A method for constructing a multi-factor coupling blasting vibration attenuation prediction model for open-pit molybdenum mines

Abstract

The application provides a kind of open-pit molybdenum ore multi-factor coupling blasting vibration attenuation prediction model construction method, it is related to the field of mine blasting engineering and blasting vibration prediction technology.The method is laid out monitoring point position to collect blasting vibration measured data and record blasting parameters;The measured data is preprocessed and the vibration characteristic parameters are extracted, and the blasting vibration sample database is constructed by correlation;The correction factors that have a significant influence on vibration attenuation are determined by multidimensional correlation analysis;Based on the attenuation relationship between charge weight and propagation distance, introduce the correction coefficient corresponding to the geological lithology, topographic conditions and charge structure, and construct a multi-factor coupling blasting vibration attenuation prediction model;Nonlinear optimization algorithm is used to calibrate model parameters, and subarea verification and iterative optimization are carried out according to engineering geology zoning, to obtain a prediction model suitable for each subarea.The application considers both physical interpretability and multi-factor coupling accuracy, and can improve the accuracy of blasting vibration prediction under complex geological conditions.

Description

Technical Field

[0001] This invention relates to the field of mining blasting engineering and blasting vibration prediction technology, and more specifically, to a method for constructing a multi-factor coupled blasting vibration attenuation prediction model for open-pit molybdenum mines. Background Technology

[0002] In open-pit mining, deep-hole blasting is the primary method for ore stripping and mining operations. With the continuous expansion of open-pit mining scale and the sustained increase in production intensity, the amount of explosive charge per blast is increasing, and the blasting energy is released in a concentrated manner. The impact of blasting vibration on the stability of stope slopes, surrounding buildings, and mining infrastructure is becoming increasingly prominent. Accurately predicting the intensity of blasting vibration is a crucial prerequisite for achieving precise control of blasting vibration and ensuring safe mine production. Currently, the Sadovsky empirical formula is widely used in engineering to predict the peak velocity of blasting vibration particles. This formula expresses the vibration velocity as a power function of the ratio of explosive charge to propagation distance. It is simple in form, has few parameters, and is applicable to scenarios with relatively uniform geological conditions.

[0003] However, large open-pit molybdenum mines often exhibit complex geological conditions, highly heterogeneous rock masses, significant topographic relief, and the presence of goaf areas left over from historical underground mining. The propagation and attenuation of blasting vibrations within the rock mass are influenced by a combination of factors, including geological lithology, topographic elevation, and charge structure. Traditional Sadovsky empirical formulas only consider charge quantity and propagation distance, failing to adequately reflect the impact of these localized factors on vibration attenuation. Directly applying general empirical parameters often leads to significant prediction deviations, making it difficult to meet the accuracy requirements under complex operating conditions. Existing research has seen some scholars attempt to directly establish a mapping relationship between multiple inputs and vibration velocities using data-driven methods such as neural networks. However, these methods suffer from poor interpretability, require large training samples, and have limited extrapolation capabilities. Furthermore, the model parameters lack clear physical meaning, hindering understanding and application by engineering technicians. Other studies have made simple modifications to Sadovsky's empirical formula, but these are mostly single-factor local adjustments. They lack a complete methodology that systematically couples multiple factors such as geological lithology, topographic conditions, and charge structure into the attenuation model, and have not developed parameter calibration and zone adaptation strategies for different engineering geological zones in open-pit molybdenum mines. Therefore, we propose a method for constructing a multi-factor coupled blasting vibration attenuation prediction model for open-pit molybdenum mines. Summary of the Invention

[0004] The present invention provides the following technical solution: a method for constructing a multi-factor coupled blasting vibration attenuation prediction model for open-pit molybdenum mines, comprising the following steps: Step 1: Set up monitoring points covering sensitive areas and different distance gradients in the open-pit molybdenum mine and surrounding areas, collect measured blasting vibration data under different blasting conditions, and record the blasting parameters corresponding to each blast. Step 2: Preprocess the measured blasting vibration data and extract vibration characteristic parameters. Then, associate and match the vibration characteristic parameters with the blasting parameters, propagation conditions, and geological properties to construct a blasting vibration sample database. Step 3: Perform multi-dimensional correlation analysis on the peak vibration velocity of each influencing factor and the particle in the blasting vibration sample database to determine the correction factors that have a significant impact on the attenuation of blasting vibration. Step 4: Based on the Sadovsky attenuation relationship, the basic model uses the charge amount and propagation distance as the dominant variables. The correction coefficients corresponding to the correction factors are introduced into the basic model to construct a multi-factor coupled blasting vibration attenuation prediction model. Step 5: Using the measured data in the blasting vibration sample database as the training set, a nonlinear optimization algorithm is used to jointly calibrate the undetermined parameters in the multi-factor coupled blasting vibration attenuation prediction model. Step 6: Verify the multi-factor coupled blasting vibration attenuation prediction model after joint calibration according to engineering geological zones. Iterate and optimize the zones with insufficient verification accuracy to obtain blasting vibration attenuation prediction models applicable to each engineering geological zone.

[0005] As a preferred technical solution of the present invention, the correction factors include geological lithology, topographic conditions, and charge structure; the correction coefficients include a geological lithology correction coefficient corresponding to the geological lithology, a topographic conditions correction coefficient corresponding to the topographic conditions, and a charge structure correction coefficient corresponding to the charge structure.

