Intelligent regulation and control method and system for anchoring force of freight cableway in complex stratum

Abstract

The invention discloses an intelligent regulation and control method and system for anchoring force of a freight cableway in a complex stratum, and relates to the technical field of cableway engineering, and the method comprises the following steps: carrying out the temperature decoupling of a wavelength offset through a self-calibration temperature decoupling method, and obtaining an actual strain value and a strain value confidence coefficient; the wavelength offset is extracted from the obtained anchor rod multi-source data set; based on the actual strain value and the multi-source data set of the anchor rod, axial force distribution serves as a solving target, and axial force data of the anchor rod are obtained through a physical constraint inversion method; axial force data is predicted in real time through a Bayesian algorithm, and a compensation regulation and control instruction is generated according to the evolution trend of the predicted axial force data. The method is used for solving the problems that the axial force prediction result is not accurate due to temperature interference and the anchoring force regulation and control means are lagged.

Description

Technical Field

[0001] This invention relates to the field of cableway engineering technology, and more specifically, to an intelligent control method and system for the anchoring force of freight cableways in complex geological formations. Background Technology

[0002] Freight cableways, as a vital mode of transportation in mountainous and complex geological environments, play an irreplaceable role in mineral extraction, material transport in mountainous areas, and infrastructure construction. Their operational safety is highly dependent on the stability of the anchoring system. As a key load-bearing component of the cableway's ground anchoring system, the anchor bolt's stress state directly determines its anchoring performance and structural safety. However, in environments with weak surrounding rock, high stress zones, and complex hydrogeological conditions, the evolution of anchoring force is often influenced by the coupling effects of multiple factors, including rock heterogeneity, pore water pressure variations, and vibration disturbances. This causes the anchoring system to exhibit nonlinear, non-stationary, and uncertain characteristics during long-term service, posing challenges to monitoring and control.

[0003] In recent years, multi-source sensing methods have been introduced into the monitoring of anchoring systems, enabling real-time acquisition of strain, temperature, pore pressure, and dynamic disturbances, providing data support for axial force distribution inversion and anchoring performance evaluation. However, existing methods generally lack adaptive decoupling processes, resulting in temperature, noise, and cross-interference often being mixed into the sensor signals, leading to a decrease in the accuracy of strain-axial force conversion. Furthermore, most prediction models are based on empirical regression or single numerical simulations, failing to systematically incorporate mechanical equilibrium conditions and shear action models, making it difficult to reflect the actual anchoring stress process.

[0004] For example, the invention patent announcement CN112095596B, entitled "Intelligent Monitoring and Early Warning System and Method for Prestressed Anchor Bolts on Slopes Based on Cloud Platform," relates to the technical field of intelligent early warning for anchor bolts. The method includes the following steps: S1, real-time acquisition of strain values ​​at various points on the anchor bolt based on several stress gauges, and uploading the strain values ​​to a cloud server via a field communication gateway; S2, construction of an average shear stress prediction model for the anchor bolt based on the acquired strain values ​​at each point; S3, calculation of the porosity stress function relationship between the anchor bolt and the anchor hole; S4, correction of the average shear stress prediction model for the anchor bolt based on the porosity stress function relationship in S3, to obtain the anchor bolt shear stress prediction model. S5. Measure the absolute value D of the difference between the anchor bolt shear stress value output by the anchor bolt shear stress prediction model and the anchor bolt shear stress value measured by the magnetic flux sensor, in chronological order. S6. Based on the type of anchor bolt accident, use an expert knowledge base to determine the authenticity of the accident and divide the absolute value D into absolute values ​​d1, d2, d3, d4, ..., dn. S7. Based on deep learning and according to the divided absolute values ​​d1, d2, d3, d4, ..., dn, construct a neural network model for accident early warning and transmit the output accident early warning information to the management personnel APP and the on-site construction APP through the neural network model for accident early warning.

[0005] The above-disclosed technical solutions have at least the following technical problems: Relying mainly on stress plates and magnetic flux sensors, it is difficult to fully characterize the anchorage stress state under complex strata. Existing models only consider the average shear stress and pore stress function, without performing distributed load inversion, and cannot obtain the fine distribution of the axial force along the entire length of the anchorage. They only remain at the level of "accident early warning" and have not proposed an active compensation and intelligent control mechanism based on the prediction results, so they cannot achieve forward-looking adjustment of the anchorage force.

