Intelligent computing processing system and method for tunnel blasting multi-parameter data

Abstract

The present application relates to the field of tunnel engineering blasting technology, and discloses a kind of intelligent computing processing system and method of tunnel blasting multi-parameter data, comprising: first, establish space mileage model and parameter mapping dictionary, based on target design mileage stake number executes interpolation operation, obtains dynamic reference parameter;Then the dynamic parameter is fused with asymmetric space compensation matrix, generates heterogeneous structure influence coefficient and deduces actual charge quantity, and after safety checking and calculation, issue blasthole arrangement parameter;Subsequently, the continuous vibration waveform data generated by actual blasting is collected, energy envelope stripping analysis is performed based on the initiation timing, and the local correction deviation is obtained;Finally, based on the local correction deviation, the compensation matrix is updated and called for the next cycle of charge calculation. The present application eliminates the step error of geological properties, realizes the differentiated charge calculation of different types of blastholes in the same section, and constructs a system based on the measured response closed-loop adaptive iteration.

Description

Technical Field

[0001] This invention relates to the field of tunnel blasting technology, specifically to an intelligent calculation and processing system and method for multi-parameter data in tunnel blasting. Background Technology

[0002] In tunnel drilling and blasting construction, the reasonable calculation of explosive charge and the design of blasting parameters are the core aspects to ensure the quality of excavation and the safety of the surrounding environment. However, existing blasting calculation and processing methods have significant limitations in terms of multi-source data fusion and dynamic parameter optimization.

[0003] Traditional geological exploration data is typically calibrated using discrete, fixed-section divisions. When selecting blasting parameters, directly using static geological boundary points can lead to abrupt changes in parameters at the interfaces. This discretization ignores the continuous and gradual changes in rock mass properties in real engineering geological environments, causing the values ​​of basic blasting parameters in geological transition zones to deviate from the actual rock mass conditions and reducing the reliability of the calculation data source.

[0004] Meanwhile, in the process of calculating the charge amount for a specific cross-section, conventional blasting designs typically employ a homogeneous calculation model for the entire cross-section, using uniform rock parameters to control the charge distribution across the entire cross-section. In reality, the free-face conditions faced by cut holes, auxiliary holes, and peripheral holes at different locations on the tunnel excavation cross-section are drastically different from the rock mass's clamping effect. The homogeneous parameter allocation fails to consider this asymmetry in the spatial stress structure, making differentiated and precise control impossible and easily leading to overcharging or insufficient rock-breaking energy in localized areas.

[0005] Furthermore, existing blasting parameter designs largely rely on static engineering experience charts or preliminary single-test blast results, lacking an adaptive closed-loop adjustment mechanism based on actual blasting response feedback. In long-distance tunneling, due to the heterogeneity of underground rock masses, if blasting parameters cannot be decoupled and inverted based on the actual continuous vibration waveform data generated in each tunneling cycle, the calculation system will be unable to adapt to the evolution of the geological environment, making it difficult for the blasting parameter system to continuously and adaptively optimize. Summary of the Invention

[0006] To address the shortcomings of existing technologies, this invention provides an intelligent calculation and processing system and method for multi-parameter data in tunnel blasting. It solves the problems of abrupt transition zone errors caused by discrete calibration of geological parameters in traditional blasting design, neglect of spatial stress differences in the calculation of homogeneous charges across the entire section, and lack of a closed-loop adaptive correction mechanism based on actual blasting response.

[0007] To achieve the above objectives, the first aspect of the present invention provides an intelligent calculation and processing system for multi-parameter data of tunnel blasting, comprising: The data initialization module is used to establish the spatial mileage model and baseline parameters. The parameter interpolation module is used to obtain the corresponding dynamic reference parameters in the spatial mileage model based on the received target design mileage station number. The charge calculation module is used to retrieve a preset asymmetric space compensation matrix and fuse the dynamic reference parameters with the asymmetric space compensation matrix to calculate the actual charge amount. The safety verification module is used to perform safety verification based on the actual charge amount, and outputs the borehole layout parameters after determining that the verification is successful. The signal extraction module is used to acquire the actual continuous vibration waveform data generated by blasting based on the borehole layout parameters, and to analyze the actual continuous vibration waveform data to obtain the local correction deviation. The matrix update module is used to update the asymmetric spatial compensation matrix based on the local correction deviation, for use in the next blasting loop.

[0008] Preferably, the benchmark parameters include a benchmark rock firmness coefficient and a benchmark rock mass structure influence coefficient that are tied to the geological section of the tunnel; The data initialization module is also used to establish a detonation time sequence vector that associates borehole type with detonation timing.

[0009] Preferably, the corresponding dynamic reference parameters are obtained, specifically including: When the distance between the target design mileage station and the geological boundary point in the spatial mileage model is less than or equal to the preset transition buffer length threshold associated with the detection accuracy of advanced geological prediction equipment, and the surrounding rock grade labels on both sides of the geological boundary point are inconsistent, the benchmark parameters on both sides of the geological boundary point are extracted and spatial distance weighted interpolation is performed to obtain the dynamic benchmark parameters. When the distance between the target design mileage station and the geological boundary point is greater than the preset transition buffer length threshold, the benchmark parameters of the current section are directly used as the dynamic benchmark parameters.

[0010] Preferably, the spatial distance weighted interpolation operation is performed by using spatial distance as a weight attenuation factor to perform first-order continuous spatial distance weighted interpolation calculation on the reference parameters on both sides of the geological boundary point, thereby generating the dynamic reference parameters with a smooth transition.

[0011] Preferably, the asymmetric spatial compensation matrix is ​​an asymmetric spatial compensation matrix that includes spatial compensation terms corresponding to various types of boreholes; The calculation of the actual charge amount specifically includes: superimposing the dynamic reference parameters with the corresponding spatial compensation terms in the asymmetric spatial compensation matrix to obtain the heterogeneous structure influence coefficient, and inputting the heterogeneous structure influence coefficient into the explosive unit consumption evolution model based on multi-parameter coupling to calculate the actual charge amount corresponding to various types of boreholes.

[0012] Preferably, a safety verification is performed based on the actual charge quantity, specifically including: The maximum single-segment charge in the actual charge is extracted and substituted into the Sadovsky vibration velocity theory model to predict the theoretical peak vibration velocity. When the theoretical peak vibration velocity exceeds the preset site safety velocity threshold corresponding to the structure type of the protected building, the output command is intercepted, and the charge calculation module is triggered to recalculate the actual charge after reducing the reference rock mass structure influence coefficient by a predetermined step size.

[0013] Preferably, before analyzing the actual continuous vibration waveform data to obtain the local correction deviation, the method further includes: Using the set detonation time of the subsequent borehole as a node, the residual energy envelope is predicted by utilizing the vibration response characteristics of the preceding borehole, and residual waveform stripping calculation is performed from the superimposed actual continuous vibration waveform data to separate and obtain the independent purified time-series feature vector of each borehole.

[0014] Preferably, the actual vibration continuous waveform data is analyzed to obtain the local correction deviation, specifically including: The independent purification time-series feature vectors are input into a one-dimensional convolutional neural network for feature extraction, and feature values ​​characterizing energy transfer efficiency are output. The feature values ​​are weighted and solved with the degree of deviation of vibration response intensity to calculate the local correction deviation.

[0015] Preferably, updating the asymmetric spatial compensation matrix based on the local correction deviation specifically includes: Based on the momentum retention coefficient, a smooth iterative inversion is performed on the historical space compensation term and the local correction deviation to generate a new asymmetric space compensation matrix. When it is determined that the root mean square rate of change of the asymmetric spatial compensation matrix is ​​less than the preset convergence threshold calculated based on the blasting tolerance error for multiple consecutive tunneling cycles, the momentum retention coefficient is automatically increased to switch to steady-state fine-tuning mode. When it is determined that the root mean square rate of change of the asymmetric spatial compensation matrix does not meet the condition that multiple consecutive tunneling cycles are less than the convergence threshold, the current momentum retention coefficient is maintained unchanged.

[0016] A second aspect of this invention provides an intelligent calculation and processing method for multi-parameter data in tunnel blasting, comprising the following steps: Establish spatial mileage model and benchmark parameters; Based on the received target design mileage station number, obtain the corresponding dynamic benchmark parameters in the spatial mileage model; The preset asymmetric space compensation matrix is ​​retrieved, and the dynamic reference parameters are fused with the asymmetric space compensation matrix to calculate the actual charge amount. A safety calculation is performed based on the actual charge amount, and the borehole layout parameters are output after the calculation is deemed to be passed. Obtain the actual continuous vibration waveform data generated by blasting based on the borehole layout parameters, and analyze the actual continuous vibration waveform data to obtain the local correction deviation. The asymmetric spatial compensation matrix is ​​updated based on the local correction deviation and then called in the next blasting loop.

[0017] This invention provides an intelligent calculation and processing system and method for multi-parameter data in tunnel blasting. It has the following beneficial effects: 1. This invention introduces a spatial mileage model and a parameter interpolation module to design mileage stations for targets located in the surrounding rock transition buffer zone. It extracts the parameters of the benchmark nodes of the geological boundary points on both sides and performs spatial distance weighted interpolation calculation. This overcomes the limitations of the discrete segment division of traditional geological exploration data, eliminates the step abrupt error of parameters at the interface of geological segments, and enables the generated dynamic benchmark parameters to objectively reflect the continuous and gradual change law of the rock mass properties on site, thereby improving the reliability of the bottom geological parameter values.

[0018] 2. This invention constructs a charge calculation mechanism based on an asymmetric spatial compensation matrix. It integrates the dynamic benchmark rock mass structure influence coefficient with the spatial compensation terms corresponding to various boreholes to generate heterogeneous structure influence coefficients to calculate the actual charge. This changes the homogeneous allocation mode of controlling the charge of the entire cross section with a single parameter in traditional blasting design. It fully considers the differences in the free face and clamping effect faced by different areas such as slotting, auxiliary and surrounding areas on the same cross section, and realizes multi-dimensional differentiated extrapolation of the charge of each borehole. It avoids the problem of local over-charge or under-charge caused by single mean calculation.

[0019] 3. Based on the acquired actual blasting vibration waveform data, this invention uses the detonation time series to perform residual waveform stripping calculation to separate the independent vibration response characteristics of each blast hole, generate local correction deviation, and rely on the momentum smoothing update mechanism to perform inversion iteration on the asymmetric spatial compensation matrix, establishing a data closed-loop channel from actual response feedback to front-end computing power compensation. This overcomes the shortcomings of traditional blasting parameters relying on static experience and being easily affected by single random errors, enabling the computing system to have the ability to adaptively correct and smoothly evolve the underlying parameters as the tunnel excavation process progresses. Attached Figure Description

[0020] Figure 1 A system module architecture diagram provided for embodiments of the present invention; Figure 2 This is a schematic diagram of the method flow provided in an embodiment of the present invention; Figure 3 A comparison diagram of the blasting vibration velocity control effects of different design methods provided in the embodiments of the present invention; Figure 4 This is a verification diagram of the adaptive convergence characteristics of the matrix update module provided in an embodiment of the present invention. Detailed Implementation

[0021] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0022] See attached document Figure 1 , Figure 1 This is a schematic diagram of the architecture of an intelligent calculation and processing system for multi-parameter data in tunnel blasting according to an embodiment of the present invention. The present invention provides an intelligent calculation and processing system for multi-parameter data in tunnel blasting, which may include: a data initialization module, a parameter interpolation module, a charge calculation module, a safety verification module, a signal extraction module, and a matrix update module.

[0023] Each computing module is deployed in a computing server containing a processor and storage media. The computing server establishes data connections with the field acquisition terminal and execution terminal via a communication network. The field acquisition terminal includes a vibration meter deployed at the blasting site to collect the full-time history signal of the actual blasting vibration. The execution terminal is used to receive and display the blasting calculation instructions and borehole layout parameters issued by the computing server.

[0024] The data initialization module is used to construct the underlying data structure, establish the tunnel's basic attributes, spatial mileage model, and parameter dictionary mapping matrix. The parameter interpolation module is connected to the data initialization module and is used to identify the location attributes of the target design mileage station, perform spatial distance weighted interpolation calculations based on the geological parameters of adjacent survey nodes, and output the dynamic benchmark parameters of the target section.

[0025] The charge calculation module establishes data connections with both the parameter interpolation module and the matrix update module. The charge calculation module integrates dynamic baseline parameters with the asymmetric spatial compensation matrix to generate heterogeneous structure influence coefficients, and uses a unit consumption model to deduce the actual charge amount for various borehole types. The safety verification module receives the actual charge amount, performs theoretical calculations using the Sadovsky vibration velocity model, generates a safety compliance judgment result, and triggers the scheme issuance command.

[0026] The signal extraction module receives raw, full-time vibration velocity waveform data transmitted from the field acquisition terminal. Based on the time series data configured in the excavation process dictionary, the module divides the observation window, performs envelope stripping and waveform decoupling calculations, outputs a purified time-series feature vector, and extracts local correction deviations based on amplitude. The matrix update module establishes an inverse analytical model of attenuation based on the local correction deviations, derives the parameter correction deviations, reconstructs the asymmetric spatial compensation matrix, and updates it to the storage medium.

[0027] See attached document Figure 2 , Figure 2 This is a flowchart illustrating an intelligent calculation and processing method for multi-parameter data in tunnel blasting according to an embodiment of the present invention. The present invention provides an intelligent calculation and processing method for multi-parameter data in tunnel blasting, comprising the following steps: S10, Read the tunnel basic attribute data and establish the mapping relationship between the surrounding rock classification dictionary and the excavation process dictionary; S20: Receive the target design mileage station number input, determine whether the target design mileage station number is in the surrounding rock transition buffer zone, and if it is in the buffer zone, extract the static parameters of the adjacent two-sided nodes and perform weighted interpolation to generate the dynamic benchmark parameters corresponding to the target design mileage station number. S30, read the preset asymmetric spatial compensation matrix recorded in the storage medium, fuse the spatial compensation terms in the preset asymmetric spatial compensation matrix with the dynamic reference parameters, obtain the independent heterogeneous influence coefficient of each spatial region, and calculate the explosive consumption per unit and the total charge of the group of holes corresponding to various types of boreholes by combining the rock firmness coefficient and explosive property parameters. S40: Obtain the peak data of the total charge of the group of holes in each delay section, calculate the theoretical peak vibration velocity, and after determining that the safety threshold condition is met, issue the hole layout parameters of the current section. S50 reads the actual continuous waveform data of vibration generated by blasting operations, extracts the waveform window according to the time sequence vector configured in the excavation process dictionary, and uses the exponential recursive decay equation to perform residual waveform stripping calculation to extract and purify vibration extreme values. S60, substitute the purified vibration extreme value into the reverse attenuation model equation to calculate the actual rock mass structure deviation, update the asymmetric spatial compensation matrix, and establish a spatial data binding relationship with the next tunneling section.

[0028] To achieve adaptive intelligent computation of multi-parameter data, the system performs underlying data initialization operations before executing blasting calculations. The data initialization module is used to construct the underlying data structure, establish basic tunnel attributes, a spatial mileage model, and a parameter dictionary mapping matrix. The specific operational logic of the data initialization module includes the following steps: S101, Establish the tunnel basic attribute ledger and spatial mileage model. In this embodiment, the system receives externally input tunnel engineering basic data and generates a tunnel basic attribute ledger. This ledger serves as the underlying foundation for subsequent data association and includes tunnel design level and standard cross-sectional profile design documents. Based on the above basic data, the system constructs a spatial mileage model reflecting the spatial distribution of the tunnel. Considering that all structural design and construction parameters in tunnel engineering are unfolded along the axial direction, as a preferred approach, the system uses the tunnel centerline as a reference and establishes a continuous one-dimensional linear mileage station sequence from the design start point to the end point as the index key. Corresponding to the three-dimensional profile of the tunnel, the system constructs a spatial coordinate set with this sequence as the vertical axis and the cross-sectional lateral offset and elevation as the horizontal axis. For the extraction and generation calculation of conventional cross-sectional profile geometric coordinates, those skilled in the art can call the development interface of conventional computer-aided design software for parsing and reading. The coordinate parsing method is a well-known technology in this field and will not be elaborated here.

[0029] S102, Configure the surrounding rock classification dictionary in the parameter dictionary mapping matrix. After completing the spatial coordinate database construction, the system establishes a parameter dictionary mapping matrix in the storage medium. This matrix consists of a surrounding rock classification dictionary and an excavation process dictionary. In the surrounding rock classification dictionary, the system establishes a mapping relationship between mileage space and benchmark geological attributes. To ensure the uniqueness of spatial parameter addressing and avoid multi-value query conflicts at the boundaries of adjacent geological sections, the system sets the mileage range of each geological section to a closed-before-open interval and binds it to the surrounding rock grade label obtained from exploration. The system retrieves preset theoretical calculation parameters for each surrounding rock grade label, including the benchmark rock strength coefficient and the benchmark rock mass structure influence coefficient. The benchmark rock strength coefficient is calculated based on the uniaxial compressive strength of the rock obtained from previous geological drilling, with a conventional value range of 1 to 20; the benchmark rock mass structure influence coefficient is determined based on the statistical data of joint and fracture development in the field, with a value range of 0.1 to 1.0. The data initialization module encapsulates the above mapping relationship and parameters into a structured data table for subsequent module addressing and calling. The mapping relationship model of the surrounding rock classification dictionary established by the data initialization module is as follows: ; In the formula, Represents the set of associated mappings for the surrounding rock classification dictionary; Indicates the geological section number; and They represent the first The starting and ending mileage of each geological section. This indicates the surrounding rock grade label associated with this section; This indicates the benchmark rock firmness coefficient corresponding to this surrounding rock grade; This represents the baseline rock mass structure influence coefficient corresponding to this surrounding rock grade.

[0030] S103, the excavation process dictionary in the configuration parameter dictionary mapping matrix. Besides geological spatial information, the alignment of process parameters is also crucial for achieving multi-source data fusion. The system establishes a parameter binding relationship between borehole type and detonation sequence in the excavation process dictionary. Specifically, the system defines a set of borehole categories applicable to the current blasting process, classifying boreholes into slotting holes, auxiliary holes, peripheral holes, and floor holes according to spatial arrangement and detonation function based on design intent. By reading the design scheme of the on-site micro-delay blasting network, the system assigns a digital detonator micro-delay detonation time to each type of borehole in the above borehole category set, generating four detonation time constants corresponding to slotting holes, auxiliary holes, peripheral holes, and floor holes respectively, according to the detonation sequence, and then combining them into a detonation time sequence vector.

[0031] Simultaneously, the system incorporates the site reference vibration wave attenuation constant into the excavation process dictionary. The site reference vibration wave attenuation constant is determined based on regression analysis of historical test blast data, typically ranging from 10 to 100. Its technical purpose is to serve as a preset parameter in subsequent time-series signal processing to guide the envelope stripping calculation of residual stress wave energy attenuation over time. Regarding the specific criteria for setting the delay time of the differential detonator, those skilled in the art can select the corresponding millisecond step difference according to engineering design specifications. The rules for selecting this step difference time are well-known in the field and will not be elaborated upon here.

[0032] Based on the completed underlying data foundation, the system activates the parameter interpolation module to process the spatial data continuously. In this embodiment, real engineering geological environments typically exhibit a continuous and gradual extension of attributes, while traditional geological exploration data often uses discrete, fixed segment divisions. To eliminate the parameter abrupt change error caused by this discrete calibration at the segment interface, the parameter interpolation module constructs a vertical evolution mechanism for geological parameters. Its general technical principle lies in smoothly transforming discrete static dictionary data into dynamic, continuous data that conforms to the gradual change law of actual rock masses through the coupling determination of spatial distance and differences in geological attributes. The specific operating logic of the parameter interpolation module includes the following steps: S201: Receive the target design mileage station input and define the surrounding rock transition buffer zone. The parameter interpolation module obtains the target design mileage station for the upcoming blasting operation. The system iterates through the surrounding rock classification dictionary established by the data initialization module, comparing the geometric positional relationship between the target design mileage station and the starting and ending mileage stations of each geological section. The parameter interpolation module extracts the coordinates of the nearest geological boundary point to the target design mileage station and calculates the absolute distance. The system has a built-in preset transition buffer zone length threshold, which is set based on the detection accuracy of advanced geological prediction equipment and the empirical length of rock mass transition zones, typically ranging from 5 to 30 meters. To avoid misjudgments caused by a single spatial distance indicator, the system introduces surrounding rock attribute differences as a second judgment dimension. Specifically, the system synchronously reads the surrounding rock grade labels on both sides of the geological boundary point. When the absolute distance is less than or equal to the preset transition buffer zone length threshold, and the surrounding rock grade labels on both sides of the geological boundary point are inconsistent, the target design mileage station is determined to be within the surrounding rock transition buffer zone, thus triggering a smooth interpolation operation command. Conversely, when the absolute distance is greater than the preset transition buffer length threshold, or when the surrounding rock grade labels on both sides of the geological boundary point are consistent, the system determines that the target design mileage station is not within the surrounding rock transition buffer. The system then directly uses the benchmark parameters of the current section as dynamic benchmark parameters, thereby saving computational resources and avoiding meaningless parameter disturbances.

[0033] S202, Extracting Low-Level Parameters from Adjacent Survey Nodes. Under the condition that the system determines the target design mileage station is within the surrounding rock transition buffer zone, the parameter interpolation module performs cross-segment memory addressing. The system uses the nearest geological boundary point as the center and extends spatially to both sides to extract the position coordinates and bound parameters of two adjacent survey nodes. As a preferred approach, the parameter interpolation module designates the node located in front of the boundary point as the first reference node and extracts the corresponding first benchmark parameters; the node located behind the boundary point is designated as the second reference node, and the corresponding second benchmark parameters are extracted. The extracted benchmark parameter types cover the benchmark rock firmness coefficient and benchmark rock mass structure influence coefficient configured in the aforementioned dictionary. This cross-segment extraction logic ensures that the algorithm can simultaneously capture the constitutive features on both sides of the geological abrupt change interface, providing a causally related data source for subsequent weighted calculations.

[0034] S203, performs spatial distance-weighted interpolation to generate dynamic reference parameters. After obtaining the parameters of the nodes on both sides, the parameter interpolation module constructs a first-order continuous spatial distance-weighted interpolation function. This function uses distance as a weight decay factor; the closer the node is to the reference node, the greater the weight contribution of that node's parameters to the final result. Its technical purpose is to transform discrete static dictionary data into dynamic continuous data that conforms to the actual geological gradation patterns. The specific interpolation calculation equations executed by the parameter interpolation module are as follows: ; In the formula, This represents the dynamic benchmark parameters output after calculation at the target design mileage station. This indicates the specific location coordinates of the target design mileage station; This indicates the position coordinates of the first reference node; This represents the static reference parameters corresponding to the first reference node; This indicates the position coordinates of the second reference node; This represents the static reference parameter corresponding to the second reference node. Based on this general interpolation equation, the system substitutes the rock firmness coefficient and the rock mass structure influence coefficient into the parameter system to calculate the dynamic rock firmness coefficient and the dynamic reference rock mass structure influence coefficient specific to the target section.

[0035] To ensure the computational completeness of the algorithm under extreme boundary conditions and prevent program exceptions, the parameter interpolation module incorporates denominator verification logic before executing the above equations. The system detects the absolute value of the difference between the position coordinates of the first and second reference nodes. This constant is typically set based on the limit resolution of the surveying instrument, and its value range can be set to 0.001 to 0.01 meters. When the absolute value of the position coordinate difference is greater than or equal to the preset minimum constant for machine precision, a first-order continuous spatial distance weighted interpolation function with distance as the weight attenuation factor is constructed for calculation. Conversely, when the absolute value of the position coordinate difference is less than the preset minimum constant for machine precision, it means that the two reference nodes are highly overlapping in space. The system directly skips calling the first-order continuous spatial distance weighted interpolation function and extracts the arithmetic mean of the first and second reference parameters as the dynamic reference parameter. Through the above processing, the system completes the smooth evolution of the geological parameters in the vertical spatial dimension and outputs the generated dynamic reference parameter to the calculation node corresponding to the next step.

[0036] In this embodiment, traditional blasting design typically uses a single parameter to control the charge distribution across the entire cross-section. To adapt to the asymmetric stress characteristics of complex geological environments, the charge calculation module introduces spatial heterogeneous computation rules. Its general technical principle lies in breaking the limitations of traditional homogeneous calculation across the entire cross-section. Considering the drastically different clamping effects and free-face conditions faced by the excavation, auxiliary, and surrounding areas on the same tunnel cross-section, this module extracts the spatial attenuation error from previous blasting cycles and performs multi-dimensional, differentiated compensation and reconstruction of the current cross-section's baseline parameters. The specific operational logic of the charge calculation module includes the following steps: S301, retrieve and parse the preset asymmetric spatial compensation matrix. The charge calculation module executes the data retrieval operation of the preset asymmetric spatial compensation matrix. To ensure the continuity and causal correlation of multi-source data in spatial conditions, the system reads the preset asymmetric spatial compensation matrix recorded in the storage medium, updated by feedback from the actual blasting cycle of the adjacent previous tunneling advance. This matrix contains spatial compensation terms corresponding to cut holes, auxiliary holes, peripheral holes, and bottom plate holes, respectively, which are used to characterize the asymmetric resistance characteristics generated by the difference in free face conditions at different spatial locations within the same blasting section. Considering the special characteristics of the initial stage of tunnel engineering, the system has built-in matrix initialization state judgment logic. When the system identifies the current target design mileage station as the first blasting section of the project, due to the lack of historical prior data support, the system forcibly initializes all spatial compensation terms in the preset asymmetric spatial compensation matrix to zero values. Conversely, when the system identifies the current target design mileage station as not the first blasting section of the project, the system normally reads and retains the preset asymmetric spatial compensation matrix recorded in the storage medium, updated by feedback from the actual blasting cycle of the adjacent previous tunneling advance. This processing mechanism ensures the computational completeness and data consistency of the algorithm process during the initial stages of engineering.

[0037] S302, The system integrates and generates borehole-specific heterogeneous influence coefficients and performs safety checks. After successfully parsing the data structure of the compensation matrix, the charge calculation module integrates the spatial compensation terms in the matrix with the dynamic reference parameters. The system uses the dynamic reference rock mass structure influence coefficient in the dynamic reference parameters as a unified calculation basis, and superimposes the spatial compensation terms corresponding to various types of boreholes in the preset asymmetric spatial compensation matrix to obtain the heterogeneous structure influence coefficients corresponding to various types of boreholes. The technical objective is to achieve asymmetric differences in rock mass structure parameters at different spatial locations within the same cross-section by superimposing error corrections caused by historical blasting feedback. The fusion calculation equations executed by the charge calculation module are as follows: ; In the formula, This indicates the current target design mileage station number. The influence coefficient of heterogeneous structure corresponding to the type of borehole; This represents the dynamic benchmark rock mass structure influence coefficient at the target design mileage station input from the preceding module. Represents the compensation matrix in asymmetric space for the first... Spatial compensation terms corresponding to borehole type; As a subscript for borehole category index, it is used to refer to slotted holes, auxiliary holes, peripheral holes, or bottom plate holes.

[0038] As a preferred approach, to prevent excessive historical feedback errors from causing the fusion results to deviate from a reasonable range, the system introduces a threshold truncation mechanism after the superposition operation. The system verifies whether the influence coefficients of each heterogeneous structure are within a preset safe range. Based on the rock mass fracture limit theory and field blasting experience, this preset safe range is set to be greater than a set lower safety threshold (usually 0.1, to prevent the coefficient from approaching zero and causing energy failure) and less than or equal to a specific upper limit (usually 1.5, to prevent excessive charge from inducing severe vibration). The aforementioned lower safety threshold and specific upper limit are the boundary values ​​of the preset safe range. When the influence coefficient of the heterogeneous structure exceeds the preset safe range, it is forcibly assigned the value of the preset safe range boundary value (specifically: when it is below the lower safety threshold, it is forcibly assigned the lower safety threshold value; if it exceeds the upper limit value, it is forcibly truncated to the specific upper limit value), thereby avoiding the mathematical singularity of negative structural coefficients or zero unit consumption due to overcompensation. Conversely, when it is determined that the influence coefficient of the heterogeneous structure is within the preset safe range, the system retains the current value of the calculated influence coefficient of the heterogeneous structure unchanged and directly enters the actual charge calculation stage.

[0039] S303, combining multi-parameter coupling to deduce the actual charge amount for various types of boreholes. After generating the heterogeneous influence coefficient, the charge amount calculation module, combined with the rock firmness coefficient and explosive property parameters, calculates the explosive consumption per unit area and the total charge amount for the group of boreholes corresponding to various types of boreholes. The system constructs an explosive consumption per unit area evolution model based on multi-parameter coupling. This model breaks through the limitations of the conventional single lookup table method, comprehensively considering the reduction and amplification effects of the borehole depth influence coefficient, charge density influence coefficient, cartridge diameter influence coefficient, and explosive energy influence coefficient on the blasting work. The system inputs the dynamic rock firmness coefficient at the target design mileage station and the independent heterogeneous structure influence coefficients of various types of boreholes into this evolution model to solve for the actual explosive consumption per unit area specific to various types of boreholes. For this solution process, the expression of the unit consumption calculation function called by the system is: ; In the formula, Indicates the first Actual explosive consumption per unit volume obtained from borehole-type calculations; This represents the unit cost calculation function for establishing a nonlinear mapping relationship; This represents the dynamic rock firmness coefficient at the target design mileage marker; Indicates the first The influence coefficient of heterogeneous structure corresponding to the type of borehole; Indicates the influence coefficient of borehole depth; Indicates the influence coefficient of charge density; Indicates the influence coefficient of the drug roll diameter; This represents the energy influence coefficient of the explosive. For the specific criteria for determining the values ​​of the aforementioned external influence coefficients and the empirical rules for algebraic transformation within the standard volume consumption function, those skilled in the art can refer to the engineering design manual and the explosive designation used on-site for matching. The basic coefficient values ​​and conventional empirical formulas for unit consumption are well-known techniques in this field and will not be elaborated upon here.

[0040] After obtaining the actual explosive consumption per borehole for each borehole type, the system performs specific explosive charge allocation calculations. Combining the load area allocated to each borehole in the design drawings with the predetermined advance depth, the system calculates the single-hole detonation volume for each borehole. Subsequently, the system multiplies the single-hole detonation volume by the actual explosive consumption per borehole type to obtain the single-hole charge amount for that borehole. Based on this, the system aggregates the single-hole charge amounts within the same differential delay detonation network segment, outputting the total charge amount for all boreholes within the corresponding delay segment. Through this in-depth deduction from cross-sectional parameters to borehole type parameters, the system distributes a parameter set containing information on the borehole layout and charge amount to the safety verification node, providing underlying data support for subsequent positive compliance assessments.

[0041] In this embodiment, because blasting operations involve violent energy release, to prevent the vibration waves generated by the actual blasting from causing structural damage to surrounding buildings, the system establishes a theoretical positive compliance judgment mechanism before formally issuing operation instructions to the site. Its general technical principle lies in predicting the peak particle vibration velocity at the target protected object in advance based on the energy propagation and attenuation law of explosive stress waves in a non-uniform medium, thereby establishing a multi-dimensional safety blocking and scheme self-checking process. The specific operating logic of the safety verification module includes the following steps: S401, extract the maximum single-stage detonation charge and three-dimensional spatial distance parameters. The safety verification module receives the total charge data of each micro-delay segment group hole from the charge calculation module. Since the micro-delay detonation network disperses the total explosive charge to different time nodes for independent detonation, engineering practice shows that the core source of the peak ground vibration is the maximum energy released by a single explosion. Based on the above causal relationship, the system executes the time-series peak scanning algorithm, traverses the charge data corresponding to all detonation time sequence vectors, and extracts the total charge of the group hole with the largest value as the maximum single-stage charge. Simultaneously, the system analyzes the spatial mileage model and external environment survey records to calculate the three-dimensional spatial straight-line distance from the center point of the current blasting face to the nearest protected structure. Through the above steps, the system completes the preparation of the source energy parameters and propagation path parameters required for theoretical verification.

[0042] S402 executes the solution of the vibration velocity theory model. After obtaining the energy and path parameters, the safety verification module calls the Sadovsky vibration velocity theory model to establish a mathematical mapping relationship between the maximum single-stage charge, the distance from the blast center to the protected structure, and the peak mass vibration velocity. The theoretical solution equations executed by the system are as follows: ; In the formula, This indicates the predicted peak particle vibration velocity at the protected building structure. Indicates the maximum amount of a single-segment charge extracted; This represents the three-dimensional straight-line distance from the blast center to the protected building or structure. This represents the vibration wave attenuation constant related to the site medium; The attenuation index represents the seismic wave attenuation index. The values ​​of the vibration wave attenuation constant and the seismic wave attenuation index, which are related to the site medium, in the above formula are significantly affected by the geological characteristics of the rock mass and the topography. Under conventional engineering conditions, the above two coefficients are obtained through nonlinear regression based on the vibration velocity data measured in previous test blasts, and their values ​​range from 50 to 300 and 1.0 to 3.0, respectively. For the specific regression algorithm for fitting the attenuation coefficient based on historical test blast data, those skilled in the art can use the conventional least squares method for numerical solution. The data fitting process is a well-known technique in this field and will not be elaborated upon here.

[0043] As a preferred approach, to prevent the division denominator in the formula from becoming invalid or the vibration velocity prediction value from diverging geometrically when the spatial distance approaches zero in scenarios such as extremely close-range blasting or cross-tunnel construction, the system incorporates distance extremum avoidance logic before executing the solution equation. The system has a built-in preset minimum safe calculation distance constant, which is determined based on the equipment's minimum resolution safety margin and the geometric dimensions of the borehole layout, with a typical value range of 5 to 15 meters. When the determined three-dimensional spatial straight-line distance is less than the preset minimum safe calculation distance constant, the spatial distance variable used for calculation is forcibly assigned to the preset minimum safe calculation distance constant, thereby ensuring the mathematical rigor and engineering feasibility of the calculation process. Conversely, when the determined three-dimensional spatial straight-line distance is greater than or equal to the preset minimum safe calculation distance constant, the system directly uses the actually calculated three-dimensional spatial straight-line distance as the spatial distance variable and substitutes it into the model for solution.

[0044] S403 executes multi-dimensional safety judgments and trigger command flows. After obtaining the theoretical predicted velocity, to avoid project delays or missed risks due to one-sided judgments relying solely on a single extreme value, the system constructs a comprehensive safety assessment logic based on interval weighting. The system retrieves a preset site safety velocity threshold from the parameter dictionary. This threshold is determined according to relevant safety regulations and the structural type of the protected building, with a typical value range of 1.0 to 3.0 cm / s. The system establishes an early warning classification judgment tree. When the theoretical peak vibration velocity is determined to be less than or equal to the preset site safety velocity threshold (the system further divides this into: predicted vibration velocities less than 80% of the threshold are considered absolutely compliant, and 80% to 100% are considered critically safe and recorded in the log), the judgment verification is passed, and the borehole layout parameters and corresponding construction guidance instructions are issued.

[0045] Conversely, when the theoretical peak vibration velocity is determined to exceed the preset site safety velocity threshold, the command is intercepted and issued, triggering the charge calculation module to reduce the reference rock mass structure influence coefficient downward by a predetermined step size and start the automatic iterative recalculation process.

[0046] In this iterative process, the predetermined step size directly affects the algorithm's convergence efficiency and the quality of the construction. Based on field experience, if the reduction step size is too large, it can easily lead to a decrease in blasting energy and under-excavation; if the reduction step size is too small, it may cause the loop to stall and increase the computational burden. Therefore, the system dynamically sets the predetermined step size to 5% to 10% of the initial input dynamic benchmark rock mass structure influence coefficient. The system implements iterative reduction based on this step size until the calculation results of the iterative output pass the compliance verification. To avoid the automatic iteration process falling into an invalid loop due to failure to meet the vibration velocity threshold under extreme working conditions, such as when extremely close to existing buildings and structures, the system incorporates a limit blocking mechanism in the iterative logic. When the reduced baseline rock mass structure influence coefficient reaches the lower safety limit (0.1) and the predicted vibration velocity still does not meet the compliance conditions, the system terminates automatic iteration, outputs an early warning command and suspends the calculation task, prompting on-site manual intervention to modify the design scheme, such as shortening the excavation advance or adjusting the blasting network layout; when the theoretical peak vibration velocity meets the compliance condition of being less than or equal to the preset site safety velocity threshold before the reduction reaches the lower safety limit, the system automatically ends the iteration process and issues updated reduced borehole layout parameters.

[0047] In this embodiment, due to the use of a differential delay blasting network on site, the surface vibration signals collected by the sensors are actually spatially superimposed stress wave signals generated by multiple blast holes detonating at different time points. To accurately evaluate the actual work done by blast holes in different areas, the system activates a time-series decoupling feature extraction algorithm. Its core technical principle lies in using the detonation time sequence vector and the site attenuation law as a basis to perform inverse envelope stripping on the superimposed waveforms, and then relying on a deep learning architecture to extract the energy evolution features hidden in the high-frequency background. The specific operating logic of the signal extraction module includes the following steps: S501 performs preprocessing and timestamp alignment of the measured vibration signals. The system receives triaxial raw vibration waveform data from field sensors via a data acquisition interface. Considering the hardware clock system barrier between the electronic detonator's detonation timing and the monitoring instrument's trigger timing, direct comparison would cause serious operational misalignment. Based on this consideration, the system extracts the rising edge of the first detonation pulse from the raw vibration waveform data and uses a cross-correlation algorithm to calculate the offset between this rising edge and the zero point of the detonation time sequence vector configured in the excavation process dictionary. After obtaining this offset, the system performs translation compensation on the time axis for the entire raw vibration waveform data. To eliminate high-frequency mechanical interference from the field excavation equipment, the system applies bandpass filtering to the translated waveform. The cutoff parameter of this filter is set according to the distribution characteristics of the main frequency band of blasting energy, and its value range is limited to between 10 Hz at the low frequency end and 150 Hz at the high frequency end.

[0048] S502 implements time-series waveform decoupling calculations based on an attenuation stripping mechanism. After acquiring the cleaned global vibration sequence, the system combines the detonation time sequence vector with site parameters to perform borehole-by-bore waveform decoupling. Using the set detonation time of subsequent boreholes as the analysis node, the system predicts the residual energy envelope of the preceding stress wave at the current moment using the peak amplitude of the preceding borehole and the site medium characteristics, and performs differential subtraction from the global sequence to separate the independent vibration response characteristics of each borehole (i.e., the purified time-series feature vector). The specific attenuation stripping calculation equations are as follows: ; In the formula, This indicates the first element extracted after the stripping operation. Type of blast hole in time Independent vibration response waveform at the location; This indicates the preprocessed global vibration sequence in time. The amplitude at that point; Indicates the first Index of historical blast holes that were detonated before this type of blast hole; Indicates the first Peak value of vibration wave induced by the preceding borehole; Indicates the first The actual detonation time of each preceding blast hole; Indicates the signal sampling time variable; This represents the attenuation constant of the reference vibration wave, reflecting the characteristics of the site medium. To ensure the physical and mathematical rigor of envelope stripping and avoid nonlinear addition anomalies caused by minor waveform phase misalignments, the system converts the global vibration sequence into a positive energy envelope signal using Hilbert transform or the sliding window root mean square method before executing the stripping model, i.e., in the above formula. and It represents the pure energy envelope amplitude, rather than the original velocity amplitude that includes alternating positive and negative phases.

[0049] As a preferred approach, considering the potential time-delay accuracy drift of the detonators in the field, if the actual detonation times of two adjacent boreholes are very close, the superposition of stress waves at the physical level will exhibit strong nonlinear resonance. In this case, the linear stripping model based on the independent attenuation assumption is no longer applicable. Therefore, the system incorporates a preset minimum time difference constant, which is determined based on the factory precision limit of the detonators used and typically ranges from two to five milliseconds. When the actual detonation time difference between two adjacent boreholes is determined to be less than the preset minimum time difference constant, an avoidance mechanism is triggered, and the single-hole stripping step is skipped. The affected boreholes are merged into a single joint detonation wave packet for overall feature extraction, thus maintaining the robustness of the algorithm under extreme differential conditions. Conversely, when the actual detonation time difference between two adjacent boreholes is determined to be greater than or equal to the preset minimum time difference constant, the system normally executes the residual waveform stripping calculation equation for each borehole to separate the independent vibration response characteristics of each borehole.

[0050] S503, a one-dimensional convolutional neural network is constructed and applied to implement deep feature mapping. After obtaining the independent vibration response waveforms of each borehole, in order to quantify the actual rock mass fracturing energy consumption within them, and considering that traditional peak extraction methods are difficult to capture hidden information over a wide frequency band, the system introduces a one-dimensional convolutional neural network for deep time-frequency feature mining. The data dimension input to this model is defined as a fixed-length one-dimensional time series tensor, which is derived from the original independent vibration response waveforms through equal-length truncation or tail-filling.

[0051] In terms of internal topology, this one-dimensional convolutional neural network consists of an input layer, three alternately cascaded one-dimensional convolutional layers and a max-pooling layer, and a fully connected output layer. During operation, the one-dimensional convolutional layers use one-dimensional convolutional kernels with a set stride to slide along the time axis, capturing the local evolution of the slope of the stress wave rise and the bandwidth of the dominant frequency. The max-pooling layer then performs a downsampling operation, further suppressing high-frequency environmental noise while compressing the feature dimension. These features are finally fed into the fully connected output layer and mapped to a univariate feature value characterizing energy transfer efficiency. The connotation of this feature value is that it refers to the proportion of explosion energy actually converted into rock mass fracturing work. After normalization, its value is constrained to a closed interval of 0.1 to 0.9.

[0052] S504 performs offline training and weight updates for the network model parameters. To endow the aforementioned neural network with generalization prediction capabilities, the system completes the training process based on historical construction records before deployment. The system is equipped with a training sample set, where the input data comes from the decoupled preprocessed one-dimensional vibration waveform tensor under similar historical rock blasting conditions; the corresponding real labels are obtained by extracting the local real excavation volume of various blast hole corresponding areas (such as the core cavity of the cut, the fracture zone of the auxiliary hole, and the surrounding formed contour) after blasting using a high-precision 3D scanner, and back-calculating the ratio with the theoretically designed blasting volume to obtain the independent and historically true energy transfer efficiency coefficient for each hole. In the error backpropagation stage, the system uses the mean square error function as the core quantification criterion to measure the deviation between the network's predicted output and the real labels. The system uses an adaptive moment estimation optimization algorithm to calculate the gradient of the loss function relative to the network weights, and iteratively updates the internal parameters of the convolution kernel with a dynamically decaying learning rate until the mean square error on the validation set falls within the preset tolerance lower bound, thereby establishing a stable nonlinear mapping relationship between the vibration waveform morphology and the underlying work efficiency.

[0053] S505 generates local correction bias based on multi-dimensional weighted fusion. The system receives energy transfer efficiency feature values ​​from the network model output for slotted holes, auxiliary holes, peripheral holes, and bottom plate holes. To avoid the prediction blind spots hidden in single-data-driven models, the system introduces the peak value of vibration waves at a purely physical level for cross-validation. The system constructs a fusion evaluation mechanism, which weights the energy transfer efficiency feature values ​​output by deep learning with the degree of deviation of vibration response intensity to obtain the local correction bias for closed-loop feedback. The specific fusion calculation equation is as follows: ; In the formula, Indicates the extraction of the generated first... Local correction deviation corresponding to the type of borehole; Indicates the network output of the first Energy transfer efficiency characteristic value of borehole-like structures; Indicates the first If the peak value of the vibration wave obtained by decoupling from the measured signal of the borehole is less than the background noise of the sensor, the system will force it to be assigned the minimum value of the background noise to avoid the denominator of the division approaching zero. This represents the empirically calibrated peak reference value in the surrounding rock environment of this section; and These represent the energy dimension weighting coefficient and the amplitude dimension weighting coefficient, respectively. The system dynamically allocates these two weighting coefficients based on the model's confidence level in the historical validation set, and their sum is always equal to one.

[0054] After completing the above calculations, the system summarizes and combines the local correction deviations for each borehole type, and overwrites them into the asymmetric spatial compensation matrix in the storage medium. This matrix will then be carried over to the charge calculation process for the next tunneling advance, achieving system-level parameter optimization and closed-loop adaptive adjustment.

[0055] In this embodiment, although the preceding module outputs the deviation based on theoretical calculations and deep learning models, in actual engineering, the heterogeneity and concealed structure of underground rock masses inevitably lead to random errors in the data from a single blasting operation. To achieve adaptive learning and anti-interference evolution of the system, the matrix update module introduces an inversion solution and closed-loop control mechanism. Its general technical principle is to abandon the traditional single-trial-and-error replacement mode. The system constructs a momentum smoothing update algorithm with historical memory effect, fusing the latest extracted measured data with historical compensation states to generate a spatial compensation matrix to guide the next tunneling cycle, thus completing the closed-loop iteration of the entire intelligent blasting design. The specific operating logic of the matrix update module includes the following steps: S601 receives multi-source extracted features and performs spatiotemporal continuity alignment. The system receives the local correction deviation of each hole type under the current blasting cycle transmitted by the signal extraction module. Considering the complex working condition of simultaneous cross-excavation of multiple working faces, the system forcibly verifies the working condition timestamp and spatial mileage coordinates before performing the underlying data overwrite. To ensure the temporal and spatial logical alignment of the feedforward data, the system reads the actual working face mileage station bound to this batch of local correction deviations and retrieves the next adjacent design mileage station to be excavated in the spatial mileage model. The system only allows the update program to start when the absolute difference between the two in space completely matches the design step size of a single advance and the order of the time tags satisfies the logical extension. Conversely, when it is determined that the absolute difference between the two in space does not match the design step size of a single advance, or the order of the time tags does not satisfy the logical extension, the system refuses to start the update program, discards the local correction deviation of the current batch, and continues to use the old version of the asymmetric spatial compensation matrix. This forward alignment and verification logic prevents data confusion and contamination between different working surfaces, ensuring that the feedforward data generated by the inversion has a real causal relationship.

[0056] S602 performs compensation parameter inversion updates based on momentum decay smoothing theory. After obtaining the verified local correction deviation, the system retrieves the old version of the asymmetric spatial compensation matrix from the previous cycle, stored in the storage medium. To prevent drastic data jumps caused by sudden geological changes or occasional sensor malfunctions, the system constructs a first-order low-pass filter smoothing evolution model. This model comprehensively considers the inertia retention weight of historical empirical parameters and the innovative absorption weight of the latest feedback, thereby iteratively calculating the compensation terms for various boreholes. The inversion update calculation equations called by the system are as follows: ; In the formula, This indicates the updated generation used in the next loop. Hole-like space compensation term; This indicates the historical application of the previous cycle. Hole-like space compensation term; This indicates the latest input of the current loop to the signal extraction module. Local correction deviation corresponding to the type of borehole; Indicates the momentum retention factor; As a subscript for borehole category index, it is used to refer to slotting holes, auxiliary holes, peripheral holes, or bottom holes. Since the properties of underground rock masses typically exhibit continuous and gradual changes over short distances, the aforementioned momentum retention coefficient directly determines the system's dependence on historical data, and its value is set between 0.6 and 0.9. For the specific dynamic value rules of this coefficient, those skilled in the art can empirically assign values ​​based on the homogeneity of the rock bedding in the field and geological prediction results. The basic mapping relationship is a well-known technique in this field and will not be elaborated upon here.

[0057] S603 implements safety constraints and avoidance of discrete divergence in updated parameters. As a preferred approach, since the above update equation is essentially a recursive expansion in the time dimension, during long-distance tunneling, if there is a continuous unidirectional systematic error, the calculated spatial compensation term may gradually accumulate until it diverges, leading to an extreme value in the charge amount of the next cycle that deviates from common sense. Therefore, the system incorporates a multi-level safety constraint mechanism. The system performs an upper limit check on the absolute value of all updated spatial compensation terms. When the absolute value of a spatial compensation term for a certain hole type is determined to be greater than a preset boundary tolerance threshold, the system immediately forces it to be truncated to that threshold. Conversely, when the absolute value of a spatial compensation term for a certain hole type is determined to be less than or equal to the preset boundary tolerance threshold, the system retains the actual value of the updated spatial compensation term. This boundary tolerance threshold is usually set to 0.5 to prevent distortion of the overall unit consumption caused by a single adjustment. Simultaneously, the system calculates the statistical variance of the parameter vector formed by the updated compensation terms for each hole type to evaluate the coordination of parameter distribution in different spatial regions. If the calculated statistical variance exceeds the preset upper limit of dispersion, the system determines that there is a risk of divergent singularity among the parameters of the multi-pore type. This upper limit of dispersion is preset based on the maximum allowable difference in rock strength on the same cross section, and the typical value range is 0.1 to 0.3. When a singularity risk is triggered, the system automatically executes a reset protocol, proportionally reducing the abnormal compensation term with the largest deviation from the mean in the matrix towards zero by a specific coefficient (usually using halving attenuation), thereby mathematically ensuring the stability of subsequent charge calculations. Conversely, if the calculated statistical variance is less than or equal to the preset upper limit of dispersion, the system determines that the parameter distribution is coordinated, does not trigger the reset protocol, and normally retains the current inversion update results.

[0058] S604 executes the comprehensive evaluation and data flow distribution of closed-loop control indicators. After completing matrix updates and safety constraints, the system needs to determine whether the blasting parameter system of the current section has converged to the optimal steady state. To avoid misjudgments caused by relying solely on a single extreme deviation, the system establishes a weighted evaluation logic based on multi-dimensional factors across the entire cross-section. The system calculates the root mean square of the rate of change between the updated asymmetric spatial compensation matrix and the old asymmetric spatial compensation matrix. To quantify the convergence degree of parameter evolution and avoid division operation crashes caused by historical terms approaching zero, the system uses the following root mean square evaluation function: ; In the formula, The root mean square rate of change represents the degree of convergence of the parameter system. This indicates the total number of borehole types; This represents a local minima smoothing factor specifically designed to prevent the denominator from being zero; it is usually fixed at a value of 10. -5 .

[0059] After obtaining the root mean square (RMS) value, the system compares it with a pre-set convergence threshold. This pre-set threshold is calculated based on the allowable error of the on-site blasting block size and the over- or under-excavation acceptance standards, and typically ranges from 2% to 5%. When the RMS rate of change is detected to be less than the pre-set convergence threshold for multiple consecutive tunneling cycles (preferably three consecutive cycles), the system automatically increases the momentum retention coefficient to enter a stable fine-tuning mode. Specifically, the system automatically increases the momentum retention coefficient in the matrix update module (S602). (For example, increasing it from the default 0.7 to 0.95) to reduce the disturbance weight of a single newly added local correction deviation on the overall matrix, thus maintaining the long-term stability of the parameter system in the steady-state phase. Conversely, when it is detected that the root mean square rate of change does not meet the condition that multiple consecutive tunneling cycles are less than the preset convergence threshold (i.e., there are cases greater than or equal to the threshold), the system maintains the current momentum retention coefficient unchanged and continues in the conventional inversion update mode. Finally, the system persistently writes the updated asymmetric spatial compensation matrix into the underlying data base and pushes it to the data flow entry of the charge calculation module. This updated matrix will serve as the compensation basis for the next heterogeneous calculation of tunneling advance, thus completing the full-element closed-loop control flow of intelligent tunnel blasting design.

[0060] Specific application examples: Project Background: A tunnel in a mountain area is currently under construction using the drill-and-blast method, with the current date being June 2026. The tunnel is about to pass through a transition zone from limestone (Class III surrounding rock) to mudstone (Class IV surrounding rock). There is a key protected structure (a surface village) approximately 45 meters vertically above the tunnel from the blasting center; the safe vibration velocity must be controlled below 2.0 cm / s.

[0061] System operation steps breakdown S10 (Data Initialization): System establishes dictionary. Class III surrounding rock section (chainage L100-L150), benchmark firmness coefficient. structural coefficient Class IV surrounding rock section (L150-L200) , The excavation process dictionary is configured with detonator differential sections (such as sections 1-3 for slotting holes, sections 5-9 for auxiliary holes, etc.).

[0062] S20 (Parameter Interpolation Calculation): Today's blasting target design chainage is L148. The system determines that this chainage is only 2 meters away from the geological boundary point (L150), which is less than the set transition buffer threshold (10 meters), triggering smooth interpolation. Calculate the dynamic benchmark parameters: closer to L150, weight biased towards Class IV surrounding rock. After distance-weighted interpolation, the target section dynamic benchmark parameters are generated: dynamic... ,dynamic (Eliminated parameter abrupt jumps).

[0063] S30 (Heterogeneous charge calculation): The system reads the asymmetric spatial compensation matrix from the previous cycle (at L145): slotting hole compensation term. Peripheral holes Generate heterogeneity influence coefficient: Cut-out holes: (Due to the large clamping force of the slotting, the system automatically compensates for the coefficient).

[0064] Peripheral holes: Substituting into the unit consumption model, the total explosive charge required for this cut hole is calculated to be 42 kg, and the total explosive charge required for the surrounding holes is 15 kg. The maximum single-stage initiation charge is extracted as follows. (In paragraph 3).

[0065] S40 (Safety Verification): Predict using the Sadovsky formula. The site attenuation constant is known. ,distance Theoretical prediction of vibration velocity: . (Safety threshold) If the system determines compliance, it will directly issue the digital borehole layout diagram and charge instructions to the on-site execution terminal.

[0066] S50 (Signal Extraction and Decoupling): After the blasting was completed, the surface vibration meter transmitted the measured waveform back. The system performed residual stripping and decoupling on the superimposed waveforms according to the micro-difference time (millisecond level). Using a 1D-CNN network to extract features, the actual fracturing energy dissipation efficiency of the cut hole was found. (Slightly lower than expected), the corresponding local correction deviation was calculated. .

[0067] S60 (Matrix Closed-Loop Update): Momentum retention coefficient is set to... Calculate the cut-out compensation term for the next cycle: The system calculates the root mean square (RMS) rate of change of the compensation matrix. If the RMS is less than 3% after three consecutive explosions, the system will increase the RMS. At a value of 0.95, the parameter system enters a steady state. The updated matrix is ​​stored in the database, awaiting retrieval tomorrow when progressing to L150.

[0068] Experimental verification and effect comparison: To verify the effectiveness of the technical solution of this invention, a comparative verification was conducted at a tunnel drilling and blasting construction site in a mountainous area for one month, accumulating 60 blasting cycles. The left tunnel used traditional empirical manual design methods, while the right tunnel deployed the intelligent calculation and processing system for multi-parameter tunnel blasting data provided by this invention. Comparative statistics were performed on peak surface particle vibration velocity, explosive consumption per unit volume, root mean square error of cross-sectional over- and under-excavation, and the time consumed in parameter design and verification.

[0069] Comparison of core evaluation indicators

[0070] See attached document Figure 3 , Figure 3 This is a comparison chart of the blasting vibration velocity control effects of different design methods in the embodiments of the present invention. Figure 3 The horizontal axis represents the number of tunneling cycles (unit: batch), and the vertical axis represents the peak ground velocity (PPR) of the surface (unit: cm / s). The figure includes a horizontal, thick black dashed line with a vertical axis value of 2.0, representing the safe vibration velocity threshold (2.0 cm / s). A light gray rectangular area is provided within the horizontal axis range of 20 to 35, representing the geological transition zone (Class III to Class IV surrounding rock). The figure includes two broken lines: one composed of a black solid line and white triangles, representing the PPR evolution trajectory of the traditional empirical manual design method; the other composed of a black solid line and dark gray circles, representing the PPR evolution trajectory of the intelligent computing system of this invention.

[0071] See attached document Figure 4 , Figure 4 This is a verification diagram of the adaptive convergence characteristics of the matrix update module in this embodiment of the invention. Figure 4 The x-axis represents the number of tunneling cycles (unit: batches), and the y-axis represents the root mean square (RMS) rate of change of the matrix (unit: %). The graph contains a series of light gray-filled bars with black borders, representing the actual matrix update RMS value for each tunneling cycle batch; a thick black line fitting the height changes of the aforementioned bars represents the momentum smoothing convergence trend; the graph also includes a horizontal black dotted line with a y-axis value of 5, representing the preset convergence threshold (5%). At the point where the momentum smoothing convergence trend (thick black line) intersects the preset convergence threshold (5%) (marked with a black pentagram), an upward vertical black dashed line indicates that steady-state fine-tuning mode is triggered at this location.

