Tunnel water inrush prediction method based on multi-source information fusion

Abstract

This invention discloses a method for predicting tunnel inrush water volume based on multi-source information fusion. The method includes acquiring geological information along the tunnel using multiple methods; extracting feature information related to lithology, geological structure, water pressure, and tunnel depth, as well as the tunnel's location coordinates, from the geological information; fusing the extracted feature information; constructing training and testing sets for a prediction model using the fused output data; using the PSO-RF algorithm as the prediction model for inrush water volume; training and testing the prediction model; and using the qualified prediction model to predict the tunnel inrush water volume. This invention combines an intelligent prediction model with multi-source geological information input to predict tunnel inrush water volume, thereby improving the accuracy of inrush water volume prediction.

Description

Technical Field

[0001] This invention relates to the field of tunnel water inrush prevention and control technology, and in particular to a method for predicting the volume of sudden water inrush in tunnels. Background Technology

[0002] Water inrush is the phenomenon of groundwater continuously flowing into a site during mining, excavation of foundation pits, or underground caverns in rock (soil) masses below the groundwater level. Sources of water inrush include groundwater (pore water, fissure water, karst water) and surface water.

[0003] Water inrush is a common engineering geological problem encountered in construction projects. It not only poses difficulties for construction but is also a significant cause of instability in the surrounding rock of underground caverns, seepage damage in dam foundation pits, or slope collapse. Therefore, predicting water inrush volume during tunnel excavation is crucial for ensuring construction safety.

[0004] In existing technologies, the prediction of water inflow relies solely on monitoring data of water inflow at the construction site, which makes it difficult to accurately predict the water inflow. Summary of the Invention

[0005] In view of this, the purpose of this invention is to provide a method for predicting tunnel inrush water volume based on multi-source information fusion, so as to solve the technical problem of improving the accuracy of water inrush volume prediction during tunnel construction.

[0006] This invention relates to a method for predicting tunnel inrush water volume based on multi-source information fusion, characterized by the following steps:

[0007] 1) Geological information along the tunnel route was obtained using geological drilling, geological survey, geological mapping, ground-penetrating radar, seismic methods, and advanced drilling.

[0008] 2) Extract the feature information related to the four attributes of lithology, geological structure, water pressure and tunnel burial depth from the geological information obtained in step 1), as well as the tunnel location coordinates corresponding to the feature information, and encode the extracted feature information as feature values;

[0009] 3) For the feature values ​​obtained in step 2), feature values ​​with the same attributes and the same tunnel location coordinates are fused, including:

[0010] Establish the recognition framework Θ = {θ1, θ2, ..., θ k}, θ j This indicates the specific identification result, that is, it is identified as belonging to one of the following categories: lithology, geological structure, water pressure, and tunnel burial depth; the characteristic values ​​obtained in step 2) are used as evidence, and each piece of evidence is E1, E2, ..., E i The corresponding probability assignment functions are m1, m2, ..., m i , i.e. m i (θj )=P(E i |θ j ), each m i The probability distribution defined on the recognition framework Θ represents the evidence E. i The degree of support for the recognition results; A is the corresponding focal element. and The fused probability assignment function m is calculated using the following formula:

[0011]

[0012] Where K is a normalization constant used to ensure that the sum of the probability allocation functions after fusion is 1:

[0013]

[0014] ④ Calculate the average evidence:

[0015]

[0016] In the formula, k is the number of focal elements; n is the number of pieces of evidence;

[0017] ⑤ Calculate the distance between a single piece of evidence and the average amount of evidence:

[0018]

[0019] The greater the distance, the weaker the correlation between the two pieces of evidence; to give high-confidence evidence a high weight coefficient, the constructor makes the weight coefficient of the evidence correlate with the distance d. i Inversely proportional, the mass function of different focal elements is:

[0020]

[0021] ③ To objectively and truthfully reflect the importance of each source of evidence, the entropy method is used to determine the weight of each piece of evidence. The expression for the weight coefficient is as follows:

[0022]

[0023] In the formula,

[0024] ⑥ Using the weighting coefficient ω i Weighting the initial sources of evidence, i.e.

[0025]

[0026] New evidence m new As a source of evidence, substitute into equation (1) to perform pairwise fusion, and output a fused data for each fusion;

[0027] The four attribute data corresponding to the same tunnel location coordinates are fused to form a sequence {a1, a2, a3, a4}, where a1 represents lithological fusion information, a2 represents geological structure fusion information, a3 represents water pressure fusion information, and a4 represents tunnel burial depth fusion information, thus obtaining a dataset composed of several sequences.