[0006] As a preferred technical solution of the present invention, the lithology is divided into multiple grades according to the integrity coefficient of the rock mass in the mining area, the uniaxial compressive strength of the rock, and the degree of weathering. Each lithology grade corresponds to a geological lithology correction coefficient value. The harder and more intact the lithology, the closer the geological lithology correction coefficient is to 1. The more broken and weak the lithology, the larger the geological lithology correction coefficient is. In this way, the blasting vibration attenuation law is adapted to the local conditions of the rock mass in the mining area by means of lithology classification.

[0007] As a preferred technical solution of the present invention, the terrain condition correction coefficient uses the ratio of the elevation difference between the monitoring point and the blast source to the horizontal distance as the independent variable; when the monitoring point is located above the blast source, the terrain condition correction coefficient is greater than 1 to reflect the elevation amplification effect of the vibration wave propagating uphill; when the monitoring point is located below the blast source, the terrain condition correction coefficient is less than 1 to reflect the accelerated attenuation of the vibration wave propagating downhill.

[0008] As a preferred technical solution of the present invention, the charge structure correction coefficient is input with the decoupling coefficient and the charge density; the charge structure correction coefficient decreases as the decoupling coefficient increases and increases as the charge density increases, so as to reflect the influence of different charge structures on the blasting energy release intensity and vibration attenuation law.

[0009] As a preferred technical solution of the present invention, the multi-dimensional correlation analysis includes: calculating the Pearson correlation coefficient between each individual influencing factor and the peak velocity of the mass point, and identifying the degree of linear correlation between each influencing factor and the peak velocity of the mass point; using grey relational analysis to calculate and rank the grey relational degree between each influencing factor and the peak velocity of the mass point; under the condition of controlling the charge amount and propagation distance, using partial correlation analysis to identify the independent contribution of geological lithology, topographic conditions and charge structure to the peak velocity of the mass point; and determining the correction factors and their priorities to be included in the multi-factor coupled blasting vibration attenuation prediction model based on the analysis results.

[0010] As a preferred technical solution of the present invention, the joint calibration takes the minimization of the root mean square error between the predicted peak vibration velocity of the mass point and the measured peak vibration velocity of the mass point as the objective function. The parameters to be determined include the attenuation coefficient, the attenuation exponent, the terrain influence coefficient, the charge structure coefficient, and the correction coefficient. Value boundary constraints are set for each parameter to be determined, and iterative optimization is performed until the objective function converges or the maximum number of iterations is reached.

[0011] As a preferred technical solution of the present invention, the nonlinear optimization algorithm is a particle swarm optimization algorithm or a Levenberg-Marquardt nonlinear least squares algorithm.

[0012] As a preferred technical solution of the present invention, the blasting vibration sample database is divided into a training set and a validation set according to engineering geological zones, wherein the training set accounts for no less than 70% and the validation set accounts for no more than 30%; the coefficient of determination, mean absolute percentage error, and root mean square error are used as evaluation indicators for zone validation; when the coefficient of determination of a certain zone is lower than 0.85 or the mean absolute percentage error is higher than 20%, at least one of the following measures is taken for iterative optimization: increasing the monitoring sample size of the zone, adjusting the correction coefficient value of the zone individually, or re-executing parameter calibration, until the validation accuracy of each zone meets the preset requirements.

[0013] As a preferred technical solution of the present invention, the blasting vibration attenuation prediction model after partition verification and optimization is encapsulated into a callable calculation interface. The calculation interface receives the charge amount, propagation distance, lithology grade, elevation difference and charge structure parameters of the blasting scheme, outputs the predicted peak vibration velocity of the mass point, and compares the predicted peak vibration velocity of the mass point with the preset vibration safety control threshold. When the predicted peak vibration velocity of the mass point exceeds the vibration safety control threshold, an early warning prompt is issued.

[0014] Compared with the prior art, the beneficial effects of the present invention are as follows: 1. By using the Sadovsky attenuation relation as the basic model and introducing geological lithology correction coefficients, terrain condition correction coefficients, and charge structure correction coefficients, the traditional two-factor model that only considers charge quantity and distance is extended into a multi-factor coupled model. This model can more accurately reflect the attenuation law of blasting vibration under complex geological conditions in open-pit molybdenum mines, which is conducive to improving prediction accuracy and solves the problem that the traditional empirical formula in the existing technology has insufficient prediction accuracy under complex working conditions. 2. By employing correlation analysis to screen correction factors in advance and nonlinear optimization algorithms to jointly calibrate model parameters, the model parameter calibration process is made systematic and traceable, avoiding the subjectivity of manual experience-based parameter tuning and solving the problem of existing technologies where model parameters rely on manual experience and lack systematic calibration methods; 3. Through partitioned verification and iterative optimization, the model can adapt to the differences in different engineering geological regions of the mining area, which is conducive to improving the local adaptation accuracy of the model in each partition and solving the problem that a single model in the existing technology cannot take into account the differences in different geological partitions. 4. By constructing correction coefficients with clear physical meaning and encapsulating the model into a callable calculation interface, it is easy for engineering technicians to understand and apply, and can directly serve the design of blasting schemes and vibration risk assessment. This solves the problems of poor interpretability and insufficient engineering practicality of data-driven methods in the existing technology. Attached Figure Description

[0015] Figure 1 This is a schematic diagram of the model construction method provided by the present invention; Figure 2 This is a schematic diagram of the correction factors and correction coefficients provided by the present invention. Detailed Implementation

[0016] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are specific implementations of the present invention and are not limited to all embodiments.

