0027] Figure 1 Flow chart of a method for controlling the rebound rate of shotcrete in a high-speed railway tunnel according to an embodiment of the present application.

[0028] Figure 2 Schematic diagram of the architecture of a method for controlling the rebound rate of shotcrete in a high-speed railway tunnel according to an embodiment of the present application.

[0029] Figure 3 The present invention is a flowchart of performing semantic association pattern analysis on each influencing factor parameter in the set of influencing factor parameters in the high-speed railway tunnel shotcrete rebound rate control method according to an embodiment of the present application to obtain a semantic association topological matrix of the influencing factor parameters.

[0030] Figure 4The present invention is a flowchart of inputting the query feature vector and the set of library feature vectors into a query response dynamic encoder based on feature decoupling to obtain the query response adaptive matching result representation matrix in the high-speed railway tunnel shotcrete rebound rate control method according to an embodiment of the present application.

[0031] Figure 5 A block diagram of a terminal device according to an embodiment of the present application. DETAILED DESCRIPTION

[0032] Below, the exemplary embodiments according to the present application will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments of the present application, and it should be understood that the present application is not limited to the exemplary embodiments described here.

[0033] Example 1

[0034] As mentioned in the above background technology, since the concrete rebound rate is affected by multiple aspects and there are implicit correlations between the various influencing factors, in the prior art, the implicit correlations between the various influencing factors are ignored and the contribution evaluation is only based on a single factor, which has essential defects. In response to the above technical problems, the technical concept of this application is to use artificial intelligence technology based on deep learning to perform correlation analysis on the various influencing factors that affect the concrete rebound rate, optimize the characteristic expression of each influencing factor based on the implicit correlation pattern between the various influencing factors, and use the expected rebound rate as a conditional constraint. Based on the adaptive response relationship between the expected rebound rate and the influencing factors, the corresponding injection parameters are intelligently predicted to achieve optimal control of the concrete rebound rate. In this way, the rationality and accuracy of the rebound rate control can be effectively improved, thereby improving the efficiency and economic benefits of the entire construction process.

[0035] Figure 1 Flow chart of a method for controlling the rebound rate of shotcrete in a high-speed railway tunnel according to an embodiment of the present application. Figure 2 FIG. 1 is a schematic diagram of the structure of a method for controlling the rebound rate of shotcrete in a high-speed railway tunnel according to an embodiment of the present application. Figure 1 and Figure 2As shown, according to the embodiment of the present application, the method for controlling the rebound rate of shotcrete in a high-speed railway tunnel includes the following steps: S110, obtaining a set of influencing factor parameters, wherein the set of influencing factor parameters includes concrete water-to-material ratio, stone particle size, ambient temperature, and ambient humidity; S120, performing parameter semantic association pattern analysis on each influencing factor parameter in the set of influencing factor parameters to obtain an influencing factor parameter semantic association topological matrix; S130, setting an expected rebound rate; S140, performing one-hot encoding on the expected rebound rate to obtain an expected rebound rate label constraint encoding vector; S150, performing fine-grained feature query response analysis on the expected rebound rate label constraint encoding vector and the influencing factor parameter semantic association topological matrix to obtain a query response adaptive matching result representation matrix; S160, generating a spray parameter decoding value based on the query response adaptive matching result representation matrix, wherein the spray parameter decoding value includes an initial acceleration value, a spray angle value, and a spray distance value.

[0036] In the above-mentioned high-speed railway tunnel shotcrete rebound rate control method, the step S110 obtains a set of influencing factor parameters, wherein the set of influencing factor parameters includes concrete water-to-material ratio, stone particle size, ambient temperature, and ambient humidity. It should be understood that the concrete rebound rate is a complex process affected by multiple factors, wherein the raw material quality control, mix design, and on-site construction environment of shotcrete can directly affect the physical properties, fluidity, and curing speed of concrete, thereby indirectly affecting the rebound rate during the spraying process. Based on this, in the technical solution of the present application, by collecting the concrete water-to-material ratio, stone particle size, and the ambient temperature and humidity of the construction site, a targeted optimization strategy is formulated to achieve the expected rebound rate control by adjusting the spraying parameters.