[0006] To address the above problems, this invention proposes a solution. Summary of the Invention

[0007] To overcome the aforementioned deficiencies of the prior art, embodiments of the present invention provide an intelligent control method and system for the anchoring force of freight cableways in complex geological formations. By decoupling temperature, the system inverts and predicts the evolution trend of axial force, and generates compensation control commands to solve the problems of inaccurate axial force prediction results due to temperature interference and the lag in anchoring force control methods.

[0008] To achieve the above objectives, the present invention provides the following technical solution: A method for intelligent control of anchorage force of freight cableways in complex geological formations includes the following steps: temperature decoupling of wavelength offset using a self-calibrated temperature decoupling method to obtain actual strain values ​​and strain confidence levels, wherein the wavelength offset is extracted from an acquired multi-source anchor dataset; based on the actual strain values ​​and the multi-source anchor dataset, axial force data of the anchors is obtained using a physical constraint inversion method, with axial force distribution as the solution objective; axial force data is predicted in real time using a Bayesian algorithm, and compensation control commands are generated based on the evolution trend of the predicted axial force data.

[0009] In a preferred embodiment, the anchor bolt multi-source dataset includes wavelength offset, temperature, pore water pressure, and vibration acceleration measured by multi-source sensors; the wavelength offset includes wavelength offset caused by strain and wavelength offset caused by temperature.

[0010] In a preferred embodiment, the step of temperature decoupling the wavelength offset using a self-calibrated temperature decoupling method to obtain the actual strain value specifically involves: obtaining a first sensor coefficient and constructing a strain coupling equation; obtaining a first wavelength offset and performing a first decoupling of the strain coupling equation using a dual-wavelength difference method to obtain a decoupled strain reference value; updating the first sensor coefficient based on a real-time feature vector and the first wavelength offset using a recursive least squares method to obtain a second sensor coefficient, wherein the real-time feature vector includes the strain reference value and the temperature change; and performing a second decoupling of the strain coupling equation based on the second sensor coefficient to obtain the actual strain value.

[0011] In a preferred embodiment, the method for obtaining the strain value confidence level specifically includes: calculating the second wavelength offset based on the second sensor coefficient using the strain coupling equation; calculating the Mahalanobis distance based on the residual between the second wavelength offset and the first wavelength offset; if the Mahalanobis distance result is greater than a first threshold, determining that the corresponding sensor is abnormal, and calculating the strain value confidence level based on the inverse proportional function of the Mahalanobis distance; if the Mahalanobis distance result is greater than a second threshold, determining that the corresponding sensor is invalid, and setting the strain value confidence level to zero.

[0012] In a preferred embodiment, the step of obtaining the axial force data of the anchor rod based on the actual strain value and the multi-source dataset of the anchor rod, with the axial force distribution as the solution objective, and through a physical constraint inversion method, specifically involves: calculating the distributed load on the anchor rod based on pore water pressure and vibration acceleration through linear coupling; establishing the axial force equilibrium equation based on the distributed load; and constructing an inversion model with the axial force distribution as the solution objective through Bayesian weighted least squares method, using the axial force equilibrium equation as the physical constraint, and outputting the axial force data.

[0013] In a preferred embodiment, the construction of the inversion model with axial force distribution as the solution objective specifically involves: concatenating several actual strain values ​​to form an actual strain vector, and generating a weight matrix based on the corresponding strain value confidence; converting the axial force data to be solved into an approximate strain vector through a strain-axial force transformation matrix; and, under the weighting of the weight matrix, minimizing the error between the actual strain vector and the approximate strain vector, and introducing a regularization term to construct the inversion model with axial force distribution as the solution objective.

[0014] In a preferred embodiment, the step of predicting axial force data in real time using a Bayesian algorithm and generating compensation and control commands based on the evolution trend of the predicted axial force data specifically involves: constructing a Bayesian linear model based on historical axial force data using a Bayesian algorithm; the Bayesian linear model consists of a state equation and an axial force observation equation, the latter being derived from the state equation; recursively solving the Bayesian linear model using a Kalman filter algorithm to obtain the predicted axial force data; and generating corresponding compensation and control commands based on the declining shape of the axial force curve when the predicted axial force data is below a warning threshold.