[0072] Conclusion Analysis: Combining the above table with Figure 3It can be seen that when facing complex geological environments, the traditional experience-based manual design method (broken line marked with white triangles) fails to adapt dynamically to geological changes when entering the geological transition zone (Class III to Class IV surrounding rock) (light gray rectangular area). This results in drastic fluctuations in vibration velocity, repeatedly exceeding the safe vibration velocity threshold (2.0 cm / s), consistent with the average of 2.15 cm / s (multiple exceeding alarms) in the table. Conversely, the intelligent calculation system of this invention (broken line marked with dark gray circles) achieves extremely stable vibration velocity evolution when traversing the same geological transition zone, thanks to the dynamic benchmark parameter generation mechanism of the parameter interpolation module. The velocity remains consistently below the safe vibration velocity threshold, successfully achieving the technical effect of maintaining a stable velocity within the 1.62-1.85 cm / s range and increasing the safety compliance rate to 100%, thus eliminating the potential for exceeding the vibration velocity limit.

[0073] Combining the above table with Figure 4 As can be seen, in the initial stage of operation, the actual matrix update RMS value (bar chart) of the system is relatively high, indicating that the system is undergoing high-intensity adaptive learning and adjustment of the initial parameters. As the cyclic batches progress, the momentum smooth convergence trend (thick black solid line) rapidly decays exponentially, and quickly falls below the preset convergence threshold (5%) near the 10th cyclic batch, triggering the steady-state fine-tuning mode (vertical dashed node). This indicates that the asymmetric spatial compensation matrix inside the system has converged to the optimal parameter solution under the current geological conditions extremely quickly and stably through closed-loop iteration. It is precisely because the system autonomously finds the heterogeneous influence coefficients of each borehole that are highly matched with the actual rock mass clamping force that it avoids blind charging and energy waste, thereby achieving the economic and construction benefits of a 20% reduction in explosive consumption per unit and an optimization of the cross-sectional over- and under-excavation mean square error from 18.5cm to 7.2cm, and a 61% improvement in excavation quality. Overall, this invention verifies its excellent performance in terms of computational accuracy, convergence efficiency, and adaptive capability.

Claims

1. An intelligent calculation and processing system for multi-parameter data in tunnel blasting, characterized in that, include: The data initialization module is used to establish the spatial mileage model and baseline parameters. The parameter interpolation module is used to obtain the corresponding dynamic reference parameters in the spatial mileage model based on the received target design mileage station number. The charge calculation module is used to retrieve a preset asymmetric space compensation matrix and fuse the dynamic reference parameters with the asymmetric space compensation matrix to calculate the actual charge amount. The safety verification module is used to perform safety verification based on the actual charge amount, and outputs the borehole layout parameters after determining that the verification is successful. The signal extraction module is used to acquire the actual continuous vibration waveform data generated by blasting based on the borehole layout parameters, and to analyze the actual continuous vibration waveform data to obtain the local correction deviation. The matrix update module is used to update the asymmetric spatial compensation matrix based on the local correction deviation, for use in the next blasting loop.

2. The intelligent calculation and processing system for multi-parameter data in tunnel blasting according to claim 1, characterized in that, The benchmark parameters include the benchmark rock firmness coefficient and the benchmark rock mass structure influence coefficient, which are tied to the geological section of the tunnel. The data initialization module is also used to establish a detonation time sequence vector that associates borehole type with detonation timing.

3. The intelligent calculation and processing system for multi-parameter data in tunnel blasting according to claim 1, characterized in that, Obtain the corresponding dynamic baseline parameters, specifically including: When the distance between the target design mileage station and the geological boundary point in the spatial mileage model is less than or equal to the preset transition buffer length threshold associated with the detection accuracy of advanced geological prediction equipment, and the surrounding rock grade labels on both sides of the geological boundary point are inconsistent, the benchmark parameters on both sides of the geological boundary point are extracted and spatial distance weighted interpolation is performed to obtain the dynamic benchmark parameters. When the distance between the target design mileage station and the geological boundary point is greater than the preset transition buffer length threshold, the benchmark parameters of the current section are directly used as the dynamic benchmark parameters.

4. The intelligent calculation and processing system for multi-parameter data in tunnel blasting according to claim 3, characterized in that, The spatial distance weighted interpolation operation is performed by using spatial distance as a weight attenuation factor to perform first-order continuous spatial distance weighted interpolation calculation on the reference parameters on both sides of the geological boundary point, thereby generating the dynamic reference parameters with a smooth transition.

5. The intelligent calculation and processing system for multi-parameter data in tunnel blasting according to claim 1, characterized in that, The asymmetric spatial compensation matrix is ​​an asymmetric spatial compensation matrix that includes spatial compensation terms corresponding to various types of boreholes. The calculation of the actual charge amount specifically includes: superimposing the dynamic reference parameters with the corresponding spatial compensation terms in the asymmetric spatial compensation matrix to obtain the heterogeneous structure influence coefficient, and inputting the heterogeneous structure influence coefficient into the explosive unit consumption evolution model based on multi-parameter coupling to calculate the actual charge amount corresponding to various types of boreholes.

6. The intelligent calculation and processing system for multi-parameter data in tunnel blasting according to claim 1, characterized in that, Safety verification is performed based on the actual charge quantity, specifically including: The maximum single-segment charge in the actual charge is extracted and substituted into the Sadovsky vibration velocity theory model to predict the theoretical peak vibration velocity. When the theoretical peak vibration velocity exceeds the preset site safety velocity threshold corresponding to the structure type of the protected building, the output command is intercepted, and the charge calculation module is triggered to recalculate the actual charge after reducing the reference rock mass structure influence coefficient by a predetermined step size.

7. The intelligent calculation and processing system for multi-parameter data in tunnel blasting according to claim 1, characterized in that, Before analyzing the actual continuous vibration waveform data to obtain the local correction deviation, the following steps are also included: Using the set detonation time of the subsequent borehole as a node, the residual energy envelope is predicted by utilizing the vibration response characteristics of the preceding borehole, and residual waveform stripping calculation is performed from the superimposed actual continuous vibration waveform data to separate and obtain the independent purified time-series feature vector of each borehole.

8. The intelligent calculation and processing system for multi-parameter data in tunnel blasting according to claim 7, characterized in that, The actual vibration continuous waveform data is analyzed to obtain the local correction deviation, specifically including: The independent purification time-series feature vectors are input into a one-dimensional convolutional neural network for feature extraction, and feature values ​​characterizing energy transfer efficiency are output. The feature values ​​are weighted and solved with the degree of deviation of vibration response intensity to calculate the local correction deviation.

9. The intelligent calculation and processing system for multi-parameter data in tunnel blasting according to claim 1, characterized in that, Updating the asymmetric space compensation matrix based on the local correction deviation specifically includes: Based on the momentum retention coefficient, a smooth iterative inversion is performed on the historical space compensation term and the local correction deviation to generate a new asymmetric space compensation matrix. When it is determined that the root mean square rate of change of the asymmetric spatial compensation matrix is ​​less than the preset convergence threshold calculated based on the blasting tolerance error for multiple consecutive tunneling cycles, the momentum retention coefficient is automatically increased to switch to steady-state fine-tuning mode. When it is determined that the root mean square rate of change of the asymmetric spatial compensation matrix does not meet the condition that multiple consecutive tunneling cycles are less than the convergence threshold, the current momentum retention coefficient is maintained unchanged.

10. A method for intelligent calculation and processing of multi-parameter data in tunnel blasting, applied to the intelligent calculation and processing system for multi-parameter data in tunnel blasting as described in any one of claims 1-9, characterized in that, Includes the following steps: Establish spatial mileage model and benchmark parameters; Based on the received target design mileage station number, obtain the corresponding dynamic benchmark parameters in the spatial mileage model; The preset asymmetric space compensation matrix is ​​retrieved, and the dynamic reference parameters are fused with the asymmetric space compensation matrix to calculate the actual charge amount. A safety calculation is performed based on the actual charge amount, and the borehole layout parameters are output after the calculation is deemed to be passed. Obtain the actual continuous vibration waveform data generated by blasting based on the borehole layout parameters, and analyze the actual continuous vibration waveform data to obtain the local correction deviation. The asymmetric spatial compensation matrix is ​​updated based on the local correction deviation and then called in the next blasting loop.

Images