[0037] In the above-mentioned high-speed railway tunnel shotcrete rebound rate control method, the step S120 performs semantic association pattern analysis on each influencing factor parameter in the set of influencing factor parameters to obtain a semantic association topological matrix of influencing factor parameters. It should be understood that since the influencing factors do not exist in isolation, but are intertwined and influence each other, for example, the temperature and humidity of the construction environment synergistically affect the curing rate of concrete, thereby affecting its rebound characteristics. In other words, simply treating each factor as an independent variable will lead to a one-sided and inaccurate understanding of the concrete rebound rate control mechanism. To this end, in the technical solution of the present application, the parameters of each influencing factor are further analyzed for association to reveal the implicit associations between the influencing factors.

[0038] Figure 3The flowchart is a method for controlling the rebound rate of shotcrete in a high-speed railway tunnel according to an embodiment of the present application, in which the semantic association pattern analysis is performed on each influencing factor parameter in the set of influencing factor parameters to obtain a semantic association topological matrix of the influencing factor parameters. Figure 3 As shown, the step S120 includes: S121, embedding and coding each influencing factor parameter in the set of influencing factor parameters to obtain a set of influencing factor parameter embedded coding vectors; S122, based on the semantic association between each influencing factor parameter embedded coding vector in the set of influencing factor parameter embedded coding vectors, performing association topological coding on the set of influencing factor parameter embedded coding vectors to obtain the influencing factor parameter semantic association topological matrix.

[0039] Specifically, the step S121 embeds and codes each influencing factor parameter in the set of influencing factor parameters to obtain a set of embedding coding vectors of influencing factor parameters. It should be understood that since different influencing factor parameters may have different data types and dimensions, in order to convert each influencing factor parameter into a unified, structured vector representation and capture its semantic information, in the technical solution of the present application, each influencing factor parameter is processed separately using the embedding coding technology, and the text items in each influencing factor parameter are semantically embedded and coded, the text content is converted into a numerical vector, and the corresponding numerical items are added to the end thereof, and information integration is performed to obtain a set of embedding coding vectors of influencing factor parameters.

[0040] In a specific example of the present application, the step S121 includes: performing semantic embedding coding on the text items of each influencing factor parameter in the set of influencing factor parameters to obtain a set of influencing factor text embedding coding vectors; adding the numerical items of each influencing factor parameter in the set of influencing factor parameters to the tail of the corresponding influencing factor text embedding coding vector in the set of influencing factor text embedding coding vectors to obtain the set of influencing factor parameter embedding coding vectors.

[0041] Specifically, in step S122, based on the semantic association between each influencing factor parameter embedded coding vector in the set of influencing factor parameter embedded coding vectors, the set of influencing factor parameter embedded coding vectors is subjected to association topological coding to obtain the influencing factor parameter semantic association topological matrix. More specifically, first, the semantic association between any two influencing factor parameter embedded coding vectors in the set of influencing factor parameter embedded coding vectors is calculated to obtain an inter-parameter association topological matrix composed of multiple semantic associations. That is, by calculating the semantic association between each influencing factor parameter embedded coding vector, the association between each influencing factor parameter is quantitatively evaluated, and the interaction strength and pattern between different influencing factors are revealed, thereby constructing an inter-parameter association topological matrix, providing a more in-depth information basis for the optimization control of the rebound rate. In the inter-parameter association topological matrix, a high association value means that there is a close mutual influence between the corresponding influencing factors, while a low association may indicate that the interaction between the two is weak or there is no direct connection.

[0042] Next, the set of influencing factor parameter embedded coding vectors is used as a node, and the inter-parameter association topology matrix is ​​used as an edge between nodes. The set of influencing factor parameter embedded coding vectors and the inter-parameter association topology matrix are input into the inter-parameter association pattern feature extractor based on the graph convolutional neural network model to obtain the influencing factor parameter semantic association topology matrix. That is, the implicit associations between the influencing factors are deeply mined and utilized through the graph convolutional neural network (GCN), and the set of influencing factor parameter embedded coding vectors is feature transferred and updated. Specifically, as a neural network running on a graph structure, GCN can effectively capture the complex relationships between nodes, and update the node representation through information transfer and aggregation between nodes, thereby revealing the potential patterns and laws in the graph structure. In the technical solution of the present application, each influencing factor parameter embedding coding vector is used as a node, the parameter association topological matrix is ​​used as the edge between nodes, and the edge weights between nodes are used to represent the association strength and directionality between each influencing factor. The node information in the graph structure is iteratively updated through a graph convolutional neural network model to achieve deep feature extraction and context information fusion of each influencing factor parameter embedding coding vector, generate an influencing factor parameter semantic association topological matrix, thereby enhancing the feature expression ability of each influencing factor parameter, and contributing to a more comprehensive understanding of the multi-factor synergistic mechanism of rebound rate.