[0015] In a preferred embodiment, the step of recursively solving the Bayesian linear model using the Kalman filter algorithm to obtain the predicted axial force data specifically involves: calculating the prior distribution at the current time step based on the posterior distribution of the previous time step using the state equation, and calculating the predicted mean axial force at the current time step using the axial force observation equation based on the prior distribution; calculating the error between the axial force data at the current time step and the predicted mean axial force, and calculating the Kalman gain matrix based on the error; correcting the prior distribution at the current time step using the Kalman gain matrix to obtain the posterior distribution at the current time step; recursively deriving the state equation based on the posterior distribution at the current time step to obtain the prior distribution of the multi-step states, and calculating the predicted axial force data using the axial force observation equation.

[0016] In a preferred embodiment, generating compensation and control commands based on the evolution trend of the predicted axial force data specifically involves: calculating a relative local steepness index based on the first derivative of the predicted axial force curve; obtaining the maximum durability percentage of the minimum point of the axial force curve through durability analysis, wherein the durability of the minimum point is determined by the difference in function values ​​between the maximum and minimum points; determining whether the curve descent pattern is a local sudden drop or a uniform gradual drop based on the relative local steepness index and the maximum durability percentage; generating a grouting compensation command if the axial force curve exhibits a local sudden drop, and generating a tensioning compensation command if the axial force curve exhibits a uniform gradual drop.

[0017] An intelligent control system for the anchoring force of a freight cableway in complex geological formations includes: a temperature decoupling module for temperature decoupling of wavelength offset using a self-calibrated temperature decoupling method to obtain actual strain values ​​and strain confidence levels, wherein the wavelength offset is extracted from an acquired multi-source dataset of anchor bolts; a constraint inversion module for obtaining axial force data of anchor bolts using a physical constraint inversion method based on actual strain values ​​and the multi-source dataset of anchor bolts, with axial force distribution as the solution objective; and a prediction and control module for real-time prediction of axial force data using a Bayesian algorithm and generating compensation and control commands based on the evolution trend of the predicted axial force data.

[0018] The technical effects and advantages of the intelligent control method and system for anchoring force of freight cableways in complex strata according to the present invention are as follows: This invention eliminates the influence of temperature on strain signals by introducing a self-calibration and temperature decoupling method; it establishes an inversion model with axial force distribution as the solution objective by combining a physical constraint inversion method, thereby ensuring the consistency between the prediction and the actual stress state; it further adopts a dynamic linear model to predict the evolution trend of future effective anchoring force, and generates compensation and control commands based on the prediction results, thereby realizing real-time reliable monitoring and intelligent dynamic control of anchoring force. Attached Figure Description

[0019] Figure 1 This is a schematic flowchart of an intelligent control method for anchoring force of a freight cableway in complex strata, provided as an embodiment of the present invention.

[0020] Figure 2 This is a schematic diagram of an intelligent control system for anchoring force of a freight cableway in complex strata, provided as an embodiment of the present invention.

[0021] Figure 3 A line graph comparing the wavelength offset before and after decoupling, provided for an embodiment of the present invention. Detailed Implementation

[0022] 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 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.

[0023] Example 1, Figure 1 This invention presents an intelligent control method for the anchoring force of freight cableways in complex geological formations, comprising the following steps: S1, the wavelength offset is decoupled by temperature using a self-calibrated temperature decoupling method to obtain the actual strain value and the confidence level of the strain value. The wavelength offset is extracted from the acquired anchor bolt multi-source dataset. S2, based on actual strain values ​​and multi-source datasets of anchor bolts, takes axial force distribution as the solution objective and obtains the axial force data of anchor bolts through physical constraint inversion method; S3 uses a Bayesian algorithm to predict axial force data in real time and generates compensation and control commands based on the evolution trend of the predicted axial force data.

[0024] This invention eliminates the influence of temperature on strain signals by introducing a self-calibration and temperature decoupling method; it establishes an inversion model with axial force distribution as the solution objective by combining a physical constraint inversion method, thereby ensuring the consistency between the prediction and the actual stress state; it further adopts a dynamic linear model to predict the evolution trend of future effective anchoring force, and generates compensation and control commands based on the prediction results, thereby realizing real-time reliable monitoring and intelligent dynamic control of anchoring force.

[0025] S1, the wavelength offset is decoupled by temperature using a self-calibrated temperature decoupling method to obtain the actual strain value and the confidence level of the strain value. The wavelength offset is extracted from the acquired anchor bolt multi-source dataset.

