Tunnel mechanical fault shutdown influence prediction method and system based on CART decision tree

Abstract

The invention provides a tunnel mechanical fault shutdown influence prediction method and system based on a CART decision tree, and belongs to the technical field of tunnel engineering. The method comprises the steps that equipment operation data and construction environment parameters in drilling and blasting method tunnel construction are collected, and a multi-dimensional feature set is constructed; performing preprocessing and feature screening on the data in the multi-dimensional feature set, and establishing a fault-shutdown associated feature matrix; based on a CART decision tree algorithm, constructing a fault shutdown influence prediction model by using the fault-shutdown association feature matrix; inputting equipment state data collected in real time and construction parameters into the prediction model, and dynamically outputting a prediction result through decision tree rule matching; and adaptively adjusting and optimizing the parameters of the prediction model according to the dynamic change of the construction stage. According to the method, multi-dimensional, high-precision and interpretable prediction on the mechanical fault shutdown influence can be realized, and intelligent support is provided for construction decision making.

Description

Technical Field

[0001] This invention belongs to the field of tunnel engineering technology, and in particular relates to a method and system for predicting the impact of tunnel mechanical failure downtime based on CART decision tree. Background Technology

[0002] The statements in this section are merely background information related to the present invention and do not necessarily constitute prior art.

[0003] In the field of tunnel construction, the drill-and-blast method has become a core technical solution for the construction of deep-buried and long tunnels due to its advantages such as strong adaptability and controllable cost. However, the construction environment of such tunnels is characterized by high stress, high humidity, and complex and variable geological conditions. Key construction machinery and equipment such as drilling rigs, loaders, and dump trucks need to operate continuously under high load and high wear conditions for a long time, leading to frequent mechanical failures such as hydraulic system failure, power system failure, and electrical component damage. According to industry statistics, construction downtime caused by mechanical failures accounts for 15%-30% of the total tunnel construction period. This not only directly causes economic losses such as equipment maintenance and schedule delays, but also triggers a chain reaction of poor coordination of subsequent processes and imbalance in resource allocation, seriously restricting the improvement of construction efficiency and the realization of overall project benefits.

[0004] Currently, the prediction of the impact of tunnel mechanical failures on work stoppages mainly relies on the subjective judgment of engineering technicians and the statistical analysis of equipment maintenance logs. This traditional method has significant technical shortcomings: on the one hand, failure records are mostly retrospective descriptions, lacking a systematic analysis of the quantitative correlation between failure characteristics and the degree of work stoppage impact, making it difficult to form reusable predictive data; on the other hand, the diverse and complex formats of multi-source heterogeneous data generated during tunnel construction, such as equipment sensor data, construction logs, and geological survey parameters, make it impossible for traditional methods to effectively integrate and utilize them. This results in prediction results being greatly influenced by the subjective experience of personnel, making it difficult to adapt to dynamically changing construction environments. Especially under complex geological conditions, the coupling effect of geological factors such as abrupt changes in surrounding rock grade and uneven distribution of fracture zones with mechanical failures often causes the deviation of traditional prediction methods to exceed 40%, seriously affecting the scientific nature of construction plans and the timeliness of emergency response.

[0005] To improve the accuracy of mechanical failure prediction, some technologies have been researched and applied in related fields. For example, failure prediction is based on support vector machines and Bayesian networks. However, these existing technologies still have significant limitations: First, they only target single equipment or specific subsystems and do not fully consider the interactive effects between the construction environment, geological conditions, and mechanical failures, thus limiting the prediction scenarios. Second, the prediction models are mostly static and cannot dynamically adjust parameters according to the progress of construction and changes in equipment operating conditions, resulting in insufficient adaptability. Third, they only focus on diagnosing whether a failure has occurred, lacking quantitative output of key engineering indicators such as downtime, impact range, and economic losses, making it difficult to directly support decisions such as construction scheduling and resource allocation. In addition, the core objective of existing technologies is mostly failure diagnosis rather than downtime impact prediction, which limits their engineering practicality and field adaptability, and cannot fundamentally solve the problem of accurately predicting the impact of mechanical failure downtime in tunnel construction. Summary of the Invention