[0028] 4) The PSO-RF algorithm model was used as the prediction model for water inflow, and the prediction model was trained and tested:

[0029] The dataset obtained in step 3) is divided into a training set and a test set. The PSO-RF algorithm model is trained using the training set, and the trained model is tested using the test set. A sequence of numbers is used as an input to the PSO-RF algorithm model. The goodness of fit and mean squared error are selected as evaluation indicators for model training to obtain a prediction model that has passed both training and testing.

[0030] 5) Use the qualified prediction model obtained in step 4) to predict the tunnel water inflow.

[0031] The beneficial effects of this invention are:

[0032] This invention relates to a method for predicting tunnel inrush water volume based on multi-source information fusion. It obtains various geological information along the tunnel through multiple detection methods and then fuses this information using a multi-source information fusion method to obtain a fused dataset that reflects the tunnel inrush water volume. This fused dataset is then used to train an inrush water volume prediction model that shows a good fit between predicted and actual values. Using this prediction model to predict inrush water volume, combined with multi-source geological information input, can improve the accuracy of inrush water volume prediction. Attached Figure Description

[0033] Figure 1 This is a flowchart of the overall process for predicting tunnel inrush water volume based on multi-source information fusion.

[0034] Figure 2 This is a flowchart of the training process for a water inflow prediction model based on the PSO-RF algorithm.

[0035] Figure 3 This is a graph showing the prediction results for the test set. Detailed Implementation

[0036] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0037] The tunnel inrush water volume prediction method based on multi-source information fusion in this embodiment includes the following steps:

[0038] 1) Geological information along the tunnel route is obtained using methods such as geological drilling, geological survey, geological mapping, ground-penetrating radar, seismic methods, and advanced drilling. This geological information includes geological profile information, topographic information, tunnel face exposure information, and geophysical information.

[0039] 2) Extract the feature information related to the four attributes of lithology, geological structure, water pressure and tunnel depth from the geological information obtained in step 1), as well as the tunnel location coordinates corresponding to the feature information, and encode the extracted feature information as feature values.

[0040] For example, the geological profile information obtained by various detection methods includes the depth of the bottom layer of the rock strata, the thickness of the rock strata, and the geological properties of the rock strata, which are lithology-related feature information. Topographic information, including topographic relief and the distribution of surface water systems, is information related to geological structure. Topographic information, including surface water level and groundwater level, is information related to water pressure. The information revealed at the tunnel face also includes information related to lithology, geological structure, and water pressure. Geophysical information, including detected fault information, fracture distribution information, and water-rich area information, belongs to geological structure information. The detected water level in water-rich areas belongs to different water pressure information. An example of extracting feature information from geological information obtained by drilling is shown in Table 1 below:

[0041] Table 1. Results of LSSK51 borehole information extraction

[0042]

[0043] The tunnel location coordinates can be extracted using CAD software. An example of tunnel coordinate extraction is shown in Table 2 below.

[0044] Table 2 Three-dimensional coordinates of a certain tunnel route

[0045]

[0046] 3) For the feature values ​​obtained in step 2), feature values ​​with the same attributes and the same tunnel location coordinates are fused, including:

[0047] Establish the recognition framework Θ = {θ1, θ2, ..., θ k}, θ j This indicates the specific identification result, that is, it is identified as belonging to one of the following categories: lithology, geological structure, water pressure, and tunnel burial depth; the characteristic values ​​obtained in step 2) are used as evidence, and each piece of evidence is E1, E2, ..., E i The corresponding probability assignment functions are m1, m2, ..., m i , i.e. m i (θ j )=P(E i |θ j ), each m iThe probability distribution defined on the recognition framework Θ represents the evidence E. i The degree of support for the recognition results; A is the corresponding focal element. and The fused probability assignment function m is calculated using the following formula:

[0048]

[0049] Where K is a normalization constant used to ensure that the sum of the probability allocation functions after fusion is 1:

[0050]

[0051] ⑦ Calculate the average evidence:

[0052]

[0053] In the formula, k is the number of focal elements; n is the number of pieces of evidence;

[0054] ⑧ Calculate the distance between a single piece of evidence and the average amount of evidence:

[0055]

[0056] The greater the distance, the weaker the correlation between the two pieces of evidence; to give high-confidence evidence a high weight coefficient, the constructor makes the weight coefficient of the evidence correlate with the distance d. i Inversely proportional, the mass function of different focal elements is:

[0057]

[0058] ③ To objectively and truthfully reflect the importance of each source of evidence, the entropy method is used to determine the weight of each piece of evidence. The expression for the weight coefficient is as follows:

[0059]

[0060] In the formula,

[0061] ⑨ Using the weighting coefficient ω i Weighting the initial sources of evidence, i.e.

[0062]

[0063] New evidence m new As a source of evidence, substitute into equation (1) to perform pairwise fusion, and output a fused data for each fusion;

[0064] The four attribute data corresponding to the same tunnel location coordinates are fused into a sequence {a1, a2, a3, a4}, where a1 represents lithological fusion information, a2 represents geological structure fusion information, a3 represents hydraulic pressure fusion information, and a4 represents tunnel burial depth fusion information. This results in a dataset composed of several sequences. Data samples are shown in Table 3 below.

[0065] Table 3 Sample of fused data

[0066]

[0067] 4) The PSO-RF algorithm model was used as the prediction model for water inflow, and the prediction model was trained and tested:

[0068] The dataset obtained in step 3) is divided into a training set and a test set. The PSO-RF algorithm model is trained using the training set, and the trained model is tested using the test set. A sequence of numbers is used as an input to the PSO-RF algorithm model. Goodness of fit and mean squared error are selected as evaluation metrics for model training, resulting in a qualified prediction model that has passed both training and testing. The model training process is as follows: Figure 2 As shown, the test results are as follows: Figure 3 As shown, the goodness of fit (R²) between the predicted and actual values ​​is... 2 The mean squared error (MSE) and the mean squared error (MSE) were 0.9711 and 4.3244, respectively, indicating a good fit. This shows that the model achieved high prediction accuracy, thus verifying the feasibility of the model for predicting water inflow at the tunnel face.

[0069] 5) Use a trained prediction model to predict the tunnel water inflow.

[0070] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A method for predicting tunnel inrush water volume based on multi-source information fusion, characterized in that: Includes the following steps: 1) Geological information along the tunnel route was obtained using geological drilling, geological survey, geological mapping, ground-penetrating radar, seismic methods, and advanced drilling. 2) Extract the feature information related to the four attributes of lithology, geological structure, water pressure and tunnel burial depth from the geological information obtained in step 1), as well as the tunnel location coordinates corresponding to the feature information, and encode the extracted feature information as feature values; 3) For the feature values ​​obtained in step 2), feature values ​​with the same attributes and the same tunnel location coordinates are fused, including: Establish the recognition framework Θ = {θ1, θ2, ..., θ k }, θ j This indicates the specific identification result, that is, it is identified as belonging to one of the following categories: lithology, geological structure, water pressure, and tunnel burial depth; the characteristic values ​​obtained in step 2) are used as evidence, and each piece of evidence is E1, E2, ..., E i The corresponding probability assignment functions are m1, m2, ..., m i , i.e. m i (θ j )=P(E i |θ j ), each m i The probability distribution defined on the recognition framework Θ represents the evidence E. i The degree of support for the recognition results; A is the corresponding focal element. and The fused probability assignment function m is calculated using the following formula: Where K is a normalization constant used to ensure that the sum of the probability allocation functions after fusion is 1: ① Calculate the average evidence: In the formula, k is the number of focal elements; n is the number of pieces of evidence; ② Calculate the distance between a single piece of evidence and the average amount of evidence: The greater the distance, the weaker the correlation between the two pieces of evidence; to give high-confidence evidence a high weight coefficient, the constructor makes the weight coefficient of the evidence correlate with the distance d. i Inversely proportional, the mass function of different focal elements is: ③ To objectively and truthfully reflect the importance of each source of evidence, the entropy method is used to determine the weight of each piece of evidence. The expression for the weight coefficient is as follows: In the formula, ③ Using the weighting coefficient ω i Weighting the initial sources of evidence, i.e. New evidence m new As a source of evidence, substitute into equation (1) to perform pairwise fusion, and output a fused data for each fusion; The four attribute data corresponding to the same tunnel location coordinates are fused to form a sequence {a1, a2, a3, a4}, where a1 represents lithological fusion information, a2 represents geological structure fusion information, a3 represents water pressure fusion information, and a4 represents tunnel burial depth fusion information, thus obtaining a dataset composed of several sequences. 4) The PSO-RF algorithm model was used as the prediction model for water inflow, and the prediction model was trained and tested: The dataset obtained in step 3) is divided into a training set and a test set. The PSO-RF algorithm model is trained using the training set, and the trained model is tested using the test set. A sequence of numbers is used as an input to the PSO-RF algorithm model. The goodness of fit and mean squared error are selected as evaluation indicators for model training to obtain a prediction model that has passed both training and testing. 5) Use the qualified prediction model obtained in step 4) to predict the tunnel water inflow.

Images