[0026] In this embodiment, the anchor bolt multi-source dataset includes wavelength offset, temperature, pore water pressure, and vibration acceleration measured by multi-source sensors.

[0027] In this embodiment, a fiber Bragg grating (FBG) sensor array, a pore water pressure sensor, and a vibration sensor are deployed along the entire length of the anchor bolt. The FBG sensor characterizes strain and temperature changes using wavelength offset, the pore water pressure sensor outputs pore water pressure, and the vibration sensor outputs acceleration signals. The data acquisition frequency is no less than 10Hz. The wavelength offset, temperature, pore water pressure, and vibration acceleration measured by the multiple sensors are stitched together to form a multi-source dataset for the anchor bolt.

[0028] In this embodiment, the step of using a self-calibrated temperature decoupling method to decouple the wavelength offset to obtain the actual strain value specifically involves: Obtain the first sensor coefficients and construct the strain coupling equation; The first wavelength offset is obtained, and the strain coupling equation is decoupled once by the dual-wavelength difference method to obtain the decoupled strain reference value. Based on the real-time feature vector and the first wavelength offset, the first sensor coefficient is updated by the recursive least squares method to obtain the second sensor coefficient. The real-time feature vector includes the strain reference value and the temperature change. Based on the second sensor coefficient, the actual strain value is obtained by performing a second decoupling on the strain coupling equation.

[0029] In this embodiment, the wavelength offset includes wavelength offset caused by strain and wavelength offset caused by temperature.

[0030] In this embodiment, based on the wavelength offset and temperature in the anchor bolt multi-source data set, and the first sensor coefficient of the sensor, a strain coupling equation is established through the wavelength offset principle of the fiber Bragg grating (FBG) sensor. The first sensor coefficient includes a strain sensitivity coefficient and a temperature sensitivity coefficient. The first wavelength offset was observed from the fiber Bragg grating (FBG) sensor. The dual-wavelength difference method was used for decoupling. By setting two strain coupling equations with different strain sensitivity coefficients, the two equations were subtracted. Since the temperature change was the same, it was assumed that their temperature sensitivity coefficients were similar. The temperature terms were then largely canceled out, thus obtaining a rough estimate of the strain reference value that eliminated most of the temperature effect. Based on this, the first sensor coefficients are updated using the real-time feature vector and the first wavelength offset as inputs, and the second sensor coefficients are obtained by using the recursive least squares method. The real-time feature vector includes the strain reference value and the temperature change. Based on the second sensor coefficient, the strain coupling equation is decoupled again using the dual-wavelength difference method to obtain the actual strain value.

[0031] It should be noted that the first sensor coefficient uses the initial nominal value provided by the manufacturer or the estimated value obtained from the previous recursive least squares iteration.

[0032] The strain coupling equations are as follows:

[0033] In the formula, This is the wavelength offset. The strain sensitivity coefficient, The strain value, This is the temperature sensitivity coefficient. This represents the change in temperature.

[0034] The specific formula for the dual-wavelength difference method is as follows:

[0035] In the formula, This represents the wavelength offset of the first FBG sensor grating. This represents the wavelength offset of the second FBG sensor grating. The strain sensitivity coefficient of the first FBG sensor grating. This represents the strain sensitivity coefficient of the second FBG sensor grating. This is the strain reference value after decoupling.

[0036] The specific formula for the recursive least squares method is as follows:

[0037] In the formula, Let be the vector of parameters to be estimated. This is the updated strain sensitivity coefficient. This is the updated temperature sensitivity coefficient; For real-time feature vectors, for The strain value at time t, for The change in temperature over time; for The amount of wavelength change observed at any given time. Let Variance be the variance matrix of the parameter estimation error. It is a forgetting factor.

[0038] Figure 3 A line graph comparing the wavelength offset before and after decoupling in an embodiment of the present invention is provided.

[0039] In this embodiment, the method for obtaining the confidence level of the strain value is specifically as follows: The second wavelength offset is calculated using the strain coupling equation based on the second sensor coefficient. The Mahalanobis distance is calculated based on the residual between the second wavelength offset and the first wavelength offset; If the Mahalanobis distance result is greater than the first threshold, the corresponding sensor is determined to be abnormal, and the confidence level of the strain value is calculated based on the inverse proportional function of the Mahalanobis distance. If the Mahalanobis distance result is greater than the second threshold, the corresponding sensor is determined to be invalid, and the confidence level of the strain value is set to zero.