[0006] To overcome the shortcomings of the existing technologies, this invention provides a method and system for predicting the impact of tunnel mechanical failure shutdowns based on CART decision trees. By integrating multi-source data to construct a multi-dimensional feature set, and using the CART decision tree algorithm to establish a dynamic prediction model, it achieves accurate quantitative assessment of shutdown risk level, shutdown duration, and economic losses, providing scientific support for tunnel construction decisions.

[0007] To achieve the above objectives, one or more embodiments of the present invention provide the following technical solutions: The first aspect of this invention provides a method for predicting the impact of tunnel mechanical failure downtime based on CART decision trees; A method for predicting the impact of tunnel mechanical failure downtime based on CART decision trees includes: Collect equipment operation data and construction environment parameters during drill-and-blast tunnel construction, and construct a multi-dimensional feature set; The data in the multidimensional feature set are preprocessed and feature-filtered to establish a fault-downtime correlation feature matrix; Based on the CART decision tree algorithm, a fault-stoppage impact prediction model is constructed using the fault-stoppage correlation feature matrix. The real-time collected equipment status data and construction parameters are input into the prediction model, and the prediction results are dynamically output through decision tree rule matching. The parameters of the prediction model are adaptively adjusted and optimized based on the dynamic changes during the construction phase.

[0008] As a further technical solution, the equipment operation data includes vibration, temperature, oil pressure and current data of the rock drilling rig and loader collected by sensors, maintenance records and fault shutdown logs.

[0009] As a further technical solution, the construction environment parameters include the surrounding rock grade, groundwater distribution, construction stage, working face temperature and humidity, and geological risk level.

[0010] As a further technical solution, the data in the multidimensional feature set is preprocessed and feature-filtered to establish a fault-downtime correlation feature matrix, including: Missing data are filled using multiple interpolation, outliers are detected and removed using box plots, and dimensional differences are eliminated by Z-score standardization. The preprocessed data were sorted by Pearson correlation coefficient and random forest feature importance to screen features that are strongly correlated with the impact of shutdown and construct a fault-shutdown correlation feature matrix.

[0011] As a further technical solution, the method of constructing a fault-stoppage impact prediction model based on the CART decision tree algorithm and utilizing the fault-stoppage correlation feature matrix includes: The dataset is recursively partitioned using the Gini coefficient as the node splitting criterion; the cost complexity pruning is used to balance the model complexity, and the shutdown risk level is defined based on the leaf node sample distribution.

[0012] As a further technical solution, the prediction results include the level of work stoppage risk, the expected duration of work stoppage, and the range of economic losses.

[0013] As a further technical solution, the parameters of the prediction model are adaptively adjusted and optimized, including: When the monthly new data and feature weight changes exceed the set threshold, incremental learning is triggered and the segmentation threshold is recalculated for nodes whose feature weight changes exceed the threshold. Retain the subtree structure in the historical decision-making path that has a verification accuracy of >85%.

[0014] The second aspect of this invention provides a tunnel mechanical failure downtime impact prediction system based on CART decision tree.

[0015] A tunnel mechanical failure downtime impact prediction system based on CART decision trees includes: The data acquisition module is configured to: collect equipment operation data and construction environment parameters during drill-and-blast tunnel construction, and construct a multi-dimensional feature set; The feature matrix construction module is configured to: preprocess and filter the data in the multidimensional feature set to establish a fault-downtime correlation feature matrix; The prediction model building module is configured to: construct a prediction model of the impact of failure and shutdown based on the CART decision tree algorithm and using the failure-shutdown correlation feature matrix. The prediction result output module is configured to: input real-time collected equipment status data and construction parameters into the prediction model, and dynamically output the prediction result through decision tree rule matching; The model optimization module is configured to adaptively adjust and optimize the parameters of the prediction model based on the dynamic changes during the construction phase.