[0017] Therefore, the following detailed description of embodiments of the present invention is not intended to limit the scope of the claimed invention, but merely illustrates some embodiments of the invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without inventive effort are within the scope of protection of the present invention.

[0018] It should be noted that, in the absence of conflict, the embodiments and features and technical solutions in the embodiments of the present invention can be combined with each other. It should be noted that similar reference numerals and letters in the following figures indicate similar items. Therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures.

[0019] Example 1: A method for constructing a multi-factor coupled blasting vibration attenuation prediction model in an open-pit molybdenum mine, comprising the following steps: Step S1, monitoring point layout and data acquisition. First, a site reconnaissance is conducted. Based on the distribution of buildings around the mining area, their distance from the blasting area, and their sensitivity, the monitoring area is divided into five functional zones: the southern permanent slope zone, the southwestern residential area, the western road and retaining wall zone, the northwestern slope and platform zone, and the northern industrial facility zone. Within each functional zone, monitoring points are deployed according to three distance gradients: near zone (within 50m of the blast source), middle zone (50m to 150m), and far zone (above 150m), for a total of 12 monitoring points throughout the mine. Each monitoring point is equipped with a three-directional velocity detector to collect vibration velocity signals in the horizontal radial, horizontal tangential, and vertical directions, with a sampling frequency of not less than 1000Hz. The monitoring equipment uses a combination of solar power and on-site power connection. The data acquisition instrument integrates a wireless transmission module, uploading vibration data to a cloud data platform in real time via a 4G network. Before each blasting operation, record the maximum single-stage charge, total charge, hole spacing, row spacing, hole diameter, plugging length, charge structure (coupled or uncoupled charge and the uncoupling coefficient), detonation sequence, and delay time. After blasting, automatically trigger the acquisition and storage of complete vibration waveforms, and extract the peak particle vibration velocity and dominant frequency of each monitoring point. Continuously collect monitoring data covering different blasting conditions, different blasting zone locations, and different seasons, accumulating vibration records for no less than 50 valid blasting events.

[0020] Step S2: Data Preprocessing and Sample Database Construction. The collected raw vibration data undergoes quality control and preprocessing. First, the three-standard-deviation criterion is used to identify outlier data. For vibration velocity sequences in the same direction at the same monitoring point, the mean and standard deviation are calculated. Data points exceeding the mean plus or minus three standard deviations are marked as outliers and removed. Outliers mainly originate from equipment false alarms, environmental interference, or recording anomalies. For a small number of missing blasting parameter records, the mean of parameters from adjacent similar blasting events is used to complete the data, which is then marked as completed data in the database. Vibration characteristic parameters are extracted from each valid vibration waveform, including: peak particle vibration velocity and resultant velocity in three directions, dominant vibration frequency (the frequency corresponding to the main peak of the power spectrum obtained through Fourier transform), vibration duration (the time interval from when the vibration velocity first exceeds the threshold to when it finally falls below the threshold), and vibration energy (the integral of the square of the vibration velocity over time). Data association and matching are performed using the blasting event number as the primary key. Vibration characteristic parameters of each monitoring point under the same blasting event are matched with the parameter records of that blast. Spatial attribute data for each monitoring point are also supplemented, including: the horizontal distance from the monitoring point to the blast source center, the elevation difference between the blast source and the monitoring point, and the lithology of the area where the monitoring point is located (classified into grades I to V based on rock mass integrity coefficient, uniaxial compressive strength, and weathering degree). This results in a blasting vibration sample database containing four categories of fields: blasting parameters, propagation conditions, geological attributes, and vibration response. Each sample record corresponds to a combination of a blasting event and a monitoring point.

[0021] Step S3: Multi-factor Correlation Analysis. A multi-dimensional correlation analysis was conducted on the influencing factors and peak particle vibration velocity in the sample database. First, the Pearson correlation coefficient between each individual factor and vibration velocity was calculated to preliminarily identify the degree of linear correlation: the ratio of the cube root of the charge amount to the distance showed a significant positive correlation with vibration velocity, making it the dominant factor; factors such as elevation difference, lithology grade, and decoupling coefficient also showed some correlation. Further, grey relational analysis was used to handle nonlinear relationships, calculating and ranking the grey relational degree between each influencing factor and vibration velocity to determine the weight of each factor's influence on vibration attenuation. Based on this, partial correlation analysis was used to analyze the independent contributions of geological lithology, topographic elevation difference, and charge structure to vibration velocity, controlling for the two dominant variables of charge amount and propagation distance. It was confirmed that these three types of factors remained statistically significant after excluding the influence of the dominant factor; therefore, they were identified as coupling correction factors and included in the model construction.

[0022] Step S4: Construction of the basic prediction model. The Sadovsky attenuation relation is used as the basic prediction model. This basic model uses the charge amount and propagation distance as the dominant variables. The charge amount is the maximum single-segment charge amount, and the propagation distance is the horizontal distance from the monitoring point to the center of the explosion source. This retains the advantages of traditional empirical formulas, such as simple form and clear physical meaning of parameters, and provides a basic framework for subsequent multi-factor coupling expansion.