[0043] In the above-mentioned high-speed railway tunnel shotcrete rebound rate control method, the step S130 sets the expected rebound rate. That is, according to the specific construction requirements and engineering goals, a reasonable expected rebound rate is set, the quality control goals and standards are clarified, and it is used as a reference in the construction process to guide the optimization and adjustment of the shotcrete parameters.

[0044] In the above-mentioned high-speed railway tunnel shotcrete rebound rate control method, the step S140 performs one-hot encoding on the expected rebound rate to obtain the expected rebound rate label constraint encoding vector. It should be understood that since the expected rebound rate is a specific numerical target, in order to perform effective feature matching analysis on it and the parameters of various influencing factors, in the technical solution of the present application, the expected rebound rate is converted into the expected rebound rate label constraint encoding vector by one-hot encoding. One-hot encoding is a coding method commonly used to process categorical data, which can convert each category data into a vector form in which only one element is 1 and the rest of the elements are all 0. Here, the expected rebound rate is processed as a category label within the possible range of rebound rate values. Through one-hot encoding, the effective connection between the expected rebound rate and the categorical feature can be achieved, and the expected rebound rate label constraint encoding vector is generated.

[0045] In the above-mentioned high-speed railway tunnel shotcrete rebound rate control method, the step S150 performs fine-grained feature query response analysis on the expected rebound rate label constraint coding vector and the influencing factor parameter semantic association topology matrix to obtain a query response adaptive matching result representation matrix. In a specific example of the present application, the expected rebound rate label constraint coding vector is used as a query feature vector, and the set of row vectors of the influencing factor parameter semantic association topology matrix is ​​used as a set of library feature vectors, and the query feature vector and the set of library feature vectors are input into a query response dynamic encoder based on feature decoupling to obtain the query response adaptive matching result representation matrix. That is, in order to achieve accurate matching and regulation between the expected rebound rate and the influencing factor parameters, a query response dynamic encoder based on feature decoupling is further introduced in the technical solution of the present application to carry out the feature interaction response analysis process. Specifically, the expected rebound rate label constraint encoding vector is used as the query feature vector, and the row vector of the influencing factor parameter semantic association topology matrix (i.e., the updated feature representation of each influencing factor parameter after being processed by the graph convolutional neural network) is used as the set of library feature vectors. The feature query response adaptive encoder first decouples the library feature vector in a fine-grained manner and splits it into vector subsequences to enhance the local information expression ability of the feature. Then, the Bi-LSTM model is used to perform bidirectional encoding on it, capture the forward and backward context information of each subsequence, and realize the local information association integration, so that each library feature vector has stronger expression ability and association at the local level. Then, the covariance matrix and the trace of the covariance matrix between the query feature and each library feature are calculated to quantify the correlation between the two, and the attention weight is generated by this. By weighting and calculating each covariance matrix, the overall matching degree between the query feature and the library feature is comprehensively evaluated, and the adaptive matching result representation matrix is ​​obtained, which provides precise guidance for the subsequent injection parameter regulation.

[0046] Figure 4 The flowchart of the method for controlling the rebound rate of shotcrete in a high-speed railway tunnel according to the embodiment of the present application is to input the query feature vector and the set of library feature vectors into the query response dynamic encoder based on feature decoupling to obtain the query response adaptive matching result representation matrix. Figure 4As shown, the step S150 includes: S151, respectively performing local semantic association enhancement on each library feature vector in the set of library feature vectors to obtain a set of library feature vector fine-grained decoupled local association feature vectors; S152, calculating the covariance matrix between the query feature vector and each library feature vector fine-grained decoupled local association feature vector in the set of library feature vector fine-grained decoupled local association feature vectors to obtain a set of query response covariance matrices; S153, calculating the matrix trace of each query response covariance matrix in the set of query response covariance matrices to obtain a set of query response trace measurement factors; S154, inputting the set of query response trace measurement factors into a Sigmoid activation function to obtain a set of query response trace attention weights; S155, using each query response trace attention weight in the set of query response trace attention weights as a weight, calculating the weighted sum of the set of query response covariance matrices to obtain the query response adaptive matching result representation matrix.