[0016] A third aspect of the present invention provides a computer-readable storage medium having a program stored thereon, which, when executed by a processor, implements the steps in the method for predicting the impact of tunnel mechanical failure shutdowns based on CART decision trees as described in the first aspect of the present invention.

[0017] A fourth aspect of the present invention provides an electronic device, including a memory, a processor, and a program stored in the memory and executable on the processor, wherein the processor executes the program to implement the steps in the method for predicting the impact of tunnel mechanical failure shutdowns based on CART decision trees as described in the first aspect of the present invention.

[0018] The above one or more technical solutions have the following beneficial effects: This invention breaks through the limitations of traditional methods that focus only on data from a single device. It integrates equipment operating status, construction environment parameters, and construction stage characteristics to construct a multi-dimensional, comprehensive fault-downtime correlation feature matrix. By using Pearson correlation coefficient and random forest feature filtering, it accurately captures key influencing factors. Combined with the strong classification capability of the CART decision tree algorithm, it greatly enhances the model's prediction accuracy and effectively solves the problems of insufficient data utilization and large prediction bias in existing technologies.

[0019] By introducing an incremental learning mechanism and a dynamic parameter adjustment strategy, when changes occur during the construction phase (such as sudden changes in surrounding rock grade or equipment load upgrades) or when 1000 new data points are added each month, the model automatically updates feature weights and recalculates the segmentation threshold, while retaining historical high-accuracy decision paths, thus achieving dynamic optimization of the tree structure. This design enables the model to adapt to the dynamic changes in tunnel construction in real time, significantly reducing prediction bias and greatly improving its applicability under complex geological and construction conditions.

[0020] Advantages of additional aspects of the invention will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention. Attached Figure Description

[0021] The accompanying drawings, which form part of this invention, are used to provide a further understanding of the invention. The illustrative embodiments of the invention and their descriptions are used to explain the invention and do not constitute an improper limitation of the invention.

[0022] Figure 1 This is a flowchart of the method in the first embodiment.

[0023] Figure 2 This is a system structure diagram of the second embodiment. Detailed Implementation

[0024] It should be noted that the following detailed descriptions are exemplary and intended to provide further illustration of the invention. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0025] It should be noted that the terminology used herein is for the purpose of describing particular implementations only and is not intended to limit the exemplary implementations of the present invention.

[0026] Where there is no conflict, the embodiments and features in the embodiments of the present invention can be combined with each other.

[0027] Example 1 This embodiment discloses a method for predicting the impact of tunnel machinery failure shutdowns based on CART decision trees. By integrating equipment operating status data, construction environment parameters, and historical fault information, a multi-dimensional feature set is constructed. The CART decision tree algorithm is used to establish a shutdown impact prediction model, which enables quantitative assessment of shutdown risk level, shutdown duration, and economic losses. The model parameters are dynamically optimized, significantly improving prediction accuracy and engineering applicability.

[0028] Specifically, such as Figure 1 As shown, the method for predicting the impact of tunnel mechanical failure downtime based on CART decision trees includes: Step S1: Collect equipment operation data and construction environment parameters during drill-and-blast tunnel construction, and construct a multi-dimensional feature set.

[0029] For equipment operation data, an IoT sensor network is deployed on core components (such as hydraulic pumps, engines, and drive shafts) of key construction machinery such as rock drilling rigs and loaders. Vibration acceleration sensors, temperature sensors, pressure sensors, and Hall current sensors are used for real-time monitoring to obtain vibration, temperature, oil pressure, and current data of the rock drilling rigs and loaders.

[0030] By interfacing with the enterprise equipment management system's API, maintenance records and downtime logs are retrieved periodically or triggered automatically. For unstructured text, natural language processing techniques are employed, using the TF-IDF algorithm combined with a predefined dictionary of tunnel machinery fault keywords to extract key fault features, which are then encoded as discrete numerical variables.