[0023] Step S5: Construction of a Multi-Factor Coupled Model. Three correction coefficients are introduced into the basic model to construct a multi-factor coupled blasting vibration attenuation prediction model. Geological Lithology Correction Coefficient: Based on the rock mass integrity coefficient, uniaxial compressive strength, and weathering degree of the mining area, lithology is divided into five levels: Level I (hard and intact) to Level V (fragmented and weak). Each level corresponds to a correction coefficient. The harder and more intact the lithology, the slower the vibration attenuates during propagation, and the closer the correction coefficient is to 1. The more fragmented and weak the lithology, the faster the vibration attenuates, and the larger the correction coefficient. This amplifies and corrects the predicted values ​​of the basic model to compensate for the deviation of the actual vibration velocity being lower than the prediction of the basic model due to faster attenuation in weak rock masses. The specific values ​​of the correction coefficients for each level are determined through subsequent parameter calibration. Terrain condition correction coefficient: The ratio of elevation difference to horizontal distance (i.e., the tangent of the slope angle) is used as the independent variable. When the monitoring point is above the blast source, the vibration wave propagates uphill with an elevation amplification effect, and the correction coefficient is greater than 1. When the monitoring point is below the blast source, the vibration wave attenuates relatively quickly downhill, and the correction coefficient is less than 1. Charge structure correction coefficient: The decoupling coefficient and charge density are used as inputs. Decoupled charges buffer the detonation effect through air gaps, which can reduce vibration intensity. Therefore, the correction coefficient decreases as the decoupling coefficient increases. Increased charge density makes energy release more concentrated, and the vibration intensity increases accordingly. The correction coefficient increases as the charge density increases.

[0024] Step S6: Joint Calibration of Model Parameters. The sample database is randomly divided into a training set (75%) and a validation set (25%) according to engineering geological zones, ensuring that both sets are distributed across each zone. Based on the training set data, a particle swarm optimization algorithm is used to jointly calibrate the model parameters. The calibration process aims to minimize the root mean square error between the predicted and measured vibration velocities. Variables to be optimized include: attenuation coefficient, attenuation exponent, topographic influence coefficient, charge structure coefficient, and correction coefficients corresponding to the five lithological grades. Reasonable boundary values ​​are set for each variable: the search range for the attenuation coefficient is 50 to 500, the search range for the attenuation exponent is 1.0 to 2.5, the search range for the topographic influence coefficient is 0 to 2.0, and the charge structure coefficient and each lithological correction coefficient are positive numbers. The particle swarm size is set to 50, and the maximum number of iterations is set to 200. The iteration terminates early when the objective function change is less than 1 x 10^-6 for 20 consecutive iterations. The optimal values ​​for each parameter are obtained after iteration convergence. As an alternative implementation method, the Levenberg-Marquardt nonlinear least squares algorithm can also be used for parameter calibration. This algorithm has a fast convergence speed when the initial parameter values ​​are reasonable and is suitable for scenarios with a large sample size.

[0025] Step S7: Zonal Validation and Iterative Optimization. The calibrated model is validated using validation set data, and three evaluation indicators are calculated: coefficient of determination (closer to 1 is better), mean absolute percentage error (lower is better), and root mean square error (lower is better). Validation indicators are calculated for each engineering geological zone. If the coefficient of determination for a zone is below 0.85 or the mean absolute percentage error is above 20%, the model for that zone is considered inaccurate, and the following iterative optimization measures are taken: First, the monitoring sample size for that zone is increased, supplementing blasting vibration data under different distance and charge combinations; second, while maintaining the overall model structure, the values ​​of the lithology correction coefficient and topography correction coefficient are allowed to be adjusted individually for that zone to adapt to its local geological characteristics; third, parameter calibration is re-executed until the validation accuracy of each zone meets the preset requirements. After zonal iterative optimization, a multi-factor coupled blasting vibration attenuation prediction model applicable to each engineering geological zone of the open-pit molybdenum mine is obtained. Compared with the traditional Sadovsky formula that only uses charge and distance, this model can improve prediction accuracy under complex geological conditions, and the prediction results are more consistent with the field measurement data.

[0026] Step S8: Model Engineering Application. The prediction model, after zonal optimization, is encapsulated into a standardized application programming interface (API). This interface supports inputting the maximum single-stage charge amount of the blasting scheme, the horizontal distance and elevation difference between the blast source and the protected target, the lithology of the protected target area, and charge structure parameters (decoupling coefficient and charge density). The output is the predicted peak particle velocity. The system automatically compares the predicted values ​​with preset vibration safety control thresholds (e.g., 2.0 cm / s for surrounding buildings, 5.0 cm / s for permanent slopes). When the predicted value exceeds the threshold, an early warning is issued, assisting blasting technicians in adjusting the maximum single-stage charge amount, optimizing the initiation network, or adjusting the charge structure during the scheme design phase. This enables pre-emptive prediction and active control of blasting vibrations.

[0027] In its application, this invention first establishes monitoring points in the open-pit molybdenum mine and surrounding areas. Three-dimensional velocity detectors continuously collect vibration data under different blasting conditions, simultaneously recording blasting parameters. The raw data undergoes quality control, preprocessing, and feature extraction. Using the blasting event number as the primary key, vibration responses are correlated with blasting parameters, propagation conditions, and geological attributes, forming a blasting vibration sample database. Next, multi-dimensional correlation analysis identifies correction factors significantly influencing vibration attenuation. Correction coefficients corresponding to geological lithology, topographic conditions, and charge structure are introduced into the basic attenuation model, which uses charge amount and propagation distance as dominant variables, constructing a multi-factor coupled blasting vibration attenuation prediction model. Using the measured data as the training set, a nonlinear optimization algorithm is employed to jointly calibrate the model parameters. The model is then validated according to engineering geological zones, with additional samples or adjusted correction coefficients for zones lacking sufficient accuracy, undergoing iterative optimization. Finally, the model is encapsulated into a callable computational interface. Inputting blasting scheme parameters outputs the predicted vibration velocity, which is compared with a preset vibration safety control threshold. When the predicted value exceeds the threshold, an early warning is issued, thus achieving pre-prediction and proactive control of blasting vibration.