[0047] More specifically, the step S151 includes: performing fine-grained decoupling on the library feature vector to obtain a subsequence of the library feature vector; inputting the subsequence of the library feature vector into a subsequence semantic enhancement encoder based on a Bi-LSTM model to obtain a fine-grained decoupled local correlation feature vector of the library feature vector.

[0048] That is, the step S150 includes: processing the query feature vector and the set of the library feature vector using the following query response encoding formula to obtain the query response adaptive matching result representation matrix, wherein the query response encoding formula is:

[0049]

[0050]

[0051]

[0052]

[0053]

[0054]

[0055]

[0056] in, represents the set of library feature vectors, , , and Respectively represent the first, second, and third and library feature vectors, The value of is the number of library feature vectors in the set of library feature vectors, Indicates decoupling processing, Indicates A subsequence of the library feature vectors, is a subsequence semantic reinforcement encoder based on the Bi-LSTM model. Indicates The library feature vector is fine-grained and decoupled from the local correlation feature vector. represents the query feature vector, represents the covariance calculation function, Indicates The query response covariance matrix, represents the matrix trace calculation function, Indicates query response trace metric, represents the Sigmoid activation function, Indicates The query response trace attention weights, represents dot product, The query response adaptive matching result representation matrix is ​​represented.

[0057] In the above-mentioned high-speed railway tunnel shotcrete rebound rate control method, the step S160 generates a spray parameter decoded value based on the query response adaptive matching result representation matrix, wherein the spray parameter decoded value includes an initial acceleration value, a spray angle value, and a spray distance value. In a specific example of the present application, the implementation method of generating a spray parameter decoded value based on the query response adaptive matching result representation matrix is ​​to input the query response adaptive matching result representation matrix into a decoder-based spray parameter generator to obtain the spray parameter decoded value. Specifically, the decoder includes an input layer, a hidden layer, and an output layer, wherein the input layer receives the query response adaptive matching result representation matrix as input; the hidden layer extracts the complex features of the input data through a multi-layer neural network structure, further explores the deep-level connections between the data, and increases the nonlinear transformation capability of the network by introducing an activation function in the hidden layer, so that it learns more complex mapping relationships. In the output layer, the features extracted from the hidden layer are mapped to the output space of the three injection parameters, namely the initial acceleration value, injection angle value and injection distance value, through three parallel fully connected layers. The optimal injection parameter configuration for the expected rebound rate under different combinations of influencing factors is intelligently predicted, providing scientific guidance for construction personnel.

[0058] Preferably, the present application takes into account that the set of the expected rebound rate label constraint encoding vector and the row vector of the influencing factor parameter semantic association topology matrix respectively express the one-hot encoding features of the expected rebound rate and the semantic features of each influencing factor parameter under the influencing factor parameter semantic association topology. In this way, when the set of the query feature vector and the library feature vector are input into the query response dynamic encoder based on feature decoupling, the query response adaptive matching result representation matrix obtained will have a complex spatial structure in the high-dimensional feature space due to the difference in constraint coverage attention weights, thereby affecting the accuracy of the injection parameter decoding value obtained by the query response adaptive matching result representation matrix through the decoder-based injection parameter generator.

[0059] Therefore, in one example of the present application, passing the query response adaptive matching result representation matrix through a decoder-based injection parameter generator to obtain the injection parameter decoding value includes:

[0060] Calculating the sum of the absolute values ​​of each eigenvalue of the query response adaptive matching result representation matrix to obtain a first query response adaptive matching result space structure value, and calculating the square root of the sum of the squares of each eigenvalue of the query response adaptive matching result representation matrix to obtain a second query response adaptive matching result space structure value;

[0061] multiplying each eigenvalue of the query response adaptive matching result representation matrix by the first query response adaptive matching result space structure value and the second query response adaptive matching result space structure value respectively to obtain a first query response adaptive matching result structure reference value and a second query response adaptive matching result structure reference value corresponding to each eigenvalue;

[0062] multiplying each eigenvalue of the query response adaptive matching result representation matrix by the scale of the query response adaptive matching result representation matrix and the square root of the scale to obtain a first query response adaptive matching result scale transformation value and a second query response adaptive matching result scale transformation value corresponding to each eigenvalue;

[0063] Dividing the first query response adaptive matching result structure reference value by the difference between the first query response adaptive matching result space structure value and the first query response adaptive matching result scale transformation value to obtain a first query response adaptive matching result transformation adjustment value;

[0064] Dividing the second query response adaptive matching result structure reference value by the difference between the second query response adaptive matching result space structure value and the second query response adaptive matching result scale transformation value to obtain a second query response adaptive matching result transformation adjustment value;

[0065] Calculating a weighted sum of the first query response adaptive matching result transformation adjustment value and the second query response adaptive matching result transformation adjustment value to obtain each eigenvalue of the optimized query response adaptive matching result representation matrix;

[0066] The optimized query response adaptive matching result representation matrix is ​​passed through a decoder-based injection parameter generator to obtain the injection parameter decoding value.

[0067] Here, the query response adaptive matching result representation matrix The optimization expression is:

[0068]

[0069]

[0070]

[0071]

[0072]

[0073]

[0074]

[0075]

[0076]

[0077] in, The query response adaptive matching result representation matrix, represents the set of real numbers, and Respectively represent the width and height of the matrix, represents the scale of the query response adaptive matching result representation matrix, The query response adaptive matching result representation matrix is ​​represented by The eigenvalues ​​at the positions, represents the spatial structure value of the adaptive matching result of the first query response, represents the spatial structure value of the adaptive matching result of the second query response, It means point multiplication by position. It means adding by position. represents the second query response adaptive matching result transformation adjustment value corresponding to each feature value, A second query response adaptive matching result transformation adjustment matrix representing each of the second query response adaptive matching result transformation adjustment values, represents the first query response adaptive matching result transformation adjustment value corresponding to each feature value, a first query response adaptive matching result transformation adjustment matrix composed of the first query response adaptive matching result transformation adjustment values, represents the weighted hyperparameter, The optimized query response adaptive matching result representation matrix is ​​represented.

[0078] Therefore, in the preferred example, with respect to the spatial structure information of the feature set of the query response adaptive matching result representation matrix in the high-dimensional space, a scale-based box transformation of each eigenvalue of the query response adaptive matching result representation matrix is ​​performed by taking the class-norm spatial structured representation of the query response adaptive matching result representation matrix as a reference window, and a spatial structure-based box attention weight adjustment of each eigenvalue of the query response adaptive matching result representation matrix is ​​implemented to ensure the invariance of spatial transformation (translation, scaling and rotation) of the query response adaptive matching result representation matrix under feature space interaction, thereby improving the convergence and generalization effects of decoding regression of the feature set of the query response adaptive matching result representation matrix under complex spatial structure representation, and improving the accuracy of the injection parameter decoding value obtained by passing it through a decoder-based injection parameter generator.

[0079] In summary, the high-speed railway tunnel shotcrete rebound rate control method according to the embodiment of the present application is explained, which uses artificial intelligence technology based on deep learning to perform correlation analysis on various influencing factors affecting the concrete rebound rate, optimizes the characteristic expression of each influencing factor based on the implicit correlation mode between each influencing factor, and uses the expected rebound rate as a conditional constraint. Based on the adaptive response relationship between the expected rebound rate and the influencing factor, the corresponding spraying parameters are intelligently predicted to achieve optimal control of the concrete rebound rate. In this way, the rationality and accuracy of the rebound rate control can be effectively improved, thereby improving the efficiency and economic benefits of the entire construction process.