[0040] In this embodiment, the second sensor coefficients updated by the recursive least squares method are re-inputted into the strain coupling equation of wavelength drift to calculate the second wavelength offset; Calculate the residual between the first wavelength offset and the second wavelength offset obtained from actual observation, and calculate the Mahalanobis distance based on the residual under the constraint of the variance matrix. If the calculated Mahalanobis distance is greater than the set first threshold, it is determined that the corresponding sensor has an abnormal offset, and the system adjusts the confidence level of the strain value of the sensor according to the inverse proportional function of the Mahalanobis distance. If the calculated Mahalanobis distance is greater than the set second threshold, the sensor is determined to be in a severely failed state. The system will set the confidence level of the strain value corresponding to the sensor to zero and remove the observation results of the sensor in subsequent calculations and inversion processes.

[0041] The formula for calculating Mahalanobis distance is as follows:

[0042] In the formula, The Mahalanobis distance, For residuals, Let be the residual variance matrix.

[0043] S2, based on actual strain values ​​and multi-source datasets of anchor bolts, takes axial force distribution as the solution objective and obtains the axial force data of anchor bolts through physical constraint inversion method.

[0044] In this embodiment, the axial force data of the anchor rod is obtained through a physical constraint inversion method based on the actual strain value and the multi-source dataset of the anchor rod, with the axial force distribution as the solution objective. Specifically: Based on pore water pressure and vibration acceleration, the distributed load on the anchor bolt is calculated through linear coupling. Establish axial force equilibrium equations based on distributed loads; Using the axial force balance equation as the physical constraint, an inversion model with axial force distribution as the solution objective is constructed through the Bayesian weighted least squares method, and axial force data is output.

[0045] In this embodiment, the construction of the inversion model with axial force distribution as the solution objective is specifically as follows: Several actual strain values ​​are concatenated to form an actual strain vector, and a weight matrix is ​​generated based on the confidence level of the corresponding strain values. The axial force data to be solved is converted into an approximate strain vector by using the strain-axial force transformation matrix; Under the weighting of the weight matrix, the goal is to minimize the error between the actual strain vector and the approximate strain vector. A regularization term is introduced to construct an inversion model with the axial force distribution as the solution objective.

[0046] In this embodiment, the target inversion model is solved using the Markov chain Monte Carlo method to obtain the axial force data over the entire length of the anchor bolt.

[0047] It should be noted that the Markov chain Monte Carlo method is existing technology and will not be described in detail in this embodiment.

[0048] The formula for distributed load is as follows:

[0049] in, For the anchor bolt in position and time Distributed load, , The coupling coefficient is... For the anchor bolt in position and time pore water pressure, For the anchor bolt in position and time The vibration acceleration.

[0050] The axial force equilibrium equations are as follows:

[0051] In the formula, For the anchor bolt in position and time axial force.

[0052] The specific formula for constructing the inversion model using the Bayesian weighted least squares method is as follows:

[0053]

[0054] In the formula, This is the actual strain vector. This is the actual strain value; This is the weight matrix. Confidence level for strain values; This is the strain-axial force transformation matrix. For elastic modulus, Let be the cross-sectional area of ​​the anchor rod perpendicular to the axis. This is an approximate strain vector. For the differential operator matrix, For regularization parameters, This represents the axial force distribution vector of the entire anchor bolt. To minimize the dependent variable value.

[0055] S3 uses a Bayesian algorithm to predict axial force data in real time and generates compensation and control commands based on the evolution trend of the predicted axial force data.

[0056] In this embodiment, the step of predicting axial force data in real time using a Bayesian algorithm and generating compensation and control commands based on the evolution trend of the predicted axial force data specifically involves: Based on historical axial force data, a Bayesian linear model is constructed using a Bayesian algorithm. The Bayesian linear model consists of a state equation and an axial force observation equation, which is derived from the state equation. The Bayesian linear model is solved recursively using the Kalman filter algorithm to obtain the predicted axial force data; When the predicted axial force data is lower than the warning threshold, a corresponding compensation control command is generated based on the downward trend of the axial force curve.

[0057] In this embodiment, the axial force observation equation is as follows:

[0058] In the formula, This represents the observed axial force value at the current moment. To design the matrix, For state vectors, To observe the noise, To observe the noise variance.