[0028] Experimental Example: Verification Experiment of Prediction Accuracy of Multi-Factor Coupled Blasting Vibration Attenuation Prediction Model I. Experimental Objective This experimental example is used to verify the prediction accuracy and engineering applicability of the multi-factor coupled blasting vibration attenuation prediction model constructed in this invention under complex geological conditions in open-pit molybdenum mines. Specifically, it includes: (1) verifying the prediction accuracy advantage of the multi-factor coupled model compared with the traditional two-factor empirical formula; (2) verifying the local adaptation accuracy of the model in each zone after zoning verification and iterative optimization according to engineering geological zones; (3) verifying whether the direction of the values ​​of each correction coefficient obtained by calibration conforms to the design law and verifying the physical interpretability of the model parameters; (4) verifying the support capability of the model for blasting scheme design and vibration risk early warning after it is encapsulated as a calculation interface.

[0029] II. Test Subjects and Test Conditions The test subject is a large open-pit molybdenum mine. The mine has complex geological conditions, strong heterogeneity of rock mass, large topographic relief, and extensive distribution of goaf areas left by historical underground mining. The mine adopts bench open-pit mining and deep-hole blasting technology, with a large amount of explosives in a single blast. It has typical conditions for blasting vibration prediction test under complex geological conditions.

[0030] The test monitoring system and data acquisition conditions are as follows: 12 monitoring points are set up in and around the mining area, covering five functional zones: the southern permanent slope area, the southwestern residential area, the western road and retaining wall area, the northwestern slope and platform area, and the northern industrial facility area. Within each functional zone, monitoring points are set up at three distance gradients: near zone (within 50m of the blast source), middle zone (50m to 150m), and far zone (above 150m). Each monitoring point is equipped with a three-dimensional velocity detector to collect vibration velocity signals in the horizontal radial, horizontal tangential, and vertical directions, respectively, with a sampling frequency of not less than 1000Hz. The data acquisition instrument integrates a wireless transmission module to upload vibration data to the cloud data platform in real time via a 4G network. Vibration records of no less than 50 valid blasting events are accumulated.

[0031] Table 1 Experimental conditions and parameter settings 1 Monitoring points Twelve stations, covering five functional zones, are deployed according to three distance gradients: near, medium, and far. 2 Sensor configuration A three-dimensional velocity detector (horizontal radial, horizontal tangential, and vertical) with a sampling frequency of at least 1000Hz. 3 Data Upload 4G wireless transmission, real-time upload to cloud data platform 4 Effective blasting events No less than 50 times 5 Sample fields Blasting parameters (maximum single-stage charge, total charge, hole spacing, row spacing, plugging length, decoupling coefficient, initiation delay), propagation conditions (horizontal distance, elevation difference), geological properties (lithology grades I to V), vibration response (peak velocity, dominant frequency, duration, vibration energy). 6 Data preprocessing Three standard deviation criteria are used to remove outliers; missing data are filled in using the mean of parameters from adjacent similar events. 7 Data partitioning The dataset is randomly divided according to engineering geological zones, with the training set comprising 75% and the validation set comprising 25%. 8 Optimization Algorithm Particle Swarm Optimization Algorithm: Particle swarm size 50, maximum number of iterations 200, premature termination occurs when the objective function changes by less than 1 x 10^-6 for 20 consecutive iterations. 9 Parameter value boundaries Attenuation coefficient: 50 to 500; Attenuation index: 1.0 to 2.5; Topographic influence coefficient: 0 to 2.0; Charge structure coefficient and lithological correction coefficients are positive. 10 Accuracy Judgment Standard The coefficient of determination should be no less than 0.85 and the mean absolute percentage error should be no more than 20%; if these conditions are not met, iterative optimization measures should be taken, such as increasing the sample size, adjusting the partition correction coefficient separately, or recalibrating. 11 Vibration safety control threshold Surrounding buildings: 2.0 cm / s; Permanent slope: 5.0 cm / s 12 Monitoring system performance Data transmission latency is no more than 3 seconds; vibration velocity monitoring accuracy is ±0.01 cm / s; over-threshold alarm push latency is no more than 30 seconds. III. Experimental Data The experimental data includes three parts: model parameter calibration results, verification accuracy indicators for each engineering geological zone, and comparisons between predicted and measured values ​​of typical blasting events. The model parameter calibration results are obtained by solving the training set data according to step five (joint model parameter calibration) of this invention; the zone verification accuracy indicators are calculated by calculating the verification set data according to step six (zone verification and iterative optimization) of this invention. The experimental data acquisition and calculation methods are performed according to the conditions listed in Part Two of this experimental example.

