Tunnel surrounding rock identification method and system

Abstract

The application provides a tunnel surrounding rock identification method and system, and relates to the technical field of tunnel construction, comprising: acquiring mixed frequency acoustic wave reflection signals collected by a sensor array arranged on a tunnel face, and hyperspectral image data of the tunnel face rock mass in the visible light to short wave infrared band synchronously acquired by a portable spectrometer; constructing a topological pattern according to the mixed frequency acoustic wave reflection signals to obtain topological invariant features; performing cross-modal fusion according to the topological invariant features and the hyperspectral image data to obtain a fused topological-spectral correlation graph; performing implicit manifold structure learning according to the fused topological-spectral correlation graph to obtain a low-dimensional feature vector; performing dynamic surrounding rock classification according to the low-dimensional feature vector to obtain a surrounding rock dynamic classification label; and performing identification result decision according to the surrounding rock dynamic classification label to obtain a tunnel surrounding rock identification result. The application realizes high-precision and high-robustness dynamic identification of the tunnel surrounding rock state.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of wireless network control, in particular to a tunnel surrounding rock identification method and system. BACKGROUND

[0002] Tunnel surrounding rock identification is a key technical link for safe construction and dynamic design of tunnel engineering. For a long time, it mainly relies on traditional methods such as geological logging and drilling core taking. These methods are intuitive but have limitations such as low efficiency, strong subjectivity and difficulty in real-time feedback. In recent years, with the development of geophysical exploration and intelligent algorithms, surrounding rock identification technologies based on single sensing data such as acoustic waves and spectra have emerged. Through signal processing or machine learning models, data features are extracted and classified, which to some extent improves the automation level of identification. However, such existing technologies usually process different physical property detection data in isolation, and fail to achieve deep-level fusion of multi-modal information from the principle level, making it difficult to capture the internal relationship between rock mass structure, mineral composition and stress state. At the same time, most methods do not consider the dynamic evolution characteristics of surrounding rock state over time, often based on static snapshots for analysis, ignoring the time sequence continuity of surrounding rock response in the tunnel excavation process, resulting in jumping of the identification results and difficulty in reflecting the evolution trend. In addition, the decision-making level often relies on simple threshold or preset rules, and the modeling ability of data itself uncertainty and geological background continuity is weak, so the reliability and robustness of the identification results under complex geological conditions need to be improved.

[0003] Based on the above shortcomings of the prior art, there is an urgent need for a tunnel surrounding rock identification method and system. SUMMARY

[0004] The purpose of the present application is to provide a tunnel surrounding rock identification method to improve the above problems. In order to achieve the above purpose, the technical scheme adopted by the present application is as follows:

[0005] In a first aspect, the present application provides a tunnel surrounding rock identification method, comprising:

[0006] acquiring mixed frequency acoustic wave reflection signals collected by a sensor array arranged at a tunnel face, and hyperspectral image data of the tunnel face rock mass in the visible light to short wave infrared band acquired synchronously by a portable spectrometer;

[0007] constructing a topological pattern according to the mixed frequency acoustic wave reflection signals, constructing an acoustic wave response persistence topological barcode by analyzing multi-frequency response, and obtaining a topological invariant feature representing the crack and pore structure of the rock mass;

[0008] performing cross-modal fusion according to the topological invariant feature and the hyperspectral image data, associating the persistence interval of the topological barcode with the mineral feature absorption band, and obtaining a fusion topological-spectrum correlation graph;

[0009] learning an implicit manifold structure according to the fusion topological-spectral correlation graph, obtaining a low-dimensional feature vector reflecting a potential instability mode of the rock mass by calculating a geodesic distance of a correlation graph node on a nonlinear manifold;

[0010] performing dynamic surrounding rock category division according to the low-dimensional feature vector, obtaining a dynamic classification label of the surrounding rock by density clustering and introducing a concept of a sliding window in time series analysis to deduce time sequence evolution characteristics of a state of the surrounding rock in a tunnel excavation process;

[0011] performing recognition result decision-making according to the dynamic classification label of the surrounding rock, obtaining a final tunnel surrounding rock recognition result by confidence weighted fusion of multiple time sequence classification results.

[0012] In a second aspect, the application further provides a tunnel surrounding rock recognition system, comprising:

[0013] an acquisition module configured to acquire mixed frequency acoustic wave reflection signals collected by a sensor array arranged on a tunnel face and high spectral image data of the rock mass on the tunnel face in a visible light to short wave infrared band acquired synchronously by a portable spectrometer;

[0014] a construction module configured to construct a topological mode according to the mixed frequency acoustic wave reflection signals, to obtain topological invariant features representing crack and pore structures of the rock mass by constructing acoustic wave response persistence topological barcodes through analysis of multi-frequency responses;

[0015] a fusion module configured to perform cross-modal fusion according to the topological invariant features and the high spectral image data, to obtain a fusion topological-spectral correlation graph by correlating persistence intervals of the topological barcodes with mineral feature absorption bands;

[0016] a learning module configured to learn an implicit manifold structure according to the fusion topological-spectral correlation graph, to obtain a low-dimensional feature vector reflecting a potential instability mode of the rock mass by calculating a geodesic distance of a correlation graph node on a nonlinear manifold;

[0017] a division module configured to perform dynamic surrounding rock category division according to the low-dimensional feature vector, to obtain a dynamic classification label of the surrounding rock by density clustering and introducing a concept of a sliding window in time series analysis to deduce time sequence evolution characteristics of a state of the surrounding rock in a tunnel excavation process;

[0018] an output module configured to perform recognition result decision-making according to the dynamic classification label of the surrounding rock, to obtain a final tunnel surrounding rock recognition result by confidence weighted fusion of multiple time sequence classification results.

[0019] The application has the following beneficial effects:

[0020] The present application realizes high-precision and high-robust dynamic identification of the tunnel surrounding rock state by constructing topological invariant features from multi-frequency acoustic waves and hyperspectral data, cross-modal fusion of the topological invariant features to learn the implicit manifold structure of the surrounding rock, and dynamic category division and confidence optimization combined with time series analysis. BRIEF DESCRIPTION OF DRAWINGS

[0021] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the following will briefly introduce the drawings needed to be used in the embodiments. It should be understood that the following drawings only show some embodiments of the present application, and therefore should not be regarded as a limitation on the scope. For those skilled in the art, other related drawings can also be obtained without creative labor on the basis of these drawings.

[0022] Figure 1 A flowchart of a tunnel surrounding rock identification method described in the embodiments of the present application;

[0023] Figure 2 A structural schematic diagram of a tunnel surrounding rock identification system described in the embodiments of the present application;

[0024] Figure 3 A structural schematic diagram of a tunnel surrounding rock identification device described in the embodiments of the present application;

[0025] Figure 4 A sensor array layout diagram.

[0026] In the figure, 800 is a tunnel surrounding rock identification device, 801 is a processor, 802 is a memory, 803 is a multimedia component, 804 is an I / O interface, 805 is a communication component, 901 is an acquisition module, 902 is a construction module, 903 is a fusion module, 904 is a learning module, 905 is a division module, and 906 is an output module. DETAILED DESCRIPTION

[0027] In order to make the purpose, technical solutions and advantages of the embodiments of the present application more clear, the following will combine the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are some of the embodiments of the present application, but not all the embodiments. The components of the embodiments of the present application described and shown in the drawings can be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present application provided in the drawings is not intended to limit the scope of the claimed present application, but only represents selected embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the present application.

[0028] It should be noted that similar reference numerals and letters refer to like items in the accompanying drawings and that, once an item is defined in one drawing, it should not require further defining and explaining in the subsequent drawings. Meanwhile, in the description of the present application, the terms "first", "second", and the like are merely used to distinguish descriptions, and cannot be understood as indicating or implying relative importance.

[0029] Embodiment 1

[0030] The embodiment provides a tunnel surrounding rock identification method.

[0031] Referring to Figure 1 , the method comprises steps S100 to S600.

[0032] Step S100, acquiring mixed frequency acoustic wave reflection signals collected by a sensor array arranged on a tunnel face, and hyperspectral image data of the face rock mass in a visible light to short wave infrared band synchronously acquired by a portable spectrometer;

[0033] It can be understood that the core of the present step is to synchronously acquire two types of physical data with essential differences: the acoustic wave reflection signals reflect the macro mechanical structure and internal defects of the rock mass, and the hyperspectral data reveal the mineral chemical composition of the rock mass surface. Specifically, the sensor array arranged on the tunnel face is used to collect acoustic wave signals, and the necessity thereof lies in that a single sensor can only provide point information, while an array arrangement can capture the response field distribution of the acoustic wave propagation in the rock mass space through multi-probe synchronous reception, which is the spatial distribution data necessary for constructing subsequent topological features; preferably, as shown in Figure 4 the sensor array can be arranged in a ring or grid shape, which can effectively cover the key areas such as the arch and sidewall of the face and ensure the uniformity of spatial sampling, avoiding the occurrence of detection blind area. The acquired mixed frequency acoustic wave reflection signals should include specifically designed low frequency components (for penetrating the rock mass and reflecting the macroscopic crack structure) and high frequency components (more sensitive to microscopic pores and small defects), which together constitute the original data set reflecting the multi-scale structural characteristics of the rock mass. Synchronously, through the hyperspectral imager located inside the face near the bottom, the hyperspectral image data of the face rock mass in the visible light to short wave infrared band are acquired, and the hyperspectral image data should cover the visible light to short wave infrared band, which has characteristic absorption bands for common altered minerals such as water-bearing minerals and clay minerals in the rock mass, thereby being able to provide detailed mineral composition information. The strict synchronization of the two types of data in space and time ensures that the physical structure and chemical composition information correspond to the same rock mass region, providing an accurate matching data basis for subsequent cross-modal deep fusion.

[0034] Step S200, constructing a topological pattern according to the mixed frequency acoustic wave reflection signal, constructing an acoustic wave response persistence topological barcode by analyzing the multi-frequency response, and obtaining a topological invariant feature representing the fracture and pore structure of the rock mass;

[0035] It should be noted that this step uses topology as a mathematical tool to characterize the acoustic wave response, which is different from the conventional signal feature extraction method. The topological pattern construction does not focus on the specific signal amplitude or frequency point, but analyzes the global invariance features such as connectivity and shape of the rock mass pore and fracture structure revealed by different frequency responses. This processing method makes the description of the rock mass structure more robust, and can bypass the details such as signal attenuation in complex media, and directly capture the inherent topological structure information.

[0036] Step S300, cross-modal fusion according to the topological invariant feature and hyperspectral image data, obtaining a fusion topological-spectral correlation graph by associating the persistence interval of the topological barcode with the mineral characteristic absorption band;

[0037] It can be understood that this step associates and maps the "persistence" feature of the topological barcode representing the physical structure of the rock mass with the spectral absorption "band" representing the mineral composition. This is essentially to explore the internal relationship between the mechanical structure integrity of the rock mass and its key mineral composition (such as altered minerals), thereby constructing a unified correlation graph that can reflect the physical state and chemical properties of the rock mass.

[0038] Step S400, implicit manifold structure learning according to the fusion topological-spectral correlation graph, obtaining a low-dimensional feature vector reflecting the potential instability mode of the rock mass by calculating the geodesic distance of the correlation graph nodes on the nonlinear manifold;

[0039] It should be noted that this step is based on a basic assumption that although the state data of the surrounding rock presents a complex distribution in a high-dimensional space, the essential factors determining the stability exist in a low-dimensional nonlinear manifold. The purpose of learning this implicit manifold structure through geodesic distance calculation is to reduce the dimension of the high-dimensional and complex correlation graph data and map it to a feature space that can more intuitively reflect the nature of the rock mass stability. This enables the subsequent classification to ignore redundant information and focus directly on the core features most relevant to the potential instability mode.

[0040] Step S500, dynamic surrounding rock classification according to the low-dimensional feature vector, obtaining a surrounding rock dynamic classification label by density clustering and introducing the concept of sliding window in time series analysis to deduce the time sequence evolution characteristics of the surrounding rock state in the tunneling process;

[0041] It can be understood that this step introduces a dynamic perspective to cope with the engineering reality that the state of surrounding rock during tunneling is not static but evolves over time. By combining the sliding window time density clustering, this method not only classifies the current state of surrounding rock, but also focuses on capturing the evolution law of state category with gradual change, mutation or continuation of the tunneling process. This makes the recognition result no longer an isolated snapshot, but a dynamic sequence containing the trend information of the behavior of surrounding rock, which is more in line with the dynamic needs of engineering decision-making.

[0042] Step S600, according to the dynamic classification label of surrounding rock, the recognition result decision is made, the multiple time sequence classification results are fused by confidence weighted, and the final tunnel surrounding rock recognition result is obtained.

[0043] It should be noted that due to the inherent uncertainty of geological identification, the classification result of a single moment may not be reliable. Therefore, by introducing confidence evaluation and optimization and fusing the decision results of multiple time windows, this method aims to improve the robustness and reliability of the final output result. This weighted decision mechanism based on time sequence context simulates the process of experienced geologists comprehensively analyzing and judging multiple information, so as to realize more prudent and reliable automatic identification at the algorithm level.

[0044] Further, step S200 includes steps S210 to S230.

[0045] Step S210, according to the mixed frequency acoustic wave reflection signal, frequency domain feature separation processing is carried out, the signal is adaptively decomposed into isolated frequency band components corresponding to macroscopic crack scattering and microscopic pore resonance respectively by matching tracking algorithm, and multi-scale acoustic wave response mode is obtained;

[0046] Step S220, according to the multi-scale acoustic wave response mode, preliminary calculation processing of topological invariant is carried out, alpha complex is constructed by constructing the spatial distribution of response amplitude of each mode, and the change of each order Betti number with distance parameter is calculated, and a multi-scale topological invariant sequence is obtained;

[0047] Step S230, according to the multi-scale topological invariant sequence, topological feature integration and coding processing is carried out, the Betti number bar code under different scales is aligned and fused according to its continuity in the direction of tunnel stress field to form acoustic wave response continuity topological bar code, and topological invariant feature is obtained.

[0048] Specifically, the flow starts with the frequency domain feature separation process in step S210, which is the key to adaptively decompose the mixed frequency signal into a series of isolated band components using the matching pursuit algorithm; this process is not a simple filtering, but according to the physical mechanism of different scale structures in the rock medium scattering of acoustic waves, the response signal is separated into different modes corresponding to the macroscopic crack scattering effect and the microscopic pore resonance effect respectively, thereby realizing the targeted analysis of the multi-scale structure of the rock mass in principle. Based on the multi-scale acoustic response modes obtained by separation, step S220 performs preliminary calculation of topological invariants, and an alpha complex is constructed for the amplitude spatial distribution received by the sensor array under each mode; the alpha complex is a mathematical tool for constructing geometric shapes based on the spatial distance and attribute threshold (in this implementation, the acoustic amplitude) of the point set, by systematically changing the distance parameter and calculating the Betti number (an invariant for quantifying the topological structure such as the number of connected regions and tunnel-like cavities) at each order as the parameter changes, a sequence of multi-scale topological invariants is obtained, and this process converts the complex spatial amplitude distribution into a quantitative topological descriptor that describes the connectivity and cavity characteristics of the rock mass structure. Step S230 then integrates the features based on the characteristics of the stress environment of the tunnel engineering, by aligning and fusing the Betti number barcodes (charts recording the emergence and disappearance of topological features with the observation scale) obtained at different scales along the dominant direction of the stress field caused by tunnel excavation, and finally forming a persistent topological barcode of acoustic response; this process not only integrates multi-scale information, but also makes the output topological invariant features contain the response characteristics of the rock mass structure to the engineering stress environment, enhancing the relevance between the features and the stability of the surrounding rock. The entire processing chain embodies the progressive idea from signal decomposition to geometric structure abstraction and then to feature fusion in the engineering context, aiming to extract topological descriptors that are sensitive to changes in rock mass structure and have clear engineering significance. The alpha complex construction formula is represented as:

[0049] ;

[0050] wherein, represents the alpha complex constructed under the scale , the amplitude threshold , and the distance scale ; represents the acoustic response mode under the th scale; represents the point set composed of the response amplitudes of mode at the spatial positions of the sensor array on the tunnel face; represents the acoustic response amplitude at point ; represents a two-dimensional circle with point as the center and as the radius; a threshold parameter based on the amplitude of the acoustic wave response; represents a simplex (such as a point, a line segment, a triangle, etc.) in an alpha complex; represents a distance scale parameter; represents that the condition must be true for all vertices of the simplex .

[0051] The formula for generating a sequence of multi-scale topological invariants is:

[0052] ;

[0053] In the formula, represents the Betti number of the complex of the th order at the scale and the distance parameter ; is the th homology group of the complex ; , is the th and the th boundary operator; is the kernel of the boundary operator , representing all -dimensional cycles; is the image of the boundary operator , representing -dimensional cycles that are actually the boundary of some -dimensional entity; represents the dimension.

[0054] Further, the step S300 includes steps S310 to S330.

[0055] Step S310, according to the hyperspectral image data, a mineral composition quantitative inversion processing is performed, the characteristic absorption band depth and area of a preset type of altered mineral are analyzed based on a nonlinear spectral unmixing model, and a mineral composition quantitative distribution map of the rock mass of the working face is obtained;

[0056] Step S320, according to the mineral composition quantitative distribution map and the topological invariant feature, a cross-modal correlation relationship construction processing is performed, a nonlinear mapping relationship between different mineral combinations and topological barcode persistence intervals is modeled through a graph attention network, and a tunnel axis direction is introduced as a spatial constraint, and a preliminary topological-spectral correlation feature tensor is obtained;

[0057] Step S330, according to the preliminary topological-spectral correlation feature tensor, a correlation relationship optimization and a graph generation processing are performed, multi-modal information is fused through a tensor decomposition method, and the topological persistence is used as a weight to optimize the correlation strength, and a fusion topological-spectral correlation graph is generated.

[0058] Specifically, the first step S310 quantitatively analyzes the hyperspectral imaging data, the core of which is to apply a nonlinear spectral unmixing model that can overcome the mineral spectral mixing effect, convert the spectral information of each pixel into mineral component percentages by accurately quantifying the key parameters such as the depth and area of the characteristic absorption band of specific altered minerals (such as chlorite, kaolinite), and generate a quantitative map that accurately reflects the mineral types and spatial abundance distribution of the tunnel face, providing a solid chemical composition data foundation for subsequent correlation with physical and mechanical characteristics. On this basis, step S320 builds a deep correlation between the two heterogeneous features, preferably using a graph attention network, considering the pixels in the mineral distribution map and the persistence intervals in the topological features as nodes in the graph, and through the attention mechanism of the network, it adaptively learns the complex nonlinear mapping relationship between different mineral combinations and the topological barcode persistence intervals representing the structure characteristics of the rock mass, rather than predefining simple rules; the key is that this process introduces the tunnel axis direction as a spatial constraint, allowing the model to consider the anisotropic characteristics inherent in the response of surrounding rock in tunnel engineering when learning the correlation, and finally output a feature tensor that contains complex cross-modal relationships. Step S330 refines and visualizes this preliminary correlation, and through tensor decomposition methods, it reduces the dimensionality and denoises the aforementioned feature tensor, stripping out the core components of the cross-modal correlation; in this process, the persistence of the topological features is used as a weight, giving stronger correlation importance to those topological structures that exist stably at multiple scales, thereby optimizing the correlation strength, and finally generating a fused topological-spectral correlation map that clearly shows the internal relationship between the physical and chemical properties of the rock mass, which provides a high-quality data foundation for subsequent manifold learning. The entire process embodies the layer-by-layer progressive relationship from data quantification to nonlinear relationship modeling and then to information condensation and visualization, aiming to deeply explore the coupling mechanism between mineral composition and rock mass structure.

[0059] Further, step S400 includes steps S410 to S430.

[0060] Step S410, according to the fused topological-spectral correlation map, a graph structure optimization process is performed, by introducing the stress difference between the tunnel axial and radial directions as anisotropy weights into the edge weight calculation of the correlation map, an anisotropic correlation map is obtained after optimization;

[0061] Step S420, according to the anisotropic correlation map, a diffusion process simulation process is performed, by simulating the diffusion process of physical quantities on the graph structure considering the stress field constraint, a diffusion affinity kernel matrix reflecting the correlation of the internal structure of the rock mass is constructed;

[0062] Step S430, according to the diffusion affinity nuclear matrix, a manifold coordinate extraction process is performed, and a low-dimensional eigenvector representing the essential structure of the surrounding rock stability is obtained by performing eigenvalue decomposition on the matrix and selecting the eigenvector corresponding to the maximum eigenvalue.

[0063] Specifically, step S410 optimizes the initial fusion graph in the engineering context, and introduces the key physical factor of the difference between the axial and radial stress caused by excavation unloading in the tunnel engineering as an anisotropic weight into the weight calculation of the connection edge between the nodes of the correlation graph; this makes the optimized graph structure no longer an abstract mathematical object reflecting only data similarity, but an "anisotropic correlation graph" deeply integrating the real stress state constraint of the rock mass, whose edge weight shows difference along the axial and radial directions of the tunnel, more accurately simulating the actual influence of the in-situ stress field on the correlation of the internal structure of the rock mass. Based on this optimized graph structure, step S420 detects the potential connectivity or affinity between nodes by simulating the diffusion process of a virtual "physical quantity" (such as heat or matter) on the anisotropic graph; in this diffusion simulation, the edge weight under the constraint of the in-situ stress determines the difficulty of "diffusion", and the diffusion affinity nuclear matrix finally constructed reflects the essential correlation strength between any two nodes (representing different local features of the rock mass) in the graph under the comprehensive consideration of the correlation of physical structure and chemical composition and the constraint of the in-situ stress field. Step S430 aims to extract a low-dimensional representation that best represents the main structural features from this affinity nuclear matrix containing complex relationships; by performing eigenvalue decomposition on the matrix and selecting the eigenvector corresponding to the maximum eigenvalue, the high-dimensional and complex correlation data is actually projected into a low-dimensional subspace, and the direction of the subspace retains the most important variation information in the data, so that the low-dimensional eigenvector can capture the essential structural pattern hidden behind the original multi-modal data that determines the stability state of the surrounding rock, providing a simple and representative feature input for subsequent dynamic classification. The entire processing chain embodies the progressive idea of embedding engineering prior knowledge (in-situ stress) into the graph structure, mining deep relationships through physical process simulation (diffusion), and extracting essential information using mathematical tools (eigenvalue decomposition).

[0064] Further, step S500 includes step S510 to step S530.

[0065] Step S510, according to the low-dimensional eigenvector, a spatiotemporal feature sequence construction process is performed, and the continuously collected low-dimensional eigenvectors are organized in the tunneling order to form a spatiotemporal feature sequence by combining the tunnel face mileage information in the tunneling direction;

[0066] Step S520, according to the spatiotemporal feature sequence, a density clustering process is performed, and a time-series density clustering tree reflecting the gradual change process of the surrounding rock state is generated by applying a density clustering algorithm considering time-series accessibility in a sliding window.

[0067] Step S530, according to the time sequence density clustering tree, a surrounding rock state deduction and label generation process is performed, the merging and splitting mode of the clustering tree branch is analyzed, and the evolution path of the cluster across the sliding window is detected, so as to obtain a surrounding rock dynamic classification label representing the evolution trend of the surrounding rock stability.

[0068] Specifically, the time-space feature sequence construction of step S510 is to combine the low-dimensional feature vector reflecting the essential characteristics of the surrounding rock with the tunneling direction and the tunnel face mileage information, and to organize the feature vectors according to the time sequence and spatial position of the tunneling; this process converts a series of isolated feature points into a continuous feature sequence in the time-space dimension, so that each feature vector has clear time sequence context and spatial positioning information, laying a data foundation for analyzing the dynamic evolution of the surrounding rock state with the engineering advancement. Based on this sequence depicting the dynamic process, step S520 performs density clustering under time sequence constraint, in a sliding time-space window, an improved density clustering algorithm is applied, which not only considers the density distribution of the feature points in the feature space, but also introduces the concept of "time sequence accessibility", that is, to preferentially classify the points that are adjacent in time-space and similar in features into a class; by continuously sliding the window and performing this clustering process, a time sequence density clustering tree can be constructed, and this tree structure records the complete genealogical relationship of the generation, continuation, merging, splitting or disappearance of different clusters (representing different surrounding rock state categories) with the tunneling process, thereby directly showing the gradual or sudden change process of the surrounding rock state. Step S530 analyzes the clustering tree to generate dynamic labels, by deeply analyzing the merging and splitting mode of the branches in the clustering tree, the key nodes of the stability transition of the surrounding rock state can be identified; at the same time, by detecting the evolution path of the clusters that can exist continuously across the continuous time window, the dominant evolution trend of the surrounding rock state can be inferred, and finally the surrounding rock dynamic classification label assigned to each time sequence node not only contains the category information of the current state, but also contains the evolution direction and stability trend information relative to the previous state, realizing the improvement from static snapshot recognition to dynamic process deduction. The whole process embodies the technical path of using time-space context information to capture and characterize the continuous evolution behavior of the surrounding rock state through time sequence clustering tree analysis.

[0069] Further, step S600 includes step S610 to step S630.

[0070] Step S610, according to the surrounding rock dynamic classification label, a confidence evaluation process is performed, by analyzing the relative position density of the corresponding low-dimensional feature vector in the dynamic cluster, the initial confidence of the surrounding rock classification of each time sequence node is calculated;

[0071] Step S620 performs confidence optimization processing according to the initial confidence and the surrounding rock dynamic classification label sequence, and obtains an optimized integrated confidence by introducing a geological coherence constraint to perform confidence propagation and smoothing on classification results that are suddenly changed in time sequence but adjacent in space.

[0072] Step S630 performs multi-hypothesis decision fusion processing according to the optimized integrated confidence, fuses optimal classification hypotheses in a current and previous multiple sliding windows through a weighted voting mechanism, and determines a class corresponding to the highest integrated confidence as the final tunnel surrounding rock recognition result.

[0073] Specifically, step S610 quantifies the certainty degree of the classification by analyzing the relative position density of the low-dimensional feature vector representing the rock mass state within the dynamic class cluster in which it is divided; if the feature vector is located in the dense core area of the class cluster, it indicates that it is highly consistent with the typical features of the class, and the initial confidence is high; if it is located in the sparse area at the edge of the class cluster, it indicates that the attribution is uncertain. This step provides a basic credibility quantification index for each recognition result. On this basis, step S620 optimizes the initial confidence using the spatio-temporal context information, and reviews the recognition result sequence by introducing the basic constraint principle of “geological coherence”. In particular, it processes classification results that are suddenly changed in time sequence but adjacent in space (tunnel face mileage). By constructing a spatial neighborhood relationship, the confidence values of high-confidence nodes with adjacent geographical positions and consistent classes are propagated and smoothed to low-confidence sudden nodes caused by local disturbance, effectively suppressing the jumping of recognition results caused by non-geological factors, thereby obtaining an integrated confidence that is more consistent in the time and space dimensions. Step S630 then performs the final integrated decision based on this, and its mechanism is no longer simply dependent on the classification result at a single time point, but rather it fuses the classification hypotheses (i.e., optimal candidate classes) with the highest confidence in the current and previous multiple sliding windows through a weighted voting method. Finally, the class with the highest integrated confidence is determined as the tunnel surrounding rock recognition result. This multi-window, multi-hypothesis decision fusion mechanism simulates the idea of observing and determining trends over a period of time, significantly improving the anti-interference ability and long-term stability of the final output result. The entire process embodies a progressive decision-making strategy from single-point uncertainty evaluation, to spatio-temporal context optimization, to multi-time window information fusion, aiming to output a more reliable and practical recognition conclusion.

[0074] Further, step S620 includes steps S621 to S623.

[0075] Step S621, according to the surrounding rock dynamic classification label sequence, time sequence mutation point detection processing is carried out, the jensen-shannon divergence of the category label distribution in the adjacent time sequence window is calculated, the non-geological time sequence mutation point caused by the temporary local disturbance of the tunneling face is recognized, and a potential unreliable classification node set is obtained;

[0076] Step S623, according to the potential unreliable classification node set and the initial confidence, spatial context confidence propagation processing is carried out, a spatial neighbor graph is constructed with the tunnel face mileage coordinate as the reference, the confidence value of the high confidence node which is adjacent in geographical position and consistent in category label is propagated to the low confidence node, and a confidence distribution modified by spatial constraint is obtained;

[0077] Step S623, according to the confidence distribution modified by spatial constraint, smoothing processing based on geological continuity prior is carried out, a Gaussian process regression model is introduced, the mileage is taken as the input variable, and the modified confidence is taken as the observation value for fitting, the geological continuity along the tunneling direction is strengthened, and an optimized integrated confidence is obtained.

[0078] Specifically, step S621, which detects outliers in the classification result sequence, is essentially about calculating the Jason-Shannon divergence of the dynamic classification label distribution of the surrounding rock within adjacent time windows. This divergence can effectively measure the degree of difference between two probability distributions. Through this calculation, nodes whose category distribution changes drastically in a short period of time can be identified. These abrupt changes are often not caused by actual changes in geological conditions, but rather by measurement or classification biases caused by brief local disturbances at the tunnel face (such as surface loosening or local water seepage). This allows for the identification of a set of potentially unreliable classification nodes that require confidence correction. Based on this identification result, step S622 focuses on local correction using spatial context. The method is to construct a spatial nearest neighbor map based on the mileage coordinates of the tunnel face. On this basis, the confidence values ​​of high-confidence nodes that are spatially adjacent and classified into the same surrounding rock category are propagated and assigned to low-confidence nodes that are identified as potentially unreliable. This process is based on the reasonable assumption that "spatial adjacent rock mass areas of the same category should have similar reliability". It effectively utilizes the spatial continuity of the tunnel surrounding rock and makes a preliminary correction to the low-confidence results caused by local non-geological factors, thus obtaining a confidence distribution corrected by spatial constraints. Step S623 further integrates fundamental geological principles at a macroscopic scale by introducing a Gaussian process regression model for global smoothing based on prior geological coherence. This model uses the mileage along the tunnel excavation direction as the input variable and the spatially corrected confidence score as the observed value for fitting. The inherent smoothing properties of the Gaussian process reinforce the prior knowledge that geological bodies along the tunnel axis typically possess continuity and gradual change, ultimately outputting an optimized integrated confidence score that is smoother in spatial distribution and more consistent with the understanding of geological laws. These three steps embody a progressive optimization strategy, from identifying unreliable points to utilizing spatial relationships for local correction, and finally to global smoothing based on geological principles.

[0079] Example 2:

[0080] like Figure 2 As shown, this embodiment provides a tunnel surrounding rock identification system, the system including:

[0081] The acquisition module 901 is used to acquire mixed frequency acoustic wave reflection signals collected by a sensor array arranged on the tunnel face, and hyperspectral image data of the tunnel face rock mass in the visible to shortwave infrared bands, which are simultaneously acquired by a portable spectrometer.

[0082] Module 902 is used to construct a topology pattern based on the mixed-frequency acoustic wave reflection signal, and to construct a persistent topology barcode of acoustic wave response by analyzing the multi-frequency response, thereby obtaining topological invariant features characterizing the fracture and pore structure of the rock mass.

[0083] The fusion module 903 is configured to perform cross-modal fusion according to the topological invariant feature and the hyperspectral image data, to obtain a fused topological-spectral association graph by associating the persistence interval of the topological barcode with the mineral feature absorption band.

[0084] The learning module 904 is configured to perform implicit manifold structure learning according to the fused topological-spectral association graph, to obtain a low-dimensional feature vector reflecting a potential instability mode of the rock mass by calculating the geodesic distance of the association graph node on the nonlinear manifold.

[0085] The division module 905 is configured to perform dynamic surrounding rock classification according to the low-dimensional feature vector, to obtain a surrounding rock dynamic classification label by density clustering and introducing the concept of a sliding window in time series analysis to deduce the time sequence evolution characteristics of the surrounding rock state in the tunneling process.

[0086] The output module 906 is configured to perform identification result decision-making according to the surrounding rock dynamic classification label, to obtain a final tunnel surrounding rock identification result by confidence weighted fusion of multiple time sequence classification results.

[0087] In an embodiment of the present application, the construction module 902 includes:

[0088] The first construction unit is configured to perform frequency domain feature separation processing according to the mixed frequency acoustic wave reflection signal, to obtain a multi-scale acoustic wave response mode by adaptively decomposing the signal into isolated frequency band components corresponding to macroscopic crack scattering and microscopic pore resonance, respectively, through a matching pursuit algorithm.

[0089] The second construction unit is configured to perform preliminary calculation processing of the topological invariant according to the multi-scale acoustic wave response mode, to obtain a multi-scale topological invariant sequence by constructing an alpha complex on the spatial distribution of the response amplitude of each mode and calculating the change of each order Betti number with the distance parameter.

[0090] The third construction unit is configured to perform integration and coding processing of the topological feature according to the multi-scale topological invariant sequence, to obtain a topological invariant feature by aligning and fusing the Betti number barcodes at different scales according to their persistence in the direction of the tunnel stress field to form an acoustic wave response persistence topological barcode.

[0091] In an embodiment of the present application, the fusion module 903 includes:

[0092] The first fusion unit is configured to perform mineral composition quantitative inversion processing according to the hyperspectral image data, to obtain a mineral composition quantitative distribution map of the tunnel face rock mass by analyzing the feature absorption band depth and area of a preset type of altered mineral based on a nonlinear spectral unmixing model.

[0093] The second fusion unit is used to construct cross-modal correlations based on the quantitative distribution map of mineral composition and topological invariant features. It models the nonlinear mapping relationship between different mineral combinations and the persistence interval of the topological barcode through graph attention network, and introduces the tunnel axis direction as a spatial constraint to obtain the preliminary topological-spectral correlation feature tensor.

[0094] The third fusion unit is used to optimize the association relationship and generate the spectrum based on the preliminary topological-spectral association feature tensor. It fuses multimodal information through tensor decomposition and uses topological persistence as a weight to optimize the association strength, generating a fused topological-spectral association graph.

[0095] Example 3:

[0096] Corresponding to the above method embodiments, this embodiment also provides a tunnel surrounding rock identification device. The tunnel surrounding rock identification device described below and the tunnel surrounding rock identification method described above can be referred to in correspondence.

[0097] Figure 3 This is a block diagram illustrating a tunnel surrounding rock identification device 800 according to an exemplary embodiment. Figure 3 As shown, the tunnel surrounding rock identification device 800 may include: a processor 801 and a memory 802. The tunnel surrounding rock identification device 800 may also include one or more of a multimedia component 803, an I / O interface 804, and a communication component 805.

[0098] The processor 801 is configured to control overall operations of the tunnel surrounding rock identification device 800 to complete all or part of the steps of the tunnel surrounding rock identification method. The memory 802 is configured to store various types of data to support the operation of the tunnel surrounding rock identification device 800. For example, the data can include instructions for any application or method operating on the tunnel surrounding rock identification device 800, and application-related data, such as contact data, sent and received messages, pictures, audio, video, and the like. The memory 802 can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic memory, flash memory, magnetic disk or optical disk. The multimedia component 803 can include a screen and an audio component. The screen can be a touch screen, for example, and the audio component is configured to output and / or input audio signals. For example, the audio component can include a microphone configured to receive external audio signals. The received audio signals can be further stored in the memory 802 or transmitted through the communication component 805. The audio component also includes at least one speaker configured to output audio signals. The I / O interface 804 provides an interface between the processor 801 and other interface modules, which can be a keyboard, a mouse, a button, and the like. The buttons can be virtual buttons or physical buttons. The communication component 805 is configured to perform wired or wireless communication between the tunnel surrounding rock identification device 800 and other devices. Wireless communication, such as Wi-Fi, Bluetooth, near field communication (NFC), 2G, 3G or 4G, or a combination of one or more of them, so the corresponding communication component 805 can include a Wi-Fi module, a Bluetooth module, and an NFC module.

[0099] In an exemplary embodiment, a tunnel surrounding rock identification device 800 can be implemented by one or more Application Specific Integrated Circuit (ASIC), Digital Signal Processor (DSP), Digital Signal Processing Device (DSPD), Programmable Logic Device (PLD), Field Programmable Gate Array (FPGA), controller, microcontroller, microprocessor or other electronic elements for executing the tunnel surrounding rock identification method described above.

[0100] In another exemplary embodiment, a computer readable storage medium including program instructions is also provided, which, when executed by a processor, implements the steps of the tunnel surrounding rock identification method described above. For example, the computer readable storage medium can be the memory 802 described above including program instructions executable by the processor 801 of the tunnel surrounding rock identification device 800 to complete the tunnel surrounding rock identification method described above.

[0101] The above description is merely a specific implementation of the present application, but the scope of protection of the present application is not limited thereto. Any person skilled in the art can easily think of changes or replacements within the technical scope disclosed by the present application, which should be covered within the scope of protection of the present application.

Claims

1. A method for identifying surrounding rock in tunnels, characterized in that, include: Acquire mixed-frequency acoustic wave reflection signals collected by a sensor array deployed at the tunnel face, and hyperspectral image data of the tunnel face rock mass in the visible to short-wave infrared bands simultaneously acquired by a portable spectrometer; Based on the mixed-frequency acoustic wave reflection signal, a topological pattern is constructed, and a persistent topological barcode of acoustic wave response is constructed by analyzing the multi-frequency response to obtain the topological invariant features characterizing the rock mass fracture and pore structure. Cross-modal fusion is performed based on the topological invariant features and the hyperspectral image data. By associating the persistence range of the topological barcode with the absorption bands of mineral features, a fused topological-spectral correlation map is obtained. Based on the fused topological-spectral correlation graph, implicit manifold structure learning is performed. By calculating the geodesic distance of the correlation graph nodes on the nonlinear manifold, a low-dimensional feature vector reflecting the potential instability mode of the rock mass is obtained. Dynamic surrounding rock classification is performed based on the low-dimensional feature vectors. Density clustering and the concept of sliding window in time series analysis are introduced to deduce the temporal evolution characteristics of the surrounding rock state during tunnel excavation, and dynamic classification labels of the surrounding rock are obtained. The identification result decision is made based on the dynamic classification label of the surrounding rock, and the final tunnel surrounding rock identification result is obtained by fusing multiple time-series classification results through confidence weighting.

2. The tunnel surrounding rock identification method according to claim 1, characterized in that, Topology pattern construction based on the mixed-frequency acoustic wave reflection signal includes: Based on the mixed-frequency acoustic wave reflection signal, frequency domain feature separation processing is performed, and the signal is adaptively decomposed into isolated frequency band components corresponding to macroscopic crack scattering and microscopic pore resonance respectively through the matching pursuit algorithm to obtain multi-scale acoustic wave response modes. The topological invariants are initially calculated based on the multi-scale acoustic response modes. An alpha complex is constructed by analyzing the spatial distribution of the response amplitudes of each mode, and the variation of each Betti number with the distance parameter is calculated to obtain a multi-scale topological invariant sequence. Based on the multi-scale topological invariant sequence, the topological features are integrated and encoded. By aligning and fusing the Betty number barcodes at different scales according to their persistence in the tunnel stress field direction, a persistent acoustic response topological barcode is formed, thus obtaining the topological invariant features.

3. The tunnel surrounding rock identification method according to claim 1, characterized in that, Cross-modal fusion based on the topological invariant features and the hyperspectral image data includes: Based on the hyperspectral image data, a quantitative inversion of mineral composition is performed. The depth and area of ​​the characteristic absorption bands of preset types of alteration minerals are analyzed by a nonlinear spectral unmixing model to obtain a quantitative distribution map of mineral composition of the face rock mass. Based on the quantitative distribution map of mineral composition and the topological invariant features, cross-modal correlation relationships are constructed. The nonlinear mapping relationship between different mineral combinations and the persistence interval of the topological barcode is modeled through graph attention network, and the tunnel axis direction is introduced as a spatial constraint to obtain the preliminary topological-spectral correlation feature tensor. Based on the preliminary topological-spectral correlation feature tensor, correlation relationship optimization and spectrum generation are performed. Multimodal information is fused through tensor decomposition, and topological persistence is used as a weight to optimize the correlation strength, generating a fused topological-spectral correlation graph.

4. The tunnel surrounding rock identification method according to claim 1, characterized in that, Learning the implicit manifold structure based on the fused topology-spectral correlation graph includes: The graph structure is optimized based on the fused topology-spectral correlation graph. The difference between the tunnel axial and radial ground stresses is introduced as anisotropic weights into the edge weight calculation of the correlation graph to obtain the optimized anisotropic correlation graph. The diffusion process is simulated based on the anisotropic correlation diagram. By simulating the diffusion process of physical quantities on the diagram structure considering stress field constraints, a diffusion affinity matrix reflecting the correlation of the internal structure of the rock mass is constructed. Based on the diffusion affinity nucleus matrix, manifold coordinates are extracted. By performing eigenvalue decomposition on the matrix and selecting the eigenvector corresponding to the largest eigenvalue, a low-dimensional eigenvector characterizing the essential structure of the surrounding rock stability is obtained.

5. The tunnel surrounding rock identification method according to claim 1, characterized in that, Dynamic surrounding rock classification is performed based on the low-dimensional feature vectors, including: Based on the low-dimensional feature vectors, a spatiotemporal feature sequence is constructed. By combining the tunnel face mileage information in the tunnel excavation direction, the continuously collected low-dimensional feature vectors are organized according to the excavation sequence to form a spatiotemporal feature sequence. Density clustering is performed based on the spatiotemporal feature sequence. By applying a density clustering algorithm that considers temporal reachability within a sliding window, a temporal density clustering tree that reflects the gradual change process of the surrounding rock state is generated. Based on the temporal density clustering tree, the surrounding rock state is inferred and a label is generated. By analyzing the merging and splitting patterns of the clustering tree branches and detecting the evolution path of clusters crossing the sliding window, dynamic classification labels of the surrounding rock that characterize the evolution trend of surrounding rock stability are obtained.

6. The tunnel surrounding rock identification method according to claim 1, characterized in that, The identification result decision is made based on the dynamic classification label of the surrounding rock, including: The confidence level is evaluated based on the dynamic classification label of the surrounding rock. The initial confidence level of the surrounding rock classification at each time node is calculated by analyzing the relative position density of the corresponding low-dimensional feature vector in its dynamic cluster. The confidence level is optimized based on the initial confidence level and the dynamic classification label sequence of the surrounding rock. By introducing geological coherence constraints, the confidence level is propagated and smoothed for classification results that are abrupt in time but adjacent in space, and the optimized integrated confidence level is obtained. Based on the optimized integrated confidence level, a multi-hypothesis decision fusion process is performed. The optimal classification hypothesis of the current and previous sliding windows is fused through a weighted voting mechanism, and the category corresponding to the highest integrated confidence level is taken as the final tunnel surrounding rock identification result.

7. The tunnel surrounding rock identification method according to claim 6, characterized in that, Based on the initial confidence level and the dynamic classification label sequence of the surrounding rock, confidence level optimization processing is performed, including: Based on the dynamic classification label sequence of the surrounding rock, a temporal abrupt change point detection process is performed. By calculating the Jason-Shannon divergence of the category label distribution within adjacent temporal windows, non-geological temporal abrupt change points caused by brief local disturbances at the tunnel face are identified, and a set of potentially unreliable classification nodes is obtained. Based on the set of potentially unreliable classification nodes and the initial confidence, spatial context confidence propagation processing is performed. By constructing a spatial nearest neighbor graph based on the tunnel face mileage coordinates, the confidence values ​​of high-confidence nodes that are geographically adjacent and have the same category label are propagated to low-confidence nodes, resulting in a confidence distribution corrected by spatial constraints. Based on the spatially constrained confidence distribution, a smoothing process based on geological continuity priors is performed. By introducing a Gaussian process regression model, mileage is used as the input variable and the corrected confidence is used as the observation value for fitting, thereby strengthening the geological continuity along the tunneling direction and obtaining the optimized integrated confidence.

8. A tunnel surrounding rock identification system, characterized in that, include: The acquisition module is used to acquire mixed-frequency acoustic wave reflection signals collected by a sensor array arranged on the tunnel face, as well as hyperspectral image data of the tunnel face rock mass in the visible to short-wave infrared bands, which are simultaneously acquired by a portable spectrometer. The construction module is used to construct a topology pattern based on the mixed frequency acoustic wave reflection signal, and to construct a persistent topology barcode of acoustic wave response by analyzing the multi-frequency response, thereby obtaining topological invariant features characterizing the rock mass fracture and pore structure. The fusion module is used to perform cross-modal fusion based on the topological invariant features and the hyperspectral image data. By associating the persistence interval of the topological barcode with the absorption band of the mineral feature, a fused topological-spectral correlation map is obtained. The learning module is used to learn the implicit manifold structure based on the fused topology-spectral correlation graph. By calculating the geodesic distance of the nodes in the correlation graph on the nonlinear manifold, a low-dimensional feature vector reflecting the potential instability mode of the rock mass is obtained. The segmentation module is used to dynamically classify the surrounding rock categories based on the low-dimensional feature vectors. It uses density clustering and introduces the concept of a sliding window from time series analysis to deduce the temporal evolution characteristics of the surrounding rock state during tunnel excavation and obtain dynamic classification labels for the surrounding rock. The output module is used to make identification decisions based on the dynamic classification labels of the surrounding rock, and obtain the final tunnel surrounding rock identification result by fusing multiple time-series classification results through confidence weighting.

9. The tunnel surrounding rock identification system according to claim 8, characterized in that, The building module includes: The first building unit is used to perform frequency domain feature separation processing on the mixed frequency acoustic wave reflection signal, and adaptively decompose the signal into isolated frequency band components corresponding to macroscopic crack scattering and microscopic pore resonance respectively through the matching pursuit algorithm to obtain multi-scale acoustic wave response modes. The second construction unit is used to perform preliminary calculations of topological invariants based on the multi-scale acoustic response modes. By constructing an alpha complex based on the spatial distribution of the response amplitudes of each mode, and calculating the changes of each order Betti number with the distance parameter, a multi-scale topological invariant sequence is obtained. The third construction unit is used to integrate and encode topological features based on the multi-scale topological invariant sequence. By aligning and fusing Betty number barcodes at different scales according to their persistence in the tunnel stress field direction, a persistent acoustic response topological barcode is formed, thereby obtaining topological invariant features.

10. The tunnel surrounding rock identification system according to claim 8, characterized in that, The fusion module includes: The first fusion unit is used to perform quantitative inversion processing of mineral composition based on the hyperspectral image data. By analyzing the depth and area of ​​the characteristic absorption bands of preset types of alteration minerals based on a nonlinear spectral unmixing model, a quantitative distribution map of mineral composition of the face rock mass is obtained. The second fusion unit is used to construct cross-modal correlations based on the quantitative distribution map of mineral components and the topological invariant features. It models the nonlinear mapping relationship between different mineral combinations and the persistence interval of the topological barcode through a graph attention network, and introduces the tunnel axis direction as a spatial constraint to obtain a preliminary topological-spectral correlation feature tensor. The third fusion unit is used to optimize the association relationship and generate the spectrum based on the preliminary topological-spectral association feature tensor. It fuses multimodal information through tensor decomposition and uses topological persistence as a weight to optimize the association strength, thereby generating a fused topological-spectral association graph.

Images