[0080] Example 2

[0081] In order to solve the problems of large rebound of shotcrete, high dosage of accelerator, low wet spraying efficiency, etc., this application intends to optimize the mix ratio of shotcrete through the research and application of new materials such as alkali-free accelerator, rebound modifier, rheology modifier, etc., and combine the spraying parameter optimization control method to greatly improve the problems of large rebound of shotcrete, material waste and extended process caused by frequent pipe blockage during construction.

[0082] 1. Selection of raw materials and mix ratio

[0083] Conventional materials such as cement, aggregates, mixing water, etc.: select materials used in on-site engineering, and select machine-made aggregates to innovate green building materials products for admixtures and additives, as follows:

[0084] Alkali-free accelerator: early strength improver and environmentally friendly alkali-free accelerator are selected. After the early strength improver and alkali-free accelerator are mixed, the setting and strength are required to meet the requirements of GB35159, especially the 1d mortar strength is not less than 7MPa; the environmentally friendly alkali-free accelerator is required to meet the iron standard requirements.

[0085] Rheology modifier: Use a special water reducer for machine-made aggregate shotcrete. The special water reducer for shotcrete is a conventional polycarboxylic acid water reducer + a rheology modifier + a special slow-setting and air-entraining component for shotcrete, which contains a rebound suppression component to reduce rebound, avoid falling blocks and increase the thickness of the spray layer once.

[0086] Green admixtures: Green admixtures are selected. The green admixtures increase the process powder of shotcrete, reduce the amount of performance powder and cement, consume industrial production powder, improve the mechanical properties of shotcrete, save energy, reduce emissions and reduce costs.

[0087] The strength grade of shotcrete is C25, the 1d strength is not less than 10MPa, and the bonding strength with rock is not less than 0.8MPa. The recommended mix ratio is as follows:

[0088]

[0089] 2. Test section materials and on-site injection conditions

[0090] Test section engineering quantity, Langjiafan Tunnel shotcrete quantity (total length 467.18m, including 170m of Class III surrounding rock, 60m of Class IV surrounding rock, 237.18m of Class V surrounding rock, and 4385.9m³ of shotcrete) Estimation: China Railway Shanghai Bureau test section engineering design quantity is 4385.9 / 467.18*2000=18776m 3 The design volume of the test section of China Railway 12th Bureau is 7510m 3 According to the conventional tunnel over-excavation injection volume of 1.8 times, the actual engineering volume is estimated to be 33796m for China Railway Shanghai Bureau 3 , China Railway 12th Bureau is 13518m 3 .

[0091] Use of special innovative materials: green admixture (China Railway 12th Bureau) is 510 tons; rheology modifier: China Railway Shanghai Bureau is 97 tons, China Railway 12th Bureau is 65 tons; modified alkali-free accelerator or environmentally friendly alkali-free accelerator: China Railway Shanghai Bureau is 650 tons, China Railway 12th Bureau is 450 tons.

[0092]

[0093]

[0094] Exemplary terminal equipment

[0095] Below, reference Figure 5 To describe the terminal device according to an embodiment of the present application. Figure 5 FIG. 1 is a block diagram of a terminal device according to an embodiment of the present application. Figure 5 As shown, the terminal device 10 includes one or more processors 11 and a memory 12 .

[0096] The processor 11 may be a central processing unit (CPU) or other forms of processing units having data processing capabilities and / or instruction execution capabilities, and may control other components in the terminal device 10 to perform desired functions.

[0097] The memory 12 may include one or more computer program products, and the computer program product may include various forms of computer-readable storage media, such as volatile memory and / or non-volatile memory. The volatile memory may include, for example, a random access memory (RAM) and / or a cache memory (cache), etc. The non-volatile memory may include, for example, a read-only memory (ROM), a hard disk, a flash memory, etc. One or more computer program instructions may be stored on the computer-readable storage medium, and the processor 11 may run the program instructions to implement the high-speed railway tunnel shotcrete rebound rate control method of each embodiment of the present application described above and / or other desired functions. Various contents such as expected rebound rate and influencing factor parameters may also be stored in the computer-readable storage medium.

[0098] In one example, the terminal device 10 may further include: an input device 13 and an output device 14 , and these components are interconnected via a bus system and / or other forms of connection mechanisms (not shown).

[0099] The input device 13 may include, for example, a keyboard, a mouse, etc.