[0059] The state equations are as follows:

[0060] For state vectors, Here is the state transition matrix. State noise, Let V be the state noise variance.

[0061] It should be noted that the design matrix is ​​a matrix that maps the state vector to the observation space; the state transition matrix is ​​a matrix that automatically evolves the state vector from the previous time step to the current time step.

[0062] In this embodiment, the step of recursively solving the Bayesian linear model using the Kalman filter algorithm to obtain the predicted axial force data specifically involves: Based on the posterior distribution of the previous time step, the prior distribution of the current time step is calculated through the state equation, and the predicted mean value of the axial force at the current time step is calculated through the axial force observation equation based on the prior distribution. Calculate the error between the current axial force data and the predicted mean axial force, and calculate the Kalman gain matrix based on the error; The prior distribution at the current time step is corrected using the Kalman gain matrix to obtain the posterior distribution at the current time step. Based on the posterior distribution of the current moment, the recursive state equation is derived to obtain the prior distribution of the multi-step state, and the predicted axial force data is calculated through the axial force observation equation.

[0063] In this embodiment, it is assumed that at time The posterior distribution is as follows:

[0064] in, for The state vector at time t, For time 1 to All data at any given time, The mean of the state estimates. Let be the variance matrix.

[0065] The prior distribution at the current moment is calculated using the state equation as follows:

[0066] In the formula, The mean of the state predictions. This represents the variance of the state prediction.

[0067] In this embodiment, the predicted mean value of the axial force at the current moment is calculated based on the prior distribution using the axial force observation equation, as follows:

[0068] In the formula, To One-step prediction value, This represents the variance of the one-step prediction.

[0069] In this embodiment, the error between the actual axial force data and the predicted average axial force is calculated as follows:

[0070] In the formula, This represents the prediction error.

[0071] In this embodiment, the formula for calculating the Kalman gain matrix is ​​as follows:

[0072] In the formula, This is the Kalman gain matrix.

[0073] In this embodiment, the posterior distribution at the current time is as follows:

[0074] In the formula, Estimate the updated mean for the state. Estimate the updated variance for the state.

[0075] In this embodiment, the state equation is recursively applied based on the posterior distribution at the current moment to obtain the multi-step state prior distribution, as follows:

[0076] In the formula, For the first The mean of the state prediction for each step. For the first The variance of the state prediction for each step. For the first The state transition matrix of the step, For the first The state noise variance of the step.

[0077] In this embodiment, the predicted axial force data is calculated using the observation equation, as follows:

[0078] In the formula, For the future The predicted value of the axial force point of the step. To predict the variance of the distribution, To design the matrix, For the first The variance of observation noise at each step.

[0079] In this embodiment, when the predicted axial force data is lower than the warning threshold, a corresponding compensation and control command is generated based on the downward trend of the axial force curve.

[0080] In this embodiment, generating compensation and control commands based on the evolution trend of the predicted axial force data specifically involves: The relative local steepness index is calculated based on the first derivative of the predicted axial force curve. The maximum durability percentage of the minimum point of the axial force curve is obtained by durability analysis. The durability of the minimum point is determined by the difference in function values ​​between the maximum and the minimum point. Based on the relative local steepness index and the percentage of maximum persistence, the curve descent pattern is determined to be either a local sudden drop or a uniform gradual drop. If the axial force curve shows a local sudden drop, a grouting compensation command will be generated. If the axial force curve shows a uniform and gradual decrease, a tension compensation command is generated.

[0081] It should be noted that the flatter and longer the axial force curve declines, the more the load is distributed among the strata, and the stronger the overall capacity (anchoring force) of the anchoring system. Conversely, the steeper and shorter the axial force curve declines, the weaker the strata, the shorter the effective anchoring section, and the weaker the overall anchoring force. When the axial force curve declines evenly and gradually, it indicates that the anchoring system is generally robust, but the tension of the anchor bolt itself needs to be fine-tuned to ensure that the entire anchor bolt section is evenly stressed and maintains the designed prestressed state. When the axial force curve shows a sudden drop in a local area, it indicates that the local bearing capacity is insufficient and needs to be quickly reinforced. In this case, for the section with a sudden drop in axial force, grouting is used to enhance the bond and bearing capacity between the anchor bolt and the surrounding strata, thereby improving the anchoring capacity of the local section.

[0082] In this embodiment, the formula for calculating the relative local steepness index is as follows:

[0083] In the formula, This is a relative local steepness index. For the first The first derivative of each sampling point To prevent small constants with a denominator of zero.

[0084] The formula for calculating the maximum durability percentage is as follows:

[0085]

[0086] In the formula, The percentage of maximum durability at the smallest point. The maximum durability at the minimum point of the curve. The sum of the durability of all minima. To prevent small constants with a denominator of zero, For minimal point durability, This represents the axial force value at a local maximum point. This represents the axial force value at a local minimum point.

[0087] In this embodiment, based on the combined result of the relative local steepness index and the maximum durability percentage, the curve is determined to be either a local sudden drop or a uniform gradual drop, as detailed below: like and It was determined to be a localized sudden drop; like and And the mean of the derivative It was determined to be a uniform, gradual descent; Other cases are marked as ambiguous and require short-term review.

[0088] In this embodiment, if the axial force curve shows a local sudden drop, a grouting compensation command is generated. If the axial force curve shows a uniform and gradual decrease, a tension compensation command is generated.

[0089] Example 2, Figure 2 This invention presents an intelligent control system for the anchoring force of a freight cableway in complex geological formations, comprising: The temperature decoupling module is used to perform temperature decoupling on the wavelength offset using a self-calibrated temperature decoupling method to obtain the actual strain value and the strain value confidence level. The wavelength offset is extracted from the acquired anchor bolt multi-source dataset. The constraint inversion module is used to obtain the axial force data of the anchor rod based on the actual strain value and the multi-source dataset of the anchor rod, with the axial force distribution as the solution objective, through the physical constraint inversion method. The predictive control module is used to predict axial force data in real time using a Bayesian algorithm and generate compensation control commands based on the evolution trend of the predicted axial force data.

[0090] The above formulas are all dimensionless calculations. The formulas are derived from software simulations based on a large amount of collected data to obtain the most recent real-world results. The preset parameters in the formulas are set by those skilled in the art according to the actual situation.

[0091] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented, in whole or in part, in the form of a computer program product.

[0092] Those skilled in the art will recognize that the modules and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0093] In addition, the functional modules in the various embodiments of this application can be integrated into one processing module, or each module can exist physically separately, or two or more modules can be integrated into one module.

[0094] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

[0095] In conclusion, the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for intelligent control of anchorage force of freight cableways in complex geological formations, characterized in that, Includes the following steps: The wavelength offset is decoupled by temperature using a self-calibrated temperature decoupling method to obtain the actual strain value and the confidence level of the strain value. The wavelength offset is extracted from the acquired anchor bolt multi-source dataset. Based on actual strain values ​​and multi-source datasets of anchor bolts, axial force distribution is used as the solution objective, and the axial force data of anchor bolts is obtained through physical constraint inversion method; The axial force data is predicted in real time using a Bayesian algorithm, and compensation control commands are generated based on the evolution trend of the predicted axial force data.

2. The intelligent control method for anchorage force of freight cableways in complex strata according to claim 1, characterized in that, The anchor bolt multi-source dataset includes wavelength offset, temperature, pore water pressure, and vibration acceleration measured by multi-source sensors; the wavelength offset includes wavelength offset caused by strain and wavelength offset caused by temperature.

3. The intelligent control method for anchorage force of freight cableways in complex strata according to claim 2, characterized in that, The method of temperature decoupling the wavelength offset using a self-calibrated temperature decoupling method to obtain the actual strain value is as follows: Obtain the first sensor coefficients and construct the strain coupling equation; The first wavelength offset is obtained, and the strain coupling equation is decoupled once by the dual-wavelength difference method to obtain the decoupled strain reference value. Based on the real-time feature vector and the first wavelength offset, the first sensor coefficient is updated by the recursive least squares method to obtain the second sensor coefficient. The real-time feature vector includes the strain reference value and the temperature change. Based on the second sensor coefficient, the actual strain value is obtained by performing a second decoupling on the strain coupling equation.

4. The intelligent control method for anchorage force of freight cableways in complex strata according to claim 3, characterized in that, The method for obtaining the confidence level of the strain value is as follows: The second wavelength offset is calculated using the strain coupling equation based on the second sensor coefficient. The Mahalanobis distance is calculated based on the residual between the second wavelength offset and the first wavelength offset; If the Mahalanobis distance result is greater than the first threshold, the corresponding sensor is determined to be abnormal, and the confidence level of the strain value is calculated based on the inverse proportional function of the Mahalanobis distance. If the Mahalanobis distance result is greater than the second threshold, the corresponding sensor is determined to be invalid, and the confidence level of the strain value is set to zero.