[0032] Table 2 Model parameter calibration results Attenuation coefficient Basic model attenuation coefficient 128.6 Decay Index Basic model decay index 1.42 Topographic influence coefficient Undetermined parameters in terrain condition correction coefficients 0.86 Charge structure coefficient 1 Undetermined parameters in the charge structure correction coefficients that vary with the decoupling coefficient 0.82 Charge structure coefficient two Undetermined parameters in the charge structure correction factor that vary with charge density 1.15 Lithology correction factor (Level I) Correction factor for hard and intact lithology 1.02 Lithology correction factor (Level II) Correction factor for harder, intact lithology 1.10 Lithology correction factor (Level III) Correction factor for medium integrity lithology 1.23 Lithology correction factor (Level IV) Correction factor for relatively fractured and weak lithology 1.41 Lithology correction factor (Level V) Correction factor for fractured and weak lithology 1.65 Note: The calibration results of all parameters in Table 2 are within the parameter value boundaries set by this invention. The attenuation coefficient is in the range of 50 to 500, the attenuation index is in the range of 1.0 to 2.5, the terrain influence coefficient is in the range of 0 to 2.0, the charge structure coefficient and each lithology correction coefficient are positive numbers, and the lithology correction coefficient increases with the lithology grade from I to V.

[0033] Table 3 Comparison of Verification Accuracy for Each Engineering Geological Zone Southern permanent slope area Multi-factor coupling model 0.92 10.8% 0.21 Traditional empirical formula 0.81 19.4% 0.38 Southwest residential area Multi-factor coupling model 0.91 11.6% 0.19 Traditional empirical formula 0.76 24.6% 0.42 Western Road and Retaining Wall Area Multi-factor coupling model 0.90 12.4% 0.24 Traditional empirical formula 0.79 21.2% 0.39 Northwest slope and plateau area Multi-factor coupling model 0.89 13.5% 0.26 Traditional empirical formula 0.78 22.8% 0.41 Northern Industrial Facilities Area Multi-factor coupling model 0.93 9.6% 0.18 Traditional empirical formula 0.82 18.6% 0.35 Overall situation in the region Multi-factor coupling model 0.91 11.8% 0.22 Traditional empirical formula 0.79 21.5% 0.39 Note: The traditional empirical formula in Table 3 refers to the Sadovsky empirical formula, which uses only the charge amount and propagation distance as variables. As shown in Table 3, the multi-factor coupling model validation indicators for the five zones and the entire region all meet the judgment criteria (coefficient of determination not less than 0.85 and mean absolute percentage error not more than 20%). The traditional empirical formula did not meet the judgment criteria in the southwestern residential area (coefficient of determination 0.76, mean absolute percentage error 24.6%), the western road and retaining wall area (21.2%), and the northwestern slope and platform area (22.8%). After zone validation and iterative optimization, it converged to the acceptable range.

[0034] Table 4 Comparison of predicted and measured values ​​for typical blasting events 1 Southern permanent slope area 200 80 +8 3.42 3.71 +8.5% 2 Southwest residential area 150 120 -5 1.69 1.90 +12.4% 3 Western Road and Retaining Wall Area 100 150 +3 1.11 0.98 -11.7% 4 Northwest slope and plateau area 180 70 +12 4.31 4.65 +7.9% 5 Northern Industrial Facilities Area 300 220 -10 0.99 0.86 -13.1% 6 Southwest residential area 110 100 -2 1.77 1.96 +10.7% 7 Western Road and Retaining Wall Area 160 130 +6 1.86 1.68 -9.7% 8 Northern Industrial Facilities Area 260 180 -15 1.28 1.09 -14.8% Note: Relative error = (predicted value minus measured value) divided by measured value multiplied by 100%. A positive value indicates an overestimation, and a negative value indicates an underestimation. Table 4 shows that the sampling events cover three distance gradients: near zone (event 4), middle zone (events 1, 2, 3, 6, 7), and far zone (events 5, 8). The absolute values ​​of the relative errors range from 7.9% to 14.8%, with an average absolute error of 11.1%, which is basically consistent with the overall average absolute percentage error of 11.8% for the entire region in Table 3. Positive and negative errors each account for half, with no obvious systematic bias.

[0035] IV. Experimental Analysis The experimental analysis was carried out according to four dimensions: overall accuracy, zone adaptation, physical meaning of parameters, and error characteristics. The analysis criteria were consistent with the evaluation system set in step six of this invention.

[0036] (I) Analysis of the overall prediction accuracy of the model As shown in Table 3, the overall validation indicators for the entire region reveal that the multi-factor coupling model has an overall coefficient of determination of 0.91, a mean absolute percentage error of 11.8%, and a root mean square error of 0.22 cm / s. In contrast, the traditional two-factor empirical formula has an overall coefficient of determination of 0.79, a mean absolute percentage error of 21.5%, and a root mean square error of 0.39 cm / s on the same validation set. The multi-factor coupling model improves the coefficient of determination by 0.12, reduces the mean absolute percentage error by 9.7 percentage points, and reduces the root mean square error by 0.17 cm / s, demonstrating superior performance across all three indicators compared to the traditional empirical formula. This indicates that the introduction of three correction factors—geological lithology, topographic conditions, and charge structure—effectively improves the characterization accuracy of blasting vibration attenuation under complex geological conditions, demonstrating a significant multi-factor coupling correction effect.