[0031] For construction environment parameters, the surrounding rock grade in front of the tunnel face is extracted from the digital database of geological survey reports; the width of the fracture zone and the rock mass integrity index are obtained using an advanced geological prediction system; and the groundwater inflow is monitored in real time by a flow meter installed in the drainage pipeline.

[0032] Based on the construction log, the current process stage is obtained, and temperature and humidity data of the working face are collected by temperature and humidity sensors deployed on the working face. A quantitative risk score is output by constructing a geological risk coefficient calculation model.

[0033] By integrating the equipment operation data with construction environment parameters, a multidimensional feature set is obtained.

[0034] Step S2: Preprocess and filter the data in the multidimensional feature set to establish a fault-downtime correlation feature matrix.

[0035] For missing values ​​in sensor data, multiple interpolation (MICE) is used to fill them with the linear regression results of the five most recent valid measurements; for missing items in operation and maintenance records and geological data, they are filled by inference through correlation data. For continuous sensor data such as hydraulic pressure and vibration acceleration, box plot detection is used to remove abnormal records that are out of range; for classified data such as geological parameters and construction stages, logical verification is used to remove contradictory data. Z-score normalization is applied to continuous characteristics (such as vibration RMS value, hydraulic oil temperature, groundwater flow rate, etc.).

[0036] Where μ is the mean and σ is the standard deviation, eliminating the influence of different units on model training; Classification features (such as surrounding rock grade and construction stage) are converted into numerical features using label coding or unique thermal coding.

[0037] Furthermore, the preprocessed data were ranked based on the importance of features using Pearson correlation coefficient and random forest. Features strongly correlated with downtime were selected using Pearson correlation coefficient, such as the cumulative duration of hydraulic oil temperature exceeding the standard and the continuous working cycle of the equipment. Random forest was used to calculate the importance of features, and features with an importance score > 0.05 were retained.

[0038] The selected features are combined into a fault-downtime correlation feature matrix, specifically including equipment status features and environmental parameter features. Equipment status features include the cumulative duration of hydraulic oil temperature exceeding the limit, the continuous operating cycle of the equipment, and the peak value of vibration acceleration. Environmental parameter features include the surrounding rock fracturing index and the geological risk coefficient. Assuming there are n features, the fault-downtime correlation feature matrix X can be represented as:

[0039] in Let m represent the i-th feature and m represent the number of samples.

[0040] The constructed fault-stoppage correlation feature matrix is ​​used in the subsequent construction of the CART decision tree model to achieve a quantitative assessment of the stoppage risk level, expected stoppage duration, and economic losses. Through the above steps, a fault-stoppage correlation feature matrix can be effectively established, providing the necessary input data for the decision tree model, thereby achieving high-precision prediction of the impact of stoppages.

[0041] Step S3: Based on the CART decision tree algorithm, construct a fault-stoppage impact prediction model using the fault-stoppage association feature matrix.

[0042] During model training, a stratified sampling method was used to divide the sample set corresponding to the fault-downtime correlation feature matrix to avoid model bias caused by uneven sample distribution. The sample set was split into training, validation, and test sets in a 7:2:1 ratio. The training set was used for model parameter learning and decision tree construction, the validation set was used for pruning strategy optimization and hyperparameter tuning, and the test set was used for final model performance evaluation. Downtime risk level was used as the stratification label to ensure that the proportion of low, medium, and high risk samples in the training, validation, and test sets was consistent with that in the original sample set, ensuring the model's adaptability to scenarios with different risk levels.

[0043] Furthermore, the CART decision tree algorithm uses the Gini coefficient as the node splitting criterion and generates the optimal binary tree structure by recursively partitioning the dataset. The core objective is to maximize the sample purity of each child node.

[0044] For the sample set D of the current node, the Gini coefficient is calculated as follows:

[0045] in, Gini coefficient; To output the number of categories for the label, For sample set The Middle The percentage of samples with class labels; the closer the Gini coefficient is to 0, the higher the sample purity and the better the classification effect.