[0100] The output device 14 can output various information to the outside, including injection parameter decoding values, etc. The output device 14 can include, for example, a display, a speaker, a printer, a communication network and a remote output device connected thereto, and the like.

[0101] Of course, to simplify, Figure 5 Only some of the components in the terminal device 10 related to the present application are shown, and components such as a bus, an input / output interface, etc. are omitted. In addition, according to specific application conditions, the terminal device 10 may also include any other appropriate components.

[0102] Exemplary computer program products and computer-readable storage media

[0103] In addition to the above-mentioned methods and systems, an embodiment of the present application may also be a computer program product, which includes computer program instructions, which, when executed by a processor, enable the processor to execute the steps of the high-speed railway tunnel shotcrete rebound rate control method according to various embodiments of the present application described in the above-mentioned exemplary method section of this specification.

[0104] The computer program product may be written in any combination of one or more programming languages ​​to write program codes for performing the operations of the embodiments of the present application, including object-oriented programming languages, such as Java, C++, etc., and conventional procedural programming languages, such as "C" language or similar programming languages. The program code may be executed entirely on the user computing device, partially on the user device, as an independent software package, partially on the user computing device and partially on a remote computing device, or entirely on a remote computing device or server.

[0105] In addition, an embodiment of the present application may also be a computer-readable storage medium on which computer program instructions are stored. When the computer program instructions are executed by a processor, the processor executes the steps of the high-speed railway tunnel shotcrete rebound rate control method according to various embodiments of the present application described in the above exemplary method section of this specification.

[0106] The computer-readable storage medium may adopt any combination of one or more readable storage media. The readable storage medium may be a readable signal storage medium or a readable storage medium. The readable storage medium may, for example, include but is not limited to an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or device, or any combination of the above. More specific examples (non-exhaustive list) of readable storage media include: an electrical connection with one or more wires, a portable 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 disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above.

[0107] Those of ordinary skill in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of the two. In order to clearly illustrate the interchangeability of hardware and software, the composition and steps of each example have been generally described in terms of function in the above description. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of the present invention.

[0108] Finally, it should be noted that the above embodiments are only used to illustrate the technical solution of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solution of the present invention can be modified or replaced by equivalents without departing from the spirit and scope of the technical solution of the present invention.

Claims

1. A method for controlling the rebound rate of shotcrete in a high-speed railway tunnel, characterized in that: include: Acquire a set of influencing factor parameters, wherein the set of influencing factor parameters includes concrete water-to-material ratio, gravel particle size, ambient temperature, and ambient humidity; Performing a semantic association pattern analysis between parameters on each influencing factor parameter in the set of influencing factor parameters to obtain an influencing factor parameter semantic association topology matrix; Set expected rebound rate; One-hot encoding the expected rebound rate to obtain an expected rebound rate label constraint encoding vector; Performing fine-grained feature query response analysis on the expected rebound rate label constraint encoding vector and the influencing factor parameter semantic association topology matrix to obtain a query response adaptive matching result representation matrix; Based on the query response adaptive matching result representation matrix, an injection parameter decoding value is generated, wherein the injection parameter decoding value includes an initial acceleration value, an injection angle value, and an injection distance value.

2. The high-speed railway tunnel shotcrete rebound rate control method according to claim 1 is characterized in that: Performing a semantic association pattern analysis between parameters on each influencing factor parameter in the set of influencing factor parameters to obtain an influencing factor parameter semantic association topology matrix includes: Embedding and coding each influencing factor parameter in the set of influencing factor parameters to obtain a set of influencing factor parameter embedded coding vectors; Based on the semantic association between each influencing factor parameter embedded coding vector in the set of influencing factor parameter embedded coding vectors, the set of influencing factor parameter embedded coding vectors is subjected to association topological coding to obtain the influencing factor parameter semantic association topological matrix.

3. The high-speed railway tunnel shotcrete rebound rate control method according to claim 2 is characterized in that: Embedding and encoding each influencing factor parameter in the set of influencing factor parameters to obtain a set of influencing factor parameter embedded coding vectors includes: Performing semantic embedding coding on the text items of each influencing factor parameter in the set of influencing factor parameters respectively to obtain a set of influencing factor text embedding coding vectors; The numerical items of each influencing factor parameter in the set of influencing factor parameters are respectively added to the tail of the corresponding influencing factor text embedding coding vector in the set of influencing factor text embedding coding vectors to obtain the set of influencing factor parameter embedding coding vectors.