5. The intelligent control method for anchorage force of freight cableways in complex strata according to claim 4, characterized in that, Based on actual strain values ​​and multi-source datasets of anchor bolts, and with axial force distribution as the solution objective, the axial force data of the anchor bolts is obtained through a physical constraint inversion method, specifically as follows: Based on pore water pressure and vibration acceleration, the distributed load on the anchor bolt is calculated through linear coupling. Establish axial force equilibrium equations based on distributed loads; Using the axial force balance equation as the physical constraint, an inversion model with axial force distribution as the solution objective is constructed through the Bayesian weighted least squares method, and axial force data is output.

6. The intelligent control method for anchorage force of freight cableways in complex strata according to claim 5, characterized in that, The inversion model constructed with axial force distribution as the solution objective is as follows: Several actual strain values ​​are concatenated to form an actual strain vector, and a weight matrix is ​​generated based on the confidence level of the corresponding strain values. The axial force data to be solved is converted into an approximate strain vector by using the strain-axial force transformation matrix; Under the weighting of the weight matrix, the goal is to minimize the error between the actual strain vector and the approximate strain vector. A regularization term is introduced to construct an inversion model with the axial force distribution as the solution objective.

7. The intelligent control method for anchorage force of freight cableways in complex strata according to claim 6, characterized in that, The process involves real-time prediction of axial force data using a Bayesian algorithm, followed by the generation of compensation and control commands based on the evolution trend of the predicted axial force data. Specifically: Based on historical axial force data, a Bayesian linear model is constructed using a Bayesian algorithm. The Bayesian linear model consists of a state equation and an axial force observation equation, which is derived from the state equation. The Bayesian linear model is solved recursively using the Kalman filter algorithm to obtain the predicted axial force data; When the predicted axial force data is lower than the warning threshold, a corresponding compensation control command is generated based on the downward trend of the axial force curve.

8. The intelligent control method for anchorage force of freight cableways in complex strata according to claim 7, characterized in that, The method of recursively solving the Bayesian linear model using the Kalman filter algorithm to obtain the predicted axial force data is as follows: Based on the posterior distribution of the previous time step, the prior distribution of the current time step is calculated through the state equation, and the predicted mean value of the axial force at the current time step is calculated through the axial force observation equation based on the prior distribution. Calculate the error between the current axial force data and the predicted mean axial force, and calculate the Kalman gain matrix based on the error; The prior distribution at the current time step is corrected using the Kalman gain matrix to obtain the posterior distribution at the current time step. Based on the posterior distribution of the current moment, the recursive state equation is derived to obtain the prior distribution of the multi-step state, and the predicted axial force data is calculated through the axial force observation equation.

9. The intelligent control method for anchorage force of freight cableways in complex strata according to claim 8, characterized in that, The process of generating compensation and control commands based on the evolution trend of predicted axial force data specifically includes: The relative local steepness index is calculated based on the first derivative of the predicted axial force curve. The maximum durability percentage of the minimum point of the axial force curve is obtained by durability analysis. The durability of the minimum point is determined by the difference in function values ​​between the maximum and the minimum point. Based on the relative local steepness index and the percentage of maximum persistence, the curve descent pattern is determined to be either a local sudden drop or a uniform gradual drop. If the axial force curve shows a local sudden drop, a grouting compensation command will be generated. If the axial force curve shows a uniform and gradual decrease, a tension compensation command is generated.

10. A system for intelligent control of anchorage force of a freight cableway in complex strata as described in any one of claims 1-9, comprising: The temperature decoupling module is used to perform temperature decoupling on the wavelength offset using a self-calibrated temperature decoupling method to obtain the actual strain value and the strain value confidence level. The wavelength offset is extracted from the acquired anchor bolt multi-source dataset. The constraint inversion module is used to obtain the axial force data of the anchor rod based on the actual strain value and the multi-source dataset of the anchor rod, with the axial force distribution as the solution objective, through the physical constraint inversion method. The predictive control module is used to predict axial force data in real time using a Bayesian algorithm and generate compensation control commands based on the evolution trend of the predicted axial force data.

Images