[0046] To address the overfitting problem in decision trees and balance model complexity with generalization ability, a cost-complexity pruning strategy is employed to optimize the generated complete decision tree. α=0.01 is set to balance the risk of overfitting, and subtrees that improve validation set accuracy by less than 2% are removed.

[0047] Specifically, first, a complete decision tree T is generated using the CART algorithm. Then, the tree is traversed from bottom to top, and the cost complexity of each subtree is calculated. For each subtree T', compute its cost complexity function: ,in For subtree Prediction error on the training set, For subtree The number of leaf nodes, This is a hyperparameter.

[0048] For satisfying subtree This involves retaining the existing branches and pruning others, thus preserving the subtree structures that improve the model's generalization performance. By adjusting... The value of can control the intensity of pruning. The larger the value, the greater the penalty to model complexity and the more thorough the pruning; conversely, the smaller the value, the greater the penalty to model complexity and the more thorough the pruning. The smaller the value, the smaller the penalty to model complexity, and the more subtree branches are retained, thus achieving the goal of balancing model complexity and prediction performance.

[0049] Finally, based on the distribution of leaf node samples, the risk level of work stoppage was classified into low risk, medium risk, and high risk.

[0050] Step S4: Input the real-time collected equipment status data and construction parameters into the prediction model, and dynamically output the prediction results through decision tree rule matching.

[0051] The system collects equipment status data and construction parameters in real time, fills in and replaces missing and abnormal data, calls the standardized parameters stored in the model, performs Z-score standardization on continuous features, and encodes categorical features according to preset rules to ensure that the data format is consistent with the model input.

[0052] The preprocessed data is input into the prediction model, and matching is performed using decision tree rules, as follows: With the goal of minimizing the Gini coefficient, the dataset is recursively divided into two parts. The formula for calculating the Gini coefficient is: ,in Let represent the proportion of samples in class k. By continuously searching for the features and their split points that minimize the Gini coefficient, a binary decision tree structure is constructed.

[0053] Cost Complexity Pruning (CCP) is used to balance model complexity and overfitting. The cost complexity function is defined as: ,in For the prediction error on the training data, The number of leaf nodes, These are hyperparameters. Adjust them... The value controls the intensity of pruning and preserves critical decision paths.

[0054] During real-time prediction, newly input equipment status data and construction environment parameters are matched on the decision tree from top to bottom. Along the decision path, the conditions are judged based on the features of the nodes, and finally a leaf node is reached. The output is based on the shutdown risk level, expected shutdown duration and economic loss range represented by the leaf node.

[0055] The final output includes the work stoppage risk level, estimated stoppage duration, economic loss range, and key risk factors. A concise pop-up notification is pushed to multiple terminals via the construction management platform, while the structured results are stored in a MySQL database, automatically generating a PDF report containing data details and trend predictions.

[0056] Step S5: Adaptively adjust and optimize the parameters of the prediction model based on the dynamic changes during the construction phase.

[0057] A dual trigger condition is set: when a critical transition occurs during the construction phase or when 1000 new valid data entries are added each month, the model parameter optimization process is automatically triggered. By monitoring the process codes of the construction scheduling system and the incremental data in the database in real time, the trigger nodes are accurately identified to ensure timely optimization.

[0058] Secondly, the parameters are adaptively adjusted. The importance score of each feature in the newly added dataset is calculated. For features with changes greater than 10%, their weight in the decision tree is readjusted to strengthen the influence of dynamically changing features. For features with significant weight changes, the Gini coefficient segmentation threshold is recalculated based on the new data, replacing the old threshold in the original decision rule to ensure the segmentation standard matches the current construction conditions. Effective subtrees with a validation accuracy greater than 85% in the historical decision paths are retained, while redundant branches whose accuracy decreases due to changes in the construction environment are pruned. Cost complexity pruning (α=0.01) balances the complexity and generalization ