A concrete spraying rebound rate optimization decision method and system

Abstract

The present application provides a concrete spraying rebound rate optimization decision method and system. Specifically, the present application first collects and extracts the time-frequency features from the spray impact acoustic emission signal, and simultaneously collects laser Doppler array measurement data during the concrete spraying process to obtain the concrete flow velocity distribution. The time-frequency features and the concrete flow velocity distribution are then subjected to multi-source fusion processing, and the principal component eigenvectors associated with the rebound are extracted. A spatial velocity field model is then constructed to obtain identification results characterizing the concrete rebound state. Cluster analysis is then performed on the principal component eigenvectors and the identification results to identify high-rebound mode categories and low-rebound mode categories to generate a risk zoning map. Finally, an adaptive control algorithm is applied to dynamically generate spray parameter adjustment instructions and output a concrete rebound rate optimization decision plan. The technical solution provided by the present application not only overcomes the feature drift problem caused by the spatiotemporal asynchrony of physical field data, but also significantly improves the overall rebound rate prediction accuracy.

Description

Technical Field

[0001] The present application relates to the field of intelligent civil engineering technology, and in particular to a concrete shotcrete rebound rate optimization decision-making method and system. Background Art

[0002] During concrete construction, concrete shotcrete rebound causes significant material waste and dust pollution, while also impacting the quality and stability of the support structure. To effectively control rebound, it is necessary to obtain real-time information on the kinetic energy distribution characteristics of the concrete shotcrete flow and the energy dissipation of aggregate impact. By precisely locating the spatial coordinates of rebound risk and dynamically outputting zone control instructions based on the movement trajectory of the shotcrete equipment, the coordinated optimization of shotcrete pressure and material delivery ratio can be achieved.

[0003] A representative technical solution currently uses a three-dimensional laser velocimeter to monitor concrete flow velocity distribution data and, based on a time-series prediction model, to determine overall rebound trends. When the predicted result exceeds a preset threshold, the system triggers a global pressure adjustment mechanism and simultaneously applies local velocity threshold control based on a fixed-scale spatial grid, attempting to balance rebound suppression with energy consumption control.

[0004] However, this technical solution still has three core limitations. First, due to the failure to integrate the acoustic emission signal characteristics, the energy of aggregate and interface peeling cannot be effectively captured, and there is a serious omission in the identification of adhesion failure type rebound; second, the static spatial grid division destroys the topological continuity of the velocity field, making it difficult to capture the sudden change characteristics of the interlayer flow state, resulting in insufficient accuracy in the spatial positioning of the rebound risk; third, the fixed control rules lack the ability to dynamically match the motion state of the equipment, which not only leads to excessive adjustment in non-risk areas and energy waste, but also makes the control effect in high-risk areas fall short of expectations. Summary of the Invention

[0005] The present application provides a concrete shotcrete rebound rate optimization decision-making method and system to solve the problems of single physical field monitoring, distortion in flow state mutation identification, and mismatch between control instructions and spatial risks in the existing technology.

[0006] In a first aspect, the present application provides a concrete shotcrete rebound rate optimization decision method, comprising:

[0007] Collecting the injection impact acoustic emission signal during the concrete spraying process and extracting the time-frequency characteristics of the injection impact acoustic emission signal;

[0008] Collect laser Doppler array measurement data during concrete spraying to obtain concrete flow velocity distribution;

[0009] Performing multi-source fusion processing on the time-frequency characteristics and the concrete flow velocity distribution, and extracting a principal component feature vector of rebound correlation from the multi-source fusion result;

[0010] Constructing a spatial velocity field model based on the concrete flow velocity distribution to obtain an identification result representing the concrete rebound state;

[0011] performing cluster analysis on the principal component eigenvectors and the identification results to identify high-resilience pattern categories and low-resilience pattern categories to generate a risk zoning map;

[0012] Based on the risk zoning map, an adaptive control algorithm is applied to dynamically generate spraying parameter adjustment instructions and output a concrete rebound rate optimization decision plan.

[0013] Optionally, collecting a jet impact acoustic emission signal during the concrete spraying process and extracting a time-frequency feature of the jet impact acoustic emission signal includes:

[0014] The original acoustic wave signal generated by the impact of concrete spraying is captured by a piezoelectric sensor array;

[0015] Converting the original sound wave signal into a first voltage change signal via a charge amplifier;

[0016] performing signal envelope separation processing on the first voltage change signal to obtain a second voltage change signal reflecting a change in signal amplitude;

[0017] Dynamically dividing the time window interval based on the peak position of the second voltage change signal, and decomposing the original sound wave signal into a preset number of frequency intervals within a single time window interval;

[0018] The amplitude distribution intensity of the original sound wave signal in each frequency interval is counted, and the amplitude distribution intensity of the frequency intervals of all time window intervals is combined to generate time-frequency features.

[0019] Optionally, laser Doppler array measurement data is collected during the concrete spraying process to obtain the concrete flow velocity distribution, including:

[0020] At least three laser detection units with spatial point distribution characteristics simultaneously receive light signals reflected by concrete particles during the spraying process, and capture signal change characteristics of the light signals generated by the movement of the concrete particles;

[0021] Correlating three-dimensional movement direction information of concrete particles according to the spatial point distribution characteristics and the signal change characteristics;

[0022] Aggregating the three-dimensional motion direction information according to the detection time series to generate instantaneous flow features with timestamps;

[0023] The instantaneous flow characteristics are mapped to the spatial position corresponding to the movement path of the spraying equipment to generate the concrete flow velocity distribution.

[0024] Optionally, performing multi-source fusion processing on the time-frequency features and the concrete flow velocity distribution, and extracting a principal component feature vector of rebound correlation from the multi-source fusion result, includes:

[0025] Receive the time-frequency characteristics and concrete flow rate distribution within the same spraying time period;

[0026] Synchronously dividing the time-frequency characteristics and the concrete flow velocity distribution into characteristic units of the same time window;

[0027] Cross-analyze the time-frequency feature unit and the flow velocity distribution unit in each time window to generate a fusion feature unit;

[0028] Integrate the fusion feature units of all time windows to generate a feature set;

[0029] A principal component feature vector associated with the concrete rebound behavior is extracted from the feature set.

[0030] Optionally, constructing a spatial velocity field model based on the concrete flow velocity distribution to obtain an identification result characterizing the concrete rebound state includes:

[0031] Determine the direction of operation based on the movement trajectory of the spraying equipment;

[0032] Taking the concrete flow velocity distribution as input, a spatial velocity field model including a spatial topological structure is constructed;

[0033] Dividing a hierarchical flow framework along the operation advancing direction in the spatial velocity field model;

[0034] Performing a flow state comparison operation on adjacent layers on the divided hierarchical flow framework, and identifying flow discontinuity feature points generated by the comparison operation;

[0035] Based on the spatial position and change direction of all flow discontinuity feature points, identification results representing the concrete rebound state are generated.

[0036] Optionally, cluster analysis is performed on the principal component eigenvector and the identification result to identify high-resilience mode categories and low-resilience mode categories to generate a risk zoning map, including:

[0037] Bind each component of the principal component eigenvector to the spatial coordinates of the identification result, create spatial state recording points, and calculate the rebound state correlation index between each spatial state recording point;

[0038] Dividing a dense area recording point set and a sparse area recording point set according to the rebound state correlation index;

[0039] The set of recording points in the dense area is marked as a high-rebound mode category, and the set of recording points in the sparse area is marked as a low-rebound mode category;

[0040] Determining a concrete spraying coverage area based on a spatial coordinate distribution range of the identification result;

[0041] The spatial coordinates of the high-rebound mode category and the low-rebound mode category are projected back to the concrete shotcrete coverage area to generate a risk zoning map covering the shotcrete surface.

[0042] Optionally, based on the risk zoning map, an adaptive control algorithm is applied to dynamically generate spraying parameter adjustment instructions and output a concrete rebound rate optimization decision plan, including:

[0043] On the risk zoning map, continuous dynamic control units are divided along the movement trajectory of the injection device, and a distribution density value of a high rebound mode category within each dynamic control unit is extracted;

[0044] Establish an association rule base between the injection outlet pressure of the injection equipment and the material delivery ratio according to the distribution density value;

[0045] generating an instruction fragment based on an entry of an association rule base of a dynamic control unit at the injection position of the current injection device;

[0046] Combining all instruction fragments along the movement trajectory of the injection device to generate a dynamic injection parameter adjustment instruction set;

[0047] The dynamic spraying parameter adjustment instruction set is bound to the spatial mapping relationship of the risk partition map to generate partition decision items with the control unit as the management granularity, and all partition decision items are integrated to output a concrete rebound rate optimization decision plan.

[0048] In a second aspect, the present application provides a concrete shotcrete rebound rate optimization decision system, comprising:

[0049] A first acquisition module is used to collect the injection impact acoustic emission signal during the concrete spraying process and extract the time-frequency characteristics of the injection impact acoustic emission signal;

[0050] The second acquisition module is used to collect laser Doppler array measurement data during the concrete spraying process to obtain the concrete flow velocity distribution;

[0051] a fusion module, configured to perform multi-source fusion processing on the time-frequency features and the concrete flow velocity distribution, and extract a principal component feature vector associated with rebound from the multi-source fusion result;

[0052] A construction module is used to construct a spatial velocity field model based on the concrete flow velocity distribution to obtain an identification result representing the concrete rebound state;

[0053] an analysis module, configured to perform cluster analysis on the principal component eigenvector and the identification result, identify high-resilience pattern categories and low-resilience pattern categories, and generate a risk zoning map;

[0054] The output module is used to dynamically generate injection parameter adjustment instructions based on the risk zoning map using an adaptive control algorithm and output a concrete rebound rate optimization decision plan.

[0055] In a third aspect, an embodiment of the present application provides a computing device comprising a processing component and a storage component; the storage component stores one or more computer instructions; the one or more computer instructions are used to be called and executed by the processing component to implement a concrete spraying rebound rate optimization decision method as described in the first aspect above.

[0056] In a fourth aspect, an embodiment of the present application provides a computer storage medium storing a computer program. When the computer program is executed by a computer, the method for optimizing the decision-making of the rebound rate of concrete spraying as described in the first aspect is implemented.

[0057] This embodiment achieves multi-physics collaborative perception by synchronously collecting acoustic emission signals and laser Doppler velocity data. It then fuses and processes the principal component features associated with rebound, overcoming the limitations of single sensor data. Ultimately, it generates a risk zoning map and dynamically adjusts injection parameters, forming a closed "perception-analysis-control" loop. Its core technological value lies in its first deep coupling of acoustic energy dissipation (time-frequency characteristics) with fluid dynamics (velocity distribution), accurately capturing two types of rebound factors: concrete adhesion failure (interfacial debonding) and kinetic energy attenuation (velocity mutation). This significantly improves the overall rebound rate prediction accuracy and provides comprehensive data support for dynamic optimization of the injection process.

[0058] Furthermore, a time window synchronization mechanism is used to align the time-frequency features and the concrete flow velocity distribution into feature units of the same time granularity. Cross-analysis is performed within each time window to generate fused feature units, and implicit rebound behavior patterns are mined through signal-flow field coupling. Finally, the feature set is integrated and the principal component eigenvector is extracted. The core breakthrough lies in overcoming the feature drift problem caused by the spatiotemporal asynchrony of physical field data. By generating fused feature units through dynamic time window matching and cross-analysis, the correlation between the acoustic emission energy peak and the flow velocity gradient is significantly improved, achieving a quantitative characterization of the peeling strength of the spray interface. Secondly, based on the spatial velocity field model of the flow velocity distribution, a hierarchical flow framework is constructed along the movement trajectory of the spraying equipment. The flow discontinuity feature points are located by comparing the flow states of adjacent layers, and a spatially bound rebound identification result is generated. The technical advantage lies in the fact that the propulsion direction of the spraying operation is used as the topological axis of the flow framework for the first time, avoiding the spatial distortion problem caused by static meshing and achieving centimeter-level positioning accuracy of the flow state mutation point. At the same time, the directional attribute (change direction) of the identification result can directly guide the real-time calibration of the spraying angle.

[0059] These and other aspects of the present application will become more readily apparent from the description of the following embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0060] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, a brief introduction will be given below to the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0061] Figure 1 A flowchart of a concrete spraying rebound rate optimization decision method provided by the present application is shown;

[0062] Figure 2 The following is a schematic diagram showing the structure of a concrete spraying rebound rate optimization decision system provided by the present application;

[0063] Figure 3 A schematic structural diagram of a computing device provided by the present application is shown. DETAILED DESCRIPTION

[0064] In order to enable those skilled in the art to better understand the solution of the present application, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application.

[0065] In some of the processes described in the specification and claims of this application and the above-mentioned figures, multiple operations that appear in a specific order are included, but it should be clearly understood that these operations may not be executed in the order in which they appear in this document or may be executed in parallel. The serial numbers of the operations, such as 101, 102, etc., are only used to distinguish between different operations, and the serial numbers themselves do not represent any order of execution. In addition, these processes may include more or fewer operations, and these operations may be executed in sequence or in parallel. It should be noted that the descriptions of "first", "second", etc. in this document are used to distinguish different messages, devices, modules, etc., and do not represent a sequential order, nor do they limit "first" and "second" to being different types.

[0066] The following will be combined with the drawings in the embodiments of this application to clearly and completely describe the technical solutions in the embodiments of this application. Obviously, the embodiments described are only part of the embodiments of this application, not all of the embodiments. Based on the embodiments in this application, all other embodiments obtained by those skilled in the art without making creative efforts are within the scope of protection of this application.

[0067] Figure 1 A flowchart of a concrete spraying rebound rate optimization decision method is provided for an embodiment of the present application. Figure 1 As shown, the method includes:

[0068] Step 101 : collecting a jet impact acoustic emission signal during a concrete spraying process, and extracting a time-frequency feature of the jet impact acoustic emission signal.

[0069] In this step, the jet impact acoustic emission signal refers to the transient stress wave generated by the collision and peeling of aggregate when the concrete spray impacts the support surface. Its physical carrier is the acoustic wave signal of mechanical vibration propagating in the medium; the time-frequency feature refers to the energy distribution matrix that integrates the time and frequency dimensions. After the acoustic wave signal is captured by the piezoelectric sensor array, it is generated through charge conversion, envelope separation, dynamic time window division and frequency domain decomposition.

[0070] In this embodiment, the original acoustic wave signal generated by the concrete spraying impact is first captured by a spatially arranged piezoelectric sensor array, and the array distribution ensures coverage of the irregular area of ​​the spraying surface; the original acoustic wave signal is then input into a charge amplifier and converted into a first voltage change signal, and transmission attenuation is eliminated through impedance matching; envelope separation processing is performed on the first voltage change signal, and the signal amplitude envelope is extracted using Hilbert transform to generate a second voltage change signal that focuses on the impact energy change; the time window boundary is dynamically determined based on the peak position of the second voltage change signal, so that each window completely contains an impact event, and the original acoustic wave signal is divided into a preset number of frequency sub-bands within the window through wavelet packet decomposition; finally, the root mean square value of the signal amplitude in each frequency sub-band is counted and integrated into a time-frequency distribution matrix according to the time window sequence, that is, the final time-frequency feature.

[0071] For example, during tunnel lining shotcrete construction, an array of nine piezoelectric sensors (three on the arch and six on the sidewalls) was deployed for the shotcrete operation of C30 concrete (water-cement ratio 0.45) in the arch area. When the concrete was sprayed at a flow rate of 8 m³ / h, the sensors captured an acoustic signal with a center frequency of 2.5 kHz and a duration of 0.1 s at the moment of impact. This signal was converted to a ±5 V voltage signal by a charge amplifier. The Hilbert transform was used to remove the carrier component above 20 kHz, generating an amplitude envelope signal. Three dynamic time windows of 0.08 s, 0.12 s, and 0.15 s were defined based on the envelope peaks. Within each window, a DB4 wavelet packet was used to decompose the signal into four frequency bands (0-1 kHz aggregate rolling, 1-2 kHz interface delamination, 2-3 kHz aggregate crushing, and 3-4 kHz cavity resonance). The root mean square value of each frequency band was calculated, generating a 12×4 time-frequency matrix. This matrix revealed a peak in the 1-2 kHz band (characteristic of adhesion failure) in the 0.12 s window, providing input for subsequent multi-source fusion.

[0072] Step 102 : Collect laser Doppler array measurement data during the concrete spraying process to obtain concrete flow velocity distribution.

[0073] In this step, the laser Doppler array refers to three or more groups of spatially distributed laser transmitting-receiving units, which capture motion information through the frequency offset of the light reflected by the particles.

[0074] In this embodiment, three groups of laser detection units distributed at spatial points are first used to synchronously emit laser beams to receive light signals reflected by concrete particles and capture the dynamic changes in the frequency and phase of the reflected light. The spatial geometric relationship between the detection units is combined with the signal frequency shift to calculate the particle motion direction vector (such as solving the three-dimensional velocity component through triangulation positioning). Multi-particle vector data is aggregated according to the detection time series to generate a set of instantaneous flow characteristics with millisecond-level timestamps. Finally, based on the spatial position changes of the real-time movement trajectory of the spraying equipment, the timestamp is mapped to the three-dimensional coordinates corresponding to the equipment movement path to generate the concrete flow velocity distribution covering the spraying surface.

[0075] For example, continuing with the tunnel lining case from step 101, laser detection units are deployed on both sides of the dome sprayer track. When concrete is sprayed, the particle reflected light generates frequency shifts of 12 kHz, 8 kHz, and 15 kHz in units A / B / C, respectively. Based on the spatial coordinates of the three units (spacing 0.5 m), the velocity vector of particle P in the coordinate system is calculated to be (1.2, 0.3, -2.1) m / s. 108 particle vector data are aggregated within this second-level window to generate an instantaneous flow feature with a timestamp of T153924. Combined with the sprayer's moving speed of 0.3 m / s along the track, T153924 is mapped to the dome coordinate area (x35, y18, z58), generating the following flow velocity distribution at this location: maximum velocity 2.8 m / s (edge ​​area) and minimum velocity 0.7 m / s (central vortex area).

[0076] Step 103 : performing multi-source fusion processing on the time-frequency features and the concrete flow velocity distribution, and extracting a principal component feature vector of rebound correlation from the multi-source fusion result.

[0077] In this step, the principal component eigenvectors are linearly independent orthogonal components extracted from the fused feature set, and their weights reflect the contribution of each feature to the concrete rebound behavior.

[0078] In this embodiment, the time-frequency feature matrix and concrete velocity distribution field with synchronized timestamps are first received, and the two are dynamically divided into time window slices of the same length based on the continuity of the time axis. Within each time window, a multi-dimensional cross-analysis is performed on the time-frequency feature unit (amplitude-frequency matrix) and the velocity distribution unit (spatial velocity vector field). A fused feature unit is generated by calculating the correlation coefficient between the energy peak of a specific frequency band of acoustic emission and the key nodes of the velocity field (such as the velocity gradient mutation point and the vorticity extreme point). The fused feature units of all time windows are integrated to construct a feature set. Finally, the principal component analysis method is used to reduce the dimensionality of the feature set, and the top K orthogonal components with the highest variance contribution rate are selected as the rebound correlation principal component eigenvectors. The direction of their eigenvalues ​​represents the degree of coupling between the material peeling strength and the flow instability.

[0079] For example, during the C30 concrete spraying process of the tunnel vault, based on the acoustic emission time-frequency characteristics obtained in step 101 (showing that the energy proportion of the 1-2kHz adhesion failure frequency band in the 0.08s time window is 62%) and the velocity distribution collected synchronously in step 102 (showing the velocity gradient of the eddy flow area at coordinates (x35, y18, z58) is 7.8s⁻¹), this step first aligns the two according to the 0.08s time window; performs cross analysis within this time window, and establishes a unified relationship between the proportion of 1-2kHz high-frequency acoustic energy and the velocity gradient. A correlation model was calculated and a strong correlation coefficient of 0.91 was obtained, generating a spatial coordinate set containing the acoustic energy-flow velocity coupling characteristics. After integrating the fused data of three consecutive time windows, the core eigenvectors PC1=0.82 (flow direction coupling) and PC2=0.71 (spatial aggregation) were extracted through principal component analysis. The abnormally high value of PC1 clearly indicates that there is a risk of aggregate-interface debonding in the coordinate area (x35.2, y18.1, z58). This eigenvector will drive the subsequent steps to accurately identify flow anomalies.

[0080] Step 104 : constructing a spatial velocity field model based on the concrete flow velocity distribution to obtain an identification result representing the concrete rebound state.

[0081] In this step, the identification result is a set of spatial positions and direction vectors of all feature points, which is used to describe the distribution of unstable flow areas on the jet surface.

[0082] In this embodiment, the operation advancement direction is first calculated based on the real-time moving trajectory coordinate sequence of the injection equipment, which serves as the topological reference axis of the spatial model; the concrete flow velocity distribution is used as input to construct a three-dimensional grid model with spatial adjacency (node ​​spacing adaptive flow gradient) to form a spatial velocity field model containing topological connections; the model is cut into hierarchical frames of equal distance or variable thickness along the advancement direction, and each frame covers the holographic flow velocity data of a specific flow section; the adjacent layer velocity vector comparison operation is performed within the frame, and the discontinuity is located by calculating the directional angle deviation and identifying the modulus difference threshold; finally, the spatial coordinates and directional vectors of all feature points are extracted to generate an identification result that identifies the flow instability area.

[0083] For example, following the tunnel vault case from step 103 (where the principal component feature PC1 = 0.82 indicates a high risk at the coordinates (x35.2, y18.1, z58)), this step first analyzes the jet trajectory to determine that the propulsion direction is a 30° positive deflection from the Y axis. A 2cm×2cm grid 3D topology model is constructed using the velocity distribution (maximum velocity 3.2 m / s) output from step 102. Seven hierarchical frames are formed at 10 cm intervals along the propulsion direction. A vector comparison is performed focusing on the intersection of frames 3 and 4 (covering coordinates (x35.2±0.1 m)): the average flow direction angle for frame 3 is 85° (modulus 2.8 m / s), and the average flow direction angle for frame 4 is 112° (modulus 1.1 m / s). The sudden change in direction angle of 27° exceeds the threshold of 20°. This point is identified as a flow discontinuity feature point, and the coordinates (x35.2, y18.1, z58.1) and the direction vector (-0.8, 0.3, -0.1) (Step 104-5). This point group constitutes the rebound state identification result.

[0084] Step 105 : performing cluster analysis on the principal component eigenvector and the identification result to identify high-resilience pattern categories and low-resilience pattern categories to generate a risk zoning map.

[0085] In this step, the high rebound mode category refers to the set of recording points in the spatial dense area divided by cluster analysis, which represents the area where the concrete rebound risk is significantly concentrated; the low rebound mode category refers to the set of recording points in the spatial sparse area divided by cluster analysis, which represents the safe area where the rebound risk is discretely distributed; the risk zoning map refers to the regional risk visualization map generated by projecting the spatial coordinates of the high rebound mode category and the low rebound mode category onto the concrete spraying coverage area.

[0086] In this embodiment, each component value of the principal component eigenvector (such as flow direction coupling and spatial aggregation) is first bound one by one to the spatial coordinates of the flow discontinuity feature points in the identification results to generate a set of spatial state record points that integrate material properties and position information; the multidimensional similarity between each record point is calculated (including principal component component differences and spatial distances) to generate a quantitative rebound state association index; according to the distribution characteristics of the index value, the record points are separated into a set of dense areas with high spatial aggregation and a set of sparse areas with low aggregation through density peak detection; the dense area set is marked as a high rebound mode category, and the sparse area is marked as a low rebound mode category; the minimum bounding box is calculated based on the spatial coordinates of all identification results to define the physical boundary of the concrete spraying coverage area; finally, the coordinate points of the two mode categories are projected into the coverage area to generate a spatial distribution map showing the risk level in color.

[0087] For example, following the tunnel vault case in step 104 (coordinates of the flow discontinuity feature point (x35.2, y18.1, z58.1) and direction vector), this step first binds the principal component feature component (PC1=0.82) to the coordinates to generate a spatial state record point A; calculates the similarity between the record point A and the adjacent point B (coordinates (x34.8, y17.9, z58.3)): principal component difference 0.05 + spatial distance 0.3m → correlation index value 0.87 (step 105-2); The density peak area centered on point A (containing 6 highly correlated points within a radius of 0.5 m) was detected from all recorded points and separated into a set of dense areas. This area was marked as a high-rebound mode category. Based on the coordinates of all 42 feature points in step 104, the coverage domain boundary [Xmin=32.1, Ymax=19.3, Zmin=56.5] was defined. The coordinates of the dense area were projected onto the coverage domain to generate a red high-risk area, and the sparse area was designated as a green safe area (step 105-6), forming a strip-shaped risk zone in the middle of the vault.

[0088] Step 106 : Based on the risk zoning map, an adaptive control algorithm is applied to dynamically generate spraying parameter adjustment instructions, and a concrete rebound rate optimization decision plan is output.

[0089] Step 106: Dynamically generate spraying parameter adjustment instructions based on the risk zoning map and output a concrete rebound rate optimization decision plan.

[0090] In this step, the adaptive control algorithm refers to a real-time decision-making system that dynamically divides spatial units according to the risk gradient; the injection parameter adjustment instruction is an atomic operation command including the pressure adjustment amount, aggregate delivery ratio and nozzle deflection angle.

[0091] In this embodiment, the real-time motion path of the spraying equipment (based on the risk zoning map output in step 105) is first obtained using trajectory tracking technology. An adaptive meshing algorithm is then used to partition the control units along the trajectory, and the unit size in high-risk areas is dynamically compressed using a gradient detection mechanism. A spatial kernel density estimation algorithm is then applied to extract the distribution density of high-rebound feature points within each unit. This is then input into a pre-trained association rule engine (based on random forest machine learning) to generate matching parameters for pressure adjustment, aggregate feed ratio correction, and nozzle deflection angle. Instruction compilation technology is used to encapsulate the parameter set into machine-readable instruction fragments. Combined with a real-time positioning system, these fragments are bound to three-dimensional coordinates to form decision items. Finally, a topological sorting algorithm is used to integrate the decision items along the entire path and serialize them into a concrete rebound rate optimization decision solution in JSON-LD format.

[0092] For example, following the tunnel vault case from step 105 (where a high-risk strip in the middle of the vault has been marked), this step acquires real-time positioning data for the jet jet along a Y+30° trajectory. A 0.8-meter compression unit is generated in the high-risk zone (1.2 meters in the conventional zone). Density estimation is used to extract the cluster characteristics of high rebound points in unit S12, triggering the association rule engine to match a control strategy for reducing pressure and increasing material. The generated "pressure -15% & delivery ratio +8%" instruction fragment is bound to the vault coordinates (35.2, 18.1, 58.1). Ultimately, the 32 units are integrated into a decision-making plan, guiding the equipment to automatically execute parameter adjustments when it reaches the S12 coordinate.

[0093] In order to solve the problem in the prior art that acoustic emission signal feature extraction is interfered with by environmental noise and cannot accurately capture the transient distribution characteristics of concrete impact energy, in some embodiments, according to step 101, the injection impact acoustic emission signal during the concrete spraying process is collected, and the time-frequency characteristics of the injection impact acoustic emission signal are extracted, including:

[0094] Step 201: capturing the original sound wave signal generated by the concrete spraying impact through a piezoelectric sensor array.

[0095] In this step, the piezoelectric sensor array refers to a multi-node sensor device arranged on the injection surface in a hexagonal topology, which converts mechanical vibration into a charge signal through the piezoelectric effect; the original acoustic wave signal is a time domain charge pulse sequence directly output by the sensor, which contains the physical vibration characteristics of the aggregate impacting the rock mass.

[0096] In this embodiment, a piezoelectric sensor array is first deployed according to the three-dimensional configuration of the spraying area, and an equidistant hexagonal topology is adopted to ensure that the sound wave covers no blind spots. When the concrete spray hits the rock mass, each sensor in the piezoelectric sensor array captures the vibration waveform in real time through the positive piezoelectric effect of the piezoelectric crystal, and each sensor generates an original sound wave signal.

[0097] Step 202: Convert the original sound wave signal into a first voltage change signal via a charge amplifier.

[0098] In this step, the charge amplifier refers to a signal conditioning device based on the principle of integral operation, which realizes the conversion of high-impedance charge signal to low-impedance voltage signal; the first voltage change signal is a standard voltage waveform processed by capacitive interference suppression, retaining the time-frequency characteristics of the sound wave.

[0099] In this embodiment, the original acoustic wave signal output from step 201 is connected to the charge amplifier input port, and sensor-amplifier impedance matching is achieved through a high-impedance input stage. An integral circuit converts the charge into a voltage, and a common-mode suppression module eliminates electromagnetic noise introduced by the transmission line. Finally, a voltage follower outputs an impedance-matched first voltage change signal.

[0100] Step 203: Perform signal envelope separation processing on the first voltage change signal to obtain a second voltage change signal reflecting the change in signal amplitude.

[0101] In this step, signal envelope separation processing refers to the digital filtering technology that extracts the instantaneous amplitude of the signal through Hilbert transform; the second voltage change signal is the envelope waveform that represents the time-varying law of the sound wave energy after stripping the carrier frequency, which is used to identify the start and end time periods of the impact event.

[0102] In this embodiment, the first voltage change signal output in step 202 is input to a digital signal processor, where a Hilbert transform algorithm is used to calculate the analytical signal, extracting its instantaneous amplitude as the envelope. A sliding mean filter is then applied to smooth the envelope fluctuations, generating a continuous second voltage change signal. The amplitude fluctuations in this signal's time-domain waveform directly correspond to the periodic variations in the impact energy of the concrete aggregate.

[0103] Step 204 : dynamically divide the time window interval based on the peak position of the second voltage change signal, and decompose the original sound wave signal into a preset number of frequency intervals within a single time window interval.

[0104] In this step, the time window interval refers to the signal analysis period adaptively defined based on the envelope peak, covering the entire life cycle of a single impact event; the frequency interval is the discrete spectrum band obtained by time-frequency decomposition, which characterizes the frequency domain distribution characteristics of the impact energy.

[0105] In this embodiment, local peaks of the second voltage variation signal output in step 203 are first identified, and time windows are dynamically divided based on adjacent peaks. Within each window, a short-time Fourier transform algorithm is applied to the original acoustic signal obtained in step 201 to decompose the time domain waveform into a predetermined number of frequency bins.

[0106] Step 205 : Count the amplitude distribution strength of the original sound wave signal in each frequency interval, and combine the amplitude distribution strength of the frequency intervals of all time window intervals to generate a time-frequency feature.

[0107] In this step, the amplitude distribution intensity refers to the total value of the sound wave energy in a single frequency interval calculated by integration operation, which represents the energy concentration in a specific frequency band; the time-frequency feature is a matrix data structure constructed by integrating the frequency domain energy distribution of the entire time window, whose row vectors correspond to the time window sequence, the column vectors represent the frequency interval, and the matrix elements store the amplitude intensity.

[0108] In this embodiment, the original acoustic signal components of each decomposed frequency interval are first extracted from the single time window interval divided in step 204. A piecewise integration algorithm is applied to calculate the absolute area value of each frequency component, generating an amplitude distribution intensity sequence in the frequency dimension. The intensity values ​​of all frequency intervals are then stacked in time window order to construct a two-dimensional intensity distribution matrix with row indexes as time windows and column indexes as frequency intervals. Ultimately, a time-frequency feature containing the spatiotemporal frequency domain energy is generated.

[0109] To address the problems in the prior art of low accuracy in reconstructing the motion trajectory of concrete particles and decoupling the velocity distribution from the spatial movement of the equipment, which results in distortion in the velocity field mapping, in some embodiments, according to step 102, collecting laser Doppler array measurement data during the concrete spraying process to obtain the concrete velocity distribution includes:

[0110] Step 301 : at least three laser detection units with spatial point distribution characteristics simultaneously receive light signals reflected by concrete particles during the spraying process, and capture signal variation characteristics of the light signals generated by the movement of the concrete particles.

[0111] In this step, the laser detection unit refers to an optical sensing device arranged in a triangulation configuration, which captures the particle reflection signal by emitting a near-infrared laser beam; the signal change characteristics are the combined parameters of the phase drift and intensity attenuation of the reflected light wave in the time domain, which characterize the motion state of the particle when it passes through the laser plane.

[0112] In this embodiment, three laser detection units are deployed along the normal to the injection trajectory, forming an equilateral triangle array. Each unit synchronously emits a near-infrared laser beam to form a detection plane. When concrete particles pass through this plane, each unit receives the reflected light signal and captures the waveform data using a high-speed photoelectric sensor. This extracts multidimensional signal variation features, including phase offset, intensity decay rate, and duration, with a time-scale accuracy of microseconds.

[0113] Step 302 : Correlating the three-dimensional movement direction information of the concrete particles according to the spatial point distribution characteristics and the signal change characteristics.

[0114] In this step, the spatial point distribution characteristics refer to the three-dimensional installation coordinate matrix of the laser unit; the three-dimensional motion direction information is the particle velocity vector solved by triangulation positioning, which includes three degree of freedom parameters: azimuth angle, pitch angle and velocity modulus.

[0115] In this embodiment, a three-dimensional coordinate system is established based on the spatial point coordinates acquired in step 301. The signal change characteristics (phase drift, intensity change rate) collected by each unit are input into the motion vector solution model. A multi-channel signal time difference cross-correlation algorithm is used to calculate the time deviation of the particle crossing each laser plane. The directional cosine value of the motion trajectory is solved in combination with spatial geometric constraints, and the three-dimensional motion direction information of the concrete particles in the inertial coordinate system is finally output.

[0116] Step 303: Aggregate the three-dimensional motion direction information according to the detection time sequence to generate instantaneous flow features with timestamps.

[0117] In this step, the detection time series refers to the particle motion observation time points arranged at fixed sampling intervals; the instantaneous flow characteristics are the distribution model of the three-dimensional velocity vectors of all particles at a single sampling moment, including the mainstream direction angle, vector discreteness and particle density parameters.

[0118] In this embodiment, the three-dimensional motion direction information output from step 302 is first sorted by millisecond timestamps, and a vector superposition algorithm is applied within each sampling time window: all particle motion direction vectors are integrated using the Rodrigues rotation formula to calculate the average azimuth and pitch angles; the vector discreteness is determined using the eigenvalue decomposition of the covariance matrix; and finally, the unit volume particle counts are fused to generate instantaneous flow characteristics with timestamps.

[0119] Step 304 : Mapping the instantaneous flow characteristics to the spatial position corresponding to the movement path of the spraying equipment to generate a concrete flow velocity distribution.

[0120] In this step, the movement path of the spraying equipment refers to the three-dimensional coordinate trajectory of the equipment recorded by the GNSS positioning system; the concrete flow velocity distribution is the data field that annotates the flow characteristic parameters in the spatial grid of the spraying path, including the mapping relationship between position coordinates and velocity vectors.

[0121] In this embodiment, a mapping function between spatial position and time is established based on the coordinate sequence of the device trajectory in the decision-making scheme of step 106. The instantaneous flow characteristics of step 303 are aligned to the device trajectory by timestamp, and the discrete time point data is continuousized through cubic spline interpolation. Coordinate offsets are compensated for by combining device posture parameters. Finally, the main flow direction angle, vector dispersion, and particle density values ​​are annotated at the spatial grid nodes covered by the spray path, resulting in a three-dimensional concrete flow velocity distribution covering the spraying area.

[0122] In order to solve the problem in the prior art of feature drift caused by spatiotemporal asynchrony of multi-source data and inability to generate rebound correlation core features for collaborative representation of physical fields, in some embodiments, according to step 103, multi-source fusion processing is performed on the time-frequency features and the concrete flow velocity distribution, and the principal component feature vector of rebound correlation is extracted from the multi-source fusion result, including:

[0123] Step 401: receiving the time-frequency characteristics and concrete flow rate distribution within the same injection time period.

[0124] In this step, the same injection time period refers to the time period from the adjustment of the injection equipment startup parameters to the completion of the current space unit operation.

[0125] In this embodiment, the time-frequency feature matrix (time window sequence - frequency interval energy distribution) generated in step 205 and the concrete flow velocity distribution (spatial coordinates - flow feature mapping) output in step 304 are synchronously received via the data bus. At the same time, a timestamp alignment engine is used to strictly match the time window sequence of the time-frequency features with the time mark points of the flow velocity distribution to ensure that the two types of data cover the same equipment operation cycle.

[0126] Step 402: Synchronously divide the time-frequency features and the concrete flow velocity distribution into feature units of the same time window.

[0127] In this step, the feature unit refers to the data block extracted by time window slice, including the acoustic wave time-frequency submatrix and the corresponding time and space velocity distribution slice; the same time window refers to the equal-length analysis period divided based on the time alignment result of step 401.

[0128] In this embodiment, according to the time alignment result of step 401, equally spaced analysis windows are divided along the time axis; time dimension slicing is performed on the time-frequency feature matrix to extract the frequency interval energy submatrix within each window; the time series of the concrete flow velocity distribution is synchronously cut to extract the flow parameter set of the spatial position within the same window; and the acoustic wave energy submatrix and the flow parameter set of the same time window are encapsulated as feature units.

[0129] Step 403 : Cross-analyze the time-frequency feature unit and the flow velocity distribution unit in each time window to generate a fusion feature unit.

[0130] In this step, cross-analysis refers to a fusion processing method that associates acoustic features with fluid motion features through a spatial correlation algorithm; the fusion feature unit is a composite data structure that integrates the sound wave energy distribution and particle motion parameters, and includes derived features such as the acoustic-fluid coupling correlation factor and the energy-direction synergy coefficient.

[0131] In this embodiment, the single time window feature unit outputted in step 402 is first read, and a spatial registration operation is performed on the time-frequency feature unit (frequency interval energy vector) and the flow velocity distribution unit (particle direction angle, discreteness, density): the spatial correlation between the acoustic energy gradient field and the particle motion vector field is calculated using the Gram matrix, and the coincidence parameters of the energy focusing area and the motion concentration area are extracted; then, a tensor product operation is applied to generate a joint distribution matrix of the acoustic fundamental frequency energy and the mainstream direction angle; and finally, a fusion feature unit including the time-frequency features, flow velocity distribution, and coupling coefficient is outputted.

[0132] Step 404: Integrate the fusion feature units of all time windows to generate a feature set.

[0133] In this step, the feature set refers to the three-dimensional tensor structure of fused feature units stacked along the time series, whose dimensions correspond to the time window index, the acoustic-flow feature dimension, and the coupling coefficient dimension, respectively, forming a dynamically evolving database covering the entire operation cycle.

[0134] In this embodiment, the fused feature units output in step 403 in the order of time windows are spatiotemporally integrated: each fused feature unit is standardized along the time axis dimension (Z-score normalized acoustic energy coefficient, radian system unified angle parameters); each unit feature is mapped to the corresponding time layer of the tensor space through a three-dimensional array stacking technique; and finally, a spatiotemporal cube structure of feature dimension (acoustic feature, flow velocity parameter, coupling coefficient) × number of time windows is constructed to form a feature set that can describe the dynamic evolution of the acoustic-fluid coupling mechanism during the injection process.

[0135] Step 405: extracting a principal component feature vector associated with the concrete rebound behavior from the feature set.

[0136] In this step, the correlation of concrete rebound behavior specifically refers to the key parameter combination in the eigenvector that reflects the rebound caused by particles detaching from the rock mass, including the coupling variation direction of the high-frequency sound scattering intensity and the particle motion discreteness.

[0137] In this embodiment, the three-dimensional tensor of the feature set generated in step 404 is first expanded into a two-dimensional feature matrix (rows: time window samples, columns: acoustic, flow, and coupling feature dimensions). A partial least squares regression algorithm is applied to construct a latent variable model of features and rebound labels, screening out a feature subspace significantly correlated with the rebound loss rate. Principal component analysis is then used to perform eigendecomposition of the covariance matrix of this subspace, retaining the orthogonal basis vectors corresponding to the largest eigenvalues. Finally, components of the basis vectors with loads exceeding a threshold are extracted to form principal component eigenvectors that map the physical mechanism of rebound. The high-frequency acoustic energy term in its directional coordinates is strongly negatively correlated with the motion discrete term.

[0138] In order to solve the problem in the prior art that static meshing destroys the topological continuity of the velocity field and cannot identify the flow state mutation point and its spatial direction attribute, in some embodiments, according to step 104, a spatial velocity field model is constructed based on the concrete velocity distribution to obtain an identification result representing the concrete rebound state, including:

[0139] Step 501: Determine the operation advancement direction based on the movement trajectory of the injection equipment.

[0140] In this step, the operation propulsion direction refers to the spatial orientation sequence of the tangent vector along the equipment motion trajectory, including the dynamic change parameters of the horizontal yaw angle and the vertical pitch angle, which are used to calibrate the main coverage axis of the concrete spraying flow in space.

[0141] In this embodiment, the jetting device GNSS trajectory coordinate sequence output from step 106 is first loaded, and coordinate fluctuations are smoothed using a time-weighted sliding window algorithm. The principles of differential geometry are then applied to calculate the direction cosines between adjacent trajectory points. The trajectory curve is parametrically differentiated to extract the projected components of the unit tangent vector in the three-dimensional coordinate system. Finally, the continuously changing horizontal yaw and vertical pitch angles of the device are output as a time series, establishing a mapping function for the device's spatial posture and propulsion direction.

[0142] Step 502: Using the concrete flow velocity distribution as input, a spatial velocity field model including a spatial topological structure is constructed.

[0143] In this example, based on concrete velocity distribution data, the spraying area is discretized into a tetrahedral grid space (Delaunay triangulation). Flow velocity parameters (mainstream direction angle, velocity modulus, and dispersion) are annotated at the grid nodes, and data at unobserved points is filled in using the Kriging interpolation algorithm. A node connectivity graph is then constructed, and the velocity gradient matrix of adjacent nodes is calculated. Ultimately, a spatial velocity field model containing gradient transfer relationships is obtained, with topological edge weights representing the spatial correlation strength of the local flow state.

[0144] Step 503: Divide the hierarchical flow framework along the operation advancing direction in the spatial velocity field model.

[0145] In this step, the hierarchical flow framework refers to a structured grid system that divides the spatial velocity field into multiple continuous levels along the operation advancement direction, each layer representing the concrete flow cross-section at a specific advancement depth; hierarchical division specifically refers to a method of generating a discretized hierarchical sequence by cutting the advancement direction through a cluster of normal planes, and the hierarchical spacing is adaptive to the velocity distribution gradient.

[0146] In this embodiment, first, based on the operation propulsion direction vector determined in step 501, its unit normal plane cluster is calculated through an orthogonalization algorithm. A series of normal planes are evenly spaced along the propulsion direction to cut the spatial velocity field model constructed in step 502. The intersection of each cutting surface and the topological grid constitutes the hierarchical boundary line. The boundary line is mapped to the three-dimensional velocity field using manifold projection technology, and the flow cross-section containing the velocity vector distribution and gradient parameters in each cutting layer is extracted. Finally, a hierarchical flow framework with spatial coherence is formed, in which adjacent layers share boundary nodes and inherit the velocity transfer relationship.

[0147] Step 504 : performing a flow state comparison operation at adjacent levels on the divided hierarchical flow framework, and identifying flow discontinuity feature points generated by the comparison operation.

[0148] In this step, the comparison of flow states at adjacent levels refers to the calculation of the direction cosine offset and modulus change rate of the velocity vector between adjacent levels; the flow discontinuity feature point is a set of spatial positions where the comparison results exceed the set threshold, representing the critical area where the particle flow undergoes collision rebound or eddy separation.

[0149] In this embodiment, the hierarchical flow framework generated in step 503 is sequentially compared (L1 → L2 → ... → Ln). First, spatial coordinate mapping functions are established for adjacent layers, and co-located nodes are associated using a nearest neighbor matching algorithm. The velocity vector angular differences and velocity modulus change rates of the matched nodes are calculated to generate a variation matrix. The Canny edge detection operator is then applied to scan the variation matrix, extracting node coordinates where the gradient abruptly changes beyond a set threshold. Finally, a set of flow discontinuity feature points and their variation parameters in three-dimensional space are output.

[0150] Step 505 : generating an identification result representing the concrete rebound state based on the spatial positions and change directions of all flow discontinuity feature points.

[0151] In this step, the identification results refer to the quantitative assessment data generated by integrating the spatial position clustering pattern and the variation direction vector, including three core outputs: rebound risk classification map, energy loss heat map and rebound probability matrix, which are used to identify the critical area and severity of rebound caused by concrete particles detaching from the rock mass.

[0152] In this embodiment, spatial density cluster analysis is first performed on the set of flow discontinuity feature points identified in step 504. Adjacent feature points with consistent variation directions are merged into rebound risk clusters using the DBSCAN algorithm. The average directional change vector of the feature points within each cluster is calculated, and the principal rebound inclination axis of the cluster is generated using the principle of vector synthesis. Simultaneously, the rebound energy loss coefficient of each cluster is calculated based on the principal component eigenvectors from step 405. Ultimately, a rebound risk grading map (colored by energy loss coefficient), a heat map (rendering intensity by particle detachment probability), and a probability matrix (marking the probability of rebound occurrence for each grid cell) bound to three-dimensional spatial coordinates are generated, completing the quantitative identification of the concrete rebound state.

[0153] To address the problem in the prior art that risk zones are separated from equipment motion trajectories and that clustering results cannot be accurately mapped to the physical space of the spraying operation, in some embodiments, according to step 105, cluster analysis is performed on the principal component eigenvectors and the identification results to identify high-rebound mode categories and low-rebound mode categories to generate a risk zone map, including:

[0154] Step 601 : Bind each component of the principal component eigenvector to the spatial coordinates of the identification result, create a spatial state recording point, and calculate the rebound state correlation index between each spatial state recording point.

[0155] In this step, the spatial state recording point refers to the three-dimensional data node that integrates the principal component characteristic component value and the spatial position, including coordinates, characteristic components and rebound state parameters; the rebound state association index refers to the feature similarity measurement value between adjacent points calculated by the spatial autocorrelation function, which characterizes the spatial propagation correlation of the rebound risk.

[0156] In this embodiment, the components of the principal component eigenvectors (such as the high-frequency acoustic energy coefficient and particle dispersion angle coefficient) from step 405 are first data-bound with the identification results (rebound probability and risk level) from the same spatial location in step 505. A spatial state record point is generated for each spatial location, containing three-dimensional coordinates, a set of characteristic components, and rebound state parameters. Subsequently, based on the principles of geographic information systems, the Euclidean distance matrix between all recorded points is calculated. A Gaussian kernel spatial autocorrelation function is applied to map the distances into correlation strength indices (the smaller the distance, the greater the index), ultimately generating a rebound state correlation index between the recorded points.

[0157] Step 602 : Divide the recording point set into a dense area and a sparse area according to the rebound state association index.

[0158] In this step, the dense area record point set refers to the spatial node cluster whose correlation index is greater than the connectivity threshold, reflecting the aggregation area with strong continuity of rebound characteristics; the sparse area record point set refers to the discrete node set whose correlation index is lower than the threshold, representing the isolated point where rebound occurs or the characteristic mutation boundary.

[0159] In this embodiment, a weighted undirected graph (nodes: spatial state record points, edge weights: association index values) is constructed based on the association index matrix generated in step 601. A graph-theoretic connected component analysis algorithm is applied to scan the network, filtering strongly connected edges based on a set connectivity threshold. The nodes of the largest connected subgraph are extracted to form the dense region record point set. The remaining unconnected nodes and weakly connected nodes are assigned to the sparse region record point set.

[0160] Step 603: Mark the dense area recording point set as a high-resilience mode category, and mark the sparse area recording point set as a low-resilience mode category.

[0161] In this embodiment, based on the set of recorded points in the dense and sparse regions demarcated in step 602, a clustered label propagation algorithm is first applied to assign category attributes to each spatial node. All nodes in the dense region are assigned a unified high-resilience pattern identifier (e.g., "HRC-1"), and their spatial coverage density parameters are calculated. Nodes in the sparse region are individually labeled with a low-resilience pattern number (e.g., "LRC-35") and event independence annotations are added. Finally, a mapping table is generated from spatial encoding to pattern category, in which high-resilience categories are associated with core rebound conduction areas, and low-resilience categories are mapped to isolated disturbance points.

[0162] Step 604: Determine a concrete spraying coverage area based on the spatial coordinate distribution range of the identification result.

[0163] In this step, the concrete spraying coverage area refers to the effective concrete adhesion area delineated within the spraying operation space based on the rebound state identification results. Its rigid boundary range is determined by the spatial distribution extreme value of the high-confidence low-rebound mode point, and the output form is a set of coordinate ranges of the permitted operating area that can guide the operation of the spraying equipment.

[0164] In this embodiment, the coordinates of all spatially recorded points in the low-rebound mode category marked in step 603 are first extracted, and a convex hull algorithm is used to generate a minimum circumscribed polygon. The spatial coordinate extrema are calculated based on the vertex coordinates of this polygon, forming a spatial envelope of the coverage domain. Simultaneously, the high-rebound mode category region marked in step 603 is defined as an absolute avoidance domain. A Boolean subtraction operation is performed within the envelope to eliminate the avoidance domain space. Finally, a set of rigid boundary coordinates for the spray coverage domain is output, defining the three-dimensional working space for safe spraying as a vertex sequence.

[0165] Step 605 : Project the spatial coordinates of the high-rebound mode category and the low-rebound mode category back to the concrete spraying coverage area to generate a risk zoning map covering the spraying surface.

[0166] In this step, spatial coordinate projection refers to the projection operation of mapping the three-dimensional coordinates of the high / low rebound pattern points to the coverage domain grid nodes, and achieving attribute inheritance through nearest neighbor matching.

[0167] In this embodiment, the concrete spraying coverage area determined in step 604 is first discretized into a grid array of equal-sized cubes, generating uniquely numbered spatial grid cells. Coordinate projection is then performed on each high / low rebound pattern point identified in step 603: the grid cell to which it belongs is located using Euclidean distance calculation, and the pattern category label and principal component eigenvalue of the point are written into the corresponding grid attribute table. For grid cells receiving multiple pattern points, a majority voting strategy is used to determine the final risk label (high / low rebound partition). Finally, a mapping table of grid coordinates and risk labels is output, forming a risk partition map covering the entire area.

[0168] To address the problem in existing technologies where fixed control rules ignore regional risk density differences, leading to energy waste in non-risk areas and control failure in high-risk areas, in some embodiments, according to step 106, based on the risk zoning map, an adaptive control algorithm is applied to dynamically generate spraying parameter adjustment instructions and output a concrete rebound rate optimization decision plan, including:

[0169] Step 701 : Divide the risk partition map into continuous dynamic control units along the movement trajectory of the injection device, and extract the distribution density value of the high rebound mode category in each dynamic control unit.

[0170] In this step, the continuous dynamic control unit refers to the columnar space segment divided on the risk zoning map along the movement trajectory of the injection equipment, and its length is adaptively adjusted with the rebound risk gradient; the distribution density value refers to the ratio of the number of high rebound mode grids in the unit to the total number of grids, which represents the rebound risk concentration in the local area.

[0171] In this embodiment, the risk zoning grid data generated in step 605 is first imported. The unit tangent vector of the trajectory is calculated based on the device movement trajectory determined in step 501. An adaptive sliding window algorithm is applied along the trajectory to divide the control units. The window length automatically expands and contracts based on the risk gradient difference between the preceding and following units (high-risk areas have windows compressed, while low-risk areas have windows expanded). Within each control unit, the ratio of the number of high-resilience mode grids to the total number of grids is calculated as the distribution density value.

[0172] Step 702: establishing an association rule base between the injection outlet pressure of the injection device and the material delivery ratio according to the distribution density value.

[0173] In this example, we first collected the distribution density values ​​of dynamic control units and their corresponding optimal control parameters (pressure adjustment coefficient and delivery ratio correction value) from historical projects. We then applied a random forest regression algorithm to train a model mapping density values ​​to parameter sets: the input density value was used, and the output pressure adjustment coefficient and delivery ratio correction coefficient were used. Model hyperparameters were optimized through ten-fold cross-validation, and decision paths were extracted to form an if-then rule set. Finally, a confidence-weighted association rule base was constructed.

[0174] Step 703 : Generate an instruction fragment based on the entry of the association rule base of the dynamic control unit where the injection position of the current injection device is located.

[0175] In this step, the injection position of the injection equipment refers to the spatial coordinates of the equipment obtained in real time through the GNSS / RTK positioning system; the instruction fragment is an atomic control command containing the pressure adjustment amount, aggregate delivery ratio and nozzle deflection angle, which is bound to a specific dynamic control unit spatial range to take effect.

[0176] In this embodiment, the real-time 3D coordinates of the spraying device are first obtained. A spatial inclusion detection algorithm is then used to match the dynamic control units identified in step 701. The corresponding entry in the association rule library for this unit in step 702 is extracted, and the pressure adjustment coefficient, delivery ratio correction factor, and nozzle deflection parameters are analyzed. The parameter values ​​are compiled into a device executable instruction format (e.g., a MODBUS register address mapping), generating an instruction fragment containing the spatial scope, parameter set, and timestamp.

[0177] Step 704 : Combining all instruction segments along the movement trajectory of the injection device to generate a dynamic injection parameter adjustment instruction set.

[0178] In this step, the dynamic injection parameter adjustment instruction set refers to a serialized set of instruction fragments arranged in sequence along the equipment movement trajectory, and its triple synchronization characteristics of space-time-control parameters ensure continuous adaptive regulation of the injection operation.

[0179] In this embodiment, the instruction segments are first topologically sorted according to the operation advancement direction in step 501 to establish a spatial order of trajectory. A timestamp alignment engine is applied to ensure seamless transitions between the effective periods of adjacent segments, and spatial overlap detection eliminates regional conflicts. The sorted instruction segment sequence is then encapsulated into a binary stream format, with a frame header checksum and device address identifier added. This ultimately generates a dynamic adjustment instruction set that can be directly written into the injection device controller.

[0180] Step 705 : Bind the dynamic spraying parameter adjustment instruction set to the spatial mapping relationship of the risk partition map, generate partition decision items with control units as the management granularity, integrate all partition decision items, and output a concrete rebound rate optimization decision plan.

[0181] In this step, the partition decision entry refers to the decision atomic unit that binds the spatial range of the dynamic control unit to its corresponding injection parameter instruction, including the unit coordinate boundary, association rule output parameters and spatial binding verification code; the concrete rebound rate optimization decision scheme is an executable file package that integrates all partition entries to achieve millisecond-level mapping of the spatial position of the injection operation to the control parameters.

[0182] In this embodiment, a spatial mapping is first established between the dynamic injection parameter adjustment instruction set output in step 704 and the risk partitioning map in step 605. The spatial scope of each instruction fragment in the instruction set is extracted, a hash matching operation is performed on the risk partitioning grid, and a spatial binding verification code is generated to ensure precise coordinate correspondence. A partitioning decision entry is created for each dynamic control unit, consisting of a sequence of unit vertex coordinates, a binary stream of parameter instructions, and a spatial verification code triple. A topological constraint algorithm is applied to optimize the sequence of these entries, ultimately integrating them into a global decision solution in JSON-LD format, enabling injection devices to retrieve parameter instructions based on real-time location indexes.

[0183] Figure 2 A structural diagram of a concrete spraying rebound rate optimization decision system is provided for an embodiment of the present application. Figure 2 As shown, the system includes:

[0184] The first acquisition module 21 is used to collect the injection impact acoustic emission signal during the concrete spraying process and extract the time-frequency characteristics of the injection impact acoustic emission signal;

[0185] The second acquisition module 22 is used to collect laser Doppler array measurement data during the concrete spraying process to obtain the concrete flow velocity distribution;

[0186] A fusion module 23 is configured to perform multi-source fusion processing on the time-frequency features and the concrete flow velocity distribution, and extract a principal component feature vector associated with rebound from the multi-source fusion result;

[0187] A construction module 24 is used to construct a spatial velocity field model based on the concrete flow velocity distribution to obtain an identification result representing the concrete rebound state;

[0188] An analysis module 25 is configured to perform cluster analysis on the principal component eigenvector and the identification result to identify high-resilience pattern categories and low-resilience pattern categories to generate a risk zoning map;

[0189] The output module 26 is used to apply the adaptive control algorithm to dynamically generate injection parameter adjustment instructions based on the risk zoning map and output a concrete rebound rate optimization decision plan.

[0190] Figure 2 The concrete spraying rebound rate optimization decision system can be executed Figure 1The implementation principles and technical effects of the concrete shotcrete rebound rate optimization decision-making method described in the illustrated embodiment are not further elaborated. The specific manner in which each module and unit performs operations in the concrete shotcrete rebound rate optimization decision-making system in the above embodiment has been described in detail in the embodiment of the method and will not be further elaborated here.

[0191] In one possible design, Figure 2 A concrete spraying rebound rate optimization decision system of the embodiment shown can be implemented as a computing device, such as Figure 3 As shown, the computing device may include a storage component 31 and a processing component 32;

[0192] The storage component 31 stores one or more computer instructions, wherein the one or more computer instructions are called and executed by the processing component 32 .

[0193] The processing component 32 is used for the above Figure 1 The embodiment provides a concrete spraying rebound rate optimization decision-making method.

[0194] The processing component 32 may include one or more processors to execute computer instructions to complete all or part of the steps in the above method. Of course, the processing component may also be implemented as one or more application-specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processing devices (DSPDs), programmable logic devices (PLDs), field programmable gate arrays (FPGAs), controllers, microcontrollers, microprocessors, or other electronic components to perform the above method.

[0195] The storage component 31 is configured to store various types of data to support operations at the terminal. The storage component can be implemented by any type of volatile or non-volatile memory 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.

[0196] Of course, a computing device may also include other components, such as input / output interfaces, display components, communication components, etc.

[0197] The input / output interface provides an interface between the processing component and the peripheral interface module, which can be an output device, an input device, etc.

[0198] The communication component is configured to facilitate, among other things, wired or wireless communications between the computing device and other devices.

[0199] Among them, the computing device can be a physical device or an elastic computing host provided by a cloud computing platform, etc. In this case, the computing device can refer to a cloud server, and the above-mentioned processing components, storage components, etc. can be basic server resources rented or purchased from the cloud computing platform.

[0200] The present application also provides a computer storage medium storing a computer program, wherein the computer program can achieve the above-mentioned Figure 1 The embodiment shown is a method for optimizing the decision-making of the rebound rate of concrete spraying.

[0201] Those skilled in the art will clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0202] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units, i.e., they may be located in one location or distributed across multiple network units. Some or all of the modules may be selected based on actual needs to achieve the objectives of the present embodiment. Persons of ordinary skill in the art will be able to understand and implement the present invention without inventive effort.

[0203] Through the above description of the embodiments, those skilled in the art will clearly understand that each embodiment can be implemented using software plus a necessary general-purpose hardware platform, or of course, hardware. Based on this understanding, the essence of the above technical solution, or the portion that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, a magnetic disk, or an optical disk, and includes a number of instructions for causing a computer device (such as a personal computer, server, or network device) to execute the methods described in each embodiment or certain portions of the embodiments.

[0204] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A concrete spraying rebound rate optimization decision method, characterized in that: include: Collecting the injection impact acoustic emission signal during the concrete spraying process and extracting the time-frequency characteristics of the injection impact acoustic emission signal; Collect laser Doppler array measurement data during concrete spraying to obtain concrete flow velocity distribution; Performing multi-source fusion processing on the time-frequency characteristics and the concrete flow velocity distribution, and extracting a principal component feature vector of rebound correlation from the multi-source fusion result; Constructing a spatial velocity field model based on the concrete flow velocity distribution to obtain an identification result representing the concrete rebound state; performing cluster analysis on the principal component eigenvectors and the identification results to identify high-resilience pattern categories and low-resilience pattern categories to generate a risk zoning map; Based on the risk zoning map, an adaptive control algorithm is applied to dynamically generate spraying parameter adjustment instructions and output a concrete rebound rate optimization decision plan.

2. The method according to claim 1, characterized in that Collecting the jet impact acoustic emission signal during the concrete spraying process and extracting the time-frequency characteristics of the jet impact acoustic emission signal, including: The original acoustic wave signal generated by the impact of concrete spraying is captured by a piezoelectric sensor array; Converting the original sound wave signal into a first voltage change signal via a charge amplifier; performing signal envelope separation processing on the first voltage change signal to obtain a second voltage change signal reflecting a change in signal amplitude; Dynamically dividing the time window interval based on the peak position of the second voltage change signal, and decomposing the original sound wave signal into a preset number of frequency intervals within a single time window interval; The amplitude distribution intensity of the original sound wave signal in each frequency interval is counted, and the amplitude distribution intensity of the frequency intervals of all time window intervals is combined to generate time-frequency features.

3. The method according to claim 1, characterized in that Collect laser Doppler array measurement data during the concrete spraying process to obtain the concrete flow velocity distribution, including: At least three laser detection units with spatial point distribution characteristics simultaneously receive light signals reflected by concrete particles during the spraying process, and capture signal change characteristics of the light signals generated by the movement of the concrete particles; Correlating three-dimensional movement direction information of concrete particles according to the spatial point distribution characteristics and the signal change characteristics; Aggregating the three-dimensional motion direction information according to the detection time series to generate instantaneous flow features with timestamps; The instantaneous flow characteristics are mapped to the spatial position corresponding to the movement path of the spraying equipment to generate the concrete flow velocity distribution.

4. The method according to claim 1, wherein Performing multi-source fusion processing on the time-frequency characteristics and the concrete flow velocity distribution, and extracting a principal component feature vector of rebound correlation from the multi-source fusion result, including: Receive the time-frequency characteristics and concrete flow rate distribution within the same spraying time period; Synchronously dividing the time-frequency characteristics and the concrete flow velocity distribution into characteristic units of the same time window; Cross-analyze the time-frequency feature unit and the flow velocity distribution unit in each time window to generate a fusion feature unit; Integrate the fusion feature units of all time windows to generate a feature set; A principal component feature vector associated with the concrete rebound behavior is extracted from the feature set.

5. The method according to claim 1, wherein A spatial velocity field model is constructed based on the concrete flow velocity distribution to obtain an identification result representing the concrete rebound state, including: Determine the direction of operation based on the movement trajectory of the spraying equipment; Taking the concrete flow velocity distribution as input, a spatial velocity field model including a spatial topological structure is constructed; Dividing a hierarchical flow framework along the operation advancing direction in the spatial velocity field model; Performing a flow state comparison operation on adjacent layers on the divided hierarchical flow framework, and identifying flow discontinuity feature points generated by the comparison operation; Based on the spatial position and change direction of all flow discontinuity feature points, identification results representing the concrete rebound state are generated.

6. The method according to claim 1, characterized in that Performing cluster analysis on the principal component eigenvectors and the identification results to identify high-resilience mode categories and low-resilience mode categories to generate a risk zoning map, including: Bind each component of the principal component eigenvector to the spatial coordinates of the identification result, create spatial state recording points, and calculate the rebound state correlation index between each spatial state recording point; Dividing a dense area recording point set and a sparse area recording point set according to the rebound state correlation index; The set of recording points in the dense area is marked as a high-rebound mode category, and the set of recording points in the sparse area is marked as a low-rebound mode category; Determining a concrete spraying coverage area based on a spatial coordinate distribution range of the identification result; The spatial coordinates of the high-rebound mode category and the low-rebound mode category are projected back to the concrete shotcrete coverage area to generate a risk zoning map covering the shotcrete surface.

7. The method according to claim 1, characterized in that Based on the risk zoning map, an adaptive control algorithm is applied to dynamically generate spraying parameter adjustment instructions and output a concrete rebound rate optimization decision plan, including: On the risk zoning map, continuous dynamic control units are divided along the movement trajectory of the injection device, and a distribution density value of a high rebound mode category within each dynamic control unit is extracted; Establish an association rule base between the injection outlet pressure of the injection equipment and the material delivery ratio according to the distribution density value; generating an instruction fragment based on an entry of an association rule base of a dynamic control unit at the injection position of the current injection device; Combining all instruction fragments along the movement trajectory of the injection device to generate a dynamic injection parameter adjustment instruction set; The dynamic spraying parameter adjustment instruction set is bound to the spatial mapping relationship of the risk partition map to generate partition decision items with the control unit as the management granularity, and all partition decision items are integrated to output a concrete rebound rate optimization decision plan.

8. A concrete spraying rebound rate optimization decision system, characterized in that: include: A first acquisition module is used to collect the injection impact acoustic emission signal during the concrete spraying process and extract the time-frequency characteristics of the injection impact acoustic emission signal; The second acquisition module is used to collect laser Doppler array measurement data during the concrete spraying process to obtain the concrete flow velocity distribution; a fusion module, configured to perform multi-source fusion processing on the time-frequency features and the concrete flow velocity distribution, and extract a principal component feature vector associated with rebound from the multi-source fusion result; A construction module is used to construct a spatial velocity field model based on the concrete flow velocity distribution to obtain an identification result representing the concrete rebound state; an analysis module, configured to perform cluster analysis on the principal component eigenvector and the identification result, identify high-resilience pattern categories and low-resilience pattern categories, and generate a risk zoning map; The output module is used to dynamically generate injection parameter adjustment instructions based on the risk zoning map using an adaptive control algorithm and output a concrete rebound rate optimization decision plan.

9. A computing device, characterized in that The method comprises a processing component and a storage component; the storage component stores one or more computer instructions; the one or more computer instructions are used to be called and executed by the processing component to implement a concrete spraying rebound rate optimization decision method according to any one of claims 1 to 7.

10. A computer storage medium, characterized in that A computer program is stored, and when the computer program is executed by a computer, the method for optimizing the decision-making of the rebound rate of concrete spraying according to any one of claims 1 to 7 is implemented.

Images