4. The high-speed railway tunnel shotcrete rebound rate control method according to claim 3 is characterized in that: Based on the semantic association between each influencing factor parameter embedded coding vector in the set of influencing factor parameter embedded coding vectors, the set of influencing factor parameter embedded coding vectors is subjected to associative topological coding to obtain the influencing factor parameter semantic association topological matrix, including: Calculating the semantic association between any two influencing factor parameter embedding coding vectors in the set of influencing factor parameter embedding coding vectors to obtain an inter-parameter association topology matrix composed of multiple semantic associations; Taking the set of influencing factor parameter embedded coding vectors as nodes and the inter-parameter association topological matrix as edges between nodes, the set of influencing factor parameter embedded coding vectors and the inter-parameter association topological matrix are input into the inter-parameter association pattern feature extractor based on the graph convolutional neural network model to obtain the influencing factor parameter semantic association topological matrix.

5. The method for controlling the rebound rate of shotcrete in a high-speed railway tunnel according to claim 4, characterized in that: Performing fine-grained feature query response analysis on the expected rebound rate label constraint encoding vector and the influencing factor parameter semantic association topology matrix to obtain a query response adaptive matching result representation matrix, including: The expected bounce rate label constraint encoding vector is used as the query feature vector, and the set of row vectors of the influencing factor parameter semantic association topology matrix is ​​used as the set of library feature vectors. The query feature vector and the set of library feature vectors are input into a query response dynamic encoder based on feature decoupling to obtain the query response adaptive matching result representation matrix.

6. The high-speed railway tunnel shotcrete rebound rate control method according to claim 5, characterized in that: Inputting the query feature vector and the set of the library feature vector into a query response dynamic encoder based on feature decoupling to obtain the query response adaptive matching result representation matrix, including: Performing local semantic association enhancement on each library feature vector in the set of library feature vectors to obtain a set of library feature vector fine-grained decoupled local association feature vectors; Calculating the covariance matrix between the query feature vector and each library feature vector fine-grained decoupled local correlation feature vector in the set of library feature vector fine-grained decoupled local correlation feature vectors to obtain a set of query response covariance matrices; Calculating the matrix trace of each query response covariance matrix in the set of query response covariance matrices to obtain a set of query response trace measurement factors; Inputting the set of query response trace measurement factors into a sigmoid activation function to obtain a set of query response trace attention weights; Taking each query response trace attention weight in the set of query response trace attention weights as a weight, a weighted sum of the set of query response covariance matrices is calculated to obtain the query response adaptive matching result representation matrix.

7. The method for controlling the rebound rate of shotcrete in a high-speed railway tunnel according to claim 6, characterized in that: Performing local semantic association enhancement on each library feature vector in the set of library feature vectors to obtain a set of library feature vector fine-grained decoupled local association feature vectors, including: Performing fine-grained decoupling on the library feature vector to obtain a subsequence of the library feature vector; The subsequence of the library feature vector is input into a subsequence semantic reinforcement encoder based on a Bi-LSTM model to obtain a fine-grained decoupled local correlation feature vector of the library feature vector.

8. The method for controlling the rebound rate of shotcrete in a high-speed railway tunnel according to claim 7, characterized in that: Based on the query response adaptive matching result representation matrix, generating injection parameter decoding values, including: The query response adaptive matching result representation matrix is ​​input into a decoder-based injection parameter generator to obtain the injection parameter decoding value.

9. A terminal device, characterized in that: include: A memory for storing instructions; A processor is coupled to the memory, and the processor is configured to execute the method for controlling the rebound rate of shotcrete in a high-speed railway tunnel according to any one of claims 1 to 8 based on the instructions stored in the memory.

10. A storage medium, characterized in that: The storage medium stores a high-speed railway tunnel shotcrete rebound rate control program, and when the high-speed railway tunnel shotcrete rebound rate control program is executed by the processor, the high-speed railway tunnel shotcrete rebound rate control method according to any one of claims 1 to 8 is implemented.

Images