[0037] (II) Analysis of the effects of partition adaptation and iterative optimization As shown in Table 3, the multi-factor coupling model achieved a determination coefficient between 0.89 and 0.93 in the five engineering geological zones, and an average absolute percentage error between 9.6% and 13.5%, both meeting the criteria. However, the traditional empirical formula failed to meet the criteria in three zones: the southwestern residential area, the western road and retaining wall area, and the northwestern slope and platform area (average absolute percentage errors of 24.6%, 21.2%, and 22.8%, respectively). For these zones with insufficient accuracy, measures were taken, including increasing the monitoring sample size, adjusting the correction coefficient for each zone individually, and recalibrating the parameters. This resulted in the determination coefficients of the three zones increasing to 0.91, 0.90, and 0.89, respectively, and the average absolute percentage errors decreasing to 11.6%, 12.4%, and 13.5%, respectively, all converging into the acceptable range. Furthermore, the accuracy differences between zones significantly narrowed. The zone validation and iterative optimization process effectively improved the model's local adaptability under complex geological conditions.

[0038] (III) Verification of the physical meaning of the correction coefficient As shown in Table 2, the lithology correction coefficient increases progressively with the rock mass, from hard and intact to broken and weak: 1.02 for Level I, 1.10 for Level II, 1.23 for Level III, 1.41 for Level IV, and 1.65 for Level V. This is consistent with the design principle that the more broken and weak the rock mass, the greater the vibration attenuation correction. Both the charge structure coefficient 1 (0.82) and the charge structure coefficient 2 (1.15) are positive. The terrain influence coefficient of 0.86 falls within the set boundary range of 0 to 2.0. The terrain correction is greater than 1 when the monitoring point is above the blast source and less than 1 when it is below the blast source, consistent with the design principle. The direction of the calibration parameter values ​​all match the design principle, indicating that the model parameters have clear physical meaning, the prediction results are interpretable, and they are easy for engineering technicians to understand and apply.

[0039] (iv) Analysis of prediction error characteristics As shown in Table 4, the absolute values ​​of the relative errors between predicted and measured values ​​ranged from 7.9% to 14.8%, with an average absolute error of 11.1%. There were 4 groups with positive relative errors (predictions were too high) and 4 groups with negative relative errors (predictions were too low), with an even distribution and no significant systematic bias. In terms of operational condition distribution, events with larger relative errors were concentrated in the far-field area (horizontal distance 180 to 220 meters), which is related to the relatively smaller sample size in the far-field area and the more complex influence of terrain and geological conditions on vibration signal attenuation. The relative errors in the near-field and mid-field areas were generally smaller. These characteristics indicate that the model prediction error is within a reasonable range and can continue to converge by supplementing the far-field monitoring samples and refining the far-field terrain correction coefficients.

[0040] V. Experimental Conclusions Based on the experimental data and analysis of this experiment, the following conclusions are drawn: Conclusion 1: The overall prediction accuracy of the multi-factor coupled model on the validation set is significantly better than that of the traditional two-factor empirical formula (the coefficient of determination increases from 0.79 to 0.91, the mean absolute percentage error decreases from 21.5% to 11.8%, and the root mean square error decreases from 0.39 cm / s to 0.22 cm / s). This verifies that under the multi-factor coupled correction of geological lithology, topographic conditions, and charge structure, the model can more accurately reflect the attenuation law of blasting vibration under complex geological conditions in open-pit molybdenum mines. Conclusion 2: After zonal verification and iterative optimization, the verification accuracy of the five engineering geological zones all met the preset judgment criteria (determination coefficient not less than 0.85 and average absolute percentage error not more than 20%). The model has the ability to adapt to different engineering geological zones and takes into account the differences in different areas of the mining area. Conclusion 3: The values ​​of the correction coefficients obtained from the calibration conform to the design rules, and the model parameters have clear physical meanings, strong interpretability, and are easy for engineering technicians to understand, verify and apply; Conclusion 4: The model can be encapsulated as a callable computing interface. By inputting blasting scheme parameters, it can output the predicted peak vibration velocity of mass points and provide early warning prompts after comparing it with the preset vibration safety control thresholds (2.0 cm / s for surrounding buildings and 5.0 cm / s for permanent slopes). It can provide quantitative support for blasting scheme design and vibration risk assessment and has engineering applicability.

[0041] Although the invention has been described in conjunction with specific features and embodiments, it is obvious that various modifications and combinations can be made therein without departing from the spirit and scope of the invention. Clearly, those skilled in the art can make various alterations and variations to the invention without departing from its spirit and scope. Therefore, if such modifications and variations fall within the scope of the claims and their equivalents, the invention is also intended to include such modifications and variations.

Claims

1. A method for constructing a multi-factor coupled blasting vibration attenuation prediction model in open-pit molybdenum mines, characterized in that, Includes the following steps: Step 1: Set up monitoring points covering sensitive areas and at different distances in the open-pit molybdenum mine and its surrounding areas, collect measured data of blasting vibration under different blasting conditions, and record the blasting parameters corresponding to each blast. Step 2: Preprocess the measured blasting vibration data and extract vibration characteristic parameters. Then, associate and match the vibration characteristic parameters with the blasting parameters, propagation conditions, and geological properties to construct a blasting vibration sample database. Step 3: Perform multi-dimensional correlation analysis on the peak vibration velocity of each influencing factor and the particle in the blasting vibration sample database to determine the correction factors that have a significant impact on the attenuation of blasting vibration. Step 4: Based on the Sadovsky attenuation relationship, the basic model uses the charge amount and propagation distance as the dominant variables. The correction coefficients corresponding to the correction factors are introduced into the basic model to construct a multi-factor coupled blasting vibration attenuation prediction model. Step 5: Using the measured data in the blasting vibration sample database as the training set, a nonlinear optimization algorithm is used to jointly calibrate the undetermined parameters in the multi-factor coupled blasting vibration attenuation prediction model. Step 6: Verify the multi-factor coupled blasting vibration attenuation prediction model according to engineering geological zones after joint calibration. Iterate and optimize the zones with insufficient verification accuracy to obtain blasting vibration attenuation prediction models applicable to each engineering geological zone.

2. The method for constructing a multi-factor coupled blasting vibration attenuation prediction model in open-pit molybdenum mines according to claim 1, characterized in that, The correction factors include geological lithology, topographic conditions, and charge structure; the correction coefficients include the geological lithology correction coefficient corresponding to the geological lithology, the topographic conditions correction coefficient corresponding to the topographic conditions, and the charge structure correction coefficient corresponding to the charge structure.

3. The method for constructing a multi-factor coupled blasting vibration attenuation prediction model in open-pit molybdenum mines according to claim 2, characterized in that, Based on the rock mass integrity coefficient, uniaxial compressive strength, and weathering degree of the mining area, the lithology is divided into multiple grades. Each lithology grade corresponds to a geological lithology correction coefficient value. The harder and more intact the lithology, the closer the geological lithology correction coefficient is to 1; the more broken and weak the lithology, the larger the geological lithology correction coefficient is.

4. The method for constructing a multi-factor coupled blasting vibration attenuation prediction model in open-pit molybdenum mines according to claim 2, characterized in that, The terrain condition correction coefficient uses the ratio of the elevation difference to the horizontal distance between the monitoring point and the blast source as the independent variable. When the monitoring point is above the blast source, the terrain condition correction coefficient is greater than 1; when the monitoring point is below the blast source, the terrain condition correction coefficient is less than 1.

5. The method for constructing a multi-factor coupled blasting vibration attenuation prediction model in open-pit molybdenum mines according to claim 2, characterized in that, The charge structure correction coefficient takes the decoupling coefficient and charge density as inputs. The charge structure correction coefficient decreases as the decoupling coefficient increases and increases as the charge density increases.

6. The method for constructing a multi-factor coupled blasting vibration attenuation prediction model in open-pit molybdenum mines according to claim 1, characterized in that, In step three, the multi-dimensional correlation analysis includes: calculating the Pearson correlation coefficient between each individual influencing factor and the peak velocity of the mass, and identifying the degree of linear correlation between each influencing factor and the peak velocity of the mass; using grey relational analysis to calculate and rank the grey relational degree between each influencing factor and the peak velocity of the mass; under the condition of controlling the charge amount and propagation distance, using partial correlation analysis to identify the independent contribution of geological lithology, topographic conditions, and charge structure to the peak velocity of the mass; and determining the correction factors and their priorities to be included in the multi-factor coupled blasting vibration attenuation prediction model based on the analysis results.

7. The method for constructing a multi-factor coupled blasting vibration attenuation prediction model in open-pit molybdenum mines according to claim 1, characterized in that, In step five, the joint calibration takes minimizing the root mean square error between the predicted peak particle vibration velocity and the measured peak particle vibration velocity as the objective function. The parameters to be determined include the attenuation coefficient, attenuation exponent, terrain influence coefficient, charge structure coefficient, and correction coefficient. Value boundary constraints are set for each parameter to be determined, and iterative optimization is performed until the objective function converges or the maximum number of iterations is reached. The value boundary of the attenuation coefficient is 50 to 500, the value boundary of the attenuation exponent is 1.0 to 2.5, the value boundary of the terrain influence coefficient is 0 to 2.0, and the charge structure coefficient and the correction coefficient are both positive numbers.

8. The method for constructing a multi-factor coupled blasting vibration attenuation prediction model in open-pit molybdenum mines according to claim 7, characterized in that, The nonlinear optimization algorithm is either particle swarm optimization or Levenberg-Marquardt nonlinear least squares algorithm.

9. The method for constructing a multi-factor coupled blasting vibration attenuation prediction model in open-pit molybdenum mines according to claim 1, characterized in that, In step six, the blasting vibration sample database is divided into a training set and a validation set according to engineering geological zones. The training set accounts for no less than 70%, and the validation set accounts for no more than 30%. The coefficient of determination, mean absolute percentage error, and root mean square error are used as evaluation indicators for zone validation. When the coefficient of determination of a certain zone is lower than 0.85 or the mean absolute percentage error is higher than 20%, at least one of the following measures is taken for iterative optimization: increasing the monitoring sample size of that zone, adjusting the correction coefficient value of that zone individually, or re-performing parameter calibration, until the validation accuracy of each zone meets the preset requirements.

10. The method for constructing a multi-factor coupled blasting vibration attenuation prediction model in open-pit molybdenum mines according to claim 1, characterized in that, Also includes: The blasting vibration attenuation prediction model, which has been verified and optimized by partitioning, is encapsulated into a callable calculation interface. The calculation interface receives the charge amount, propagation distance, lithology grade, elevation difference and charge structure parameters of the blasting scheme, outputs the predicted peak vibration velocity of the mass particles, and compares the predicted peak vibration velocity of the mass particles with the preset vibration safety control threshold. When the predicted peak vibration velocity of the mass particles exceeds the vibration safety control threshold, an early warning is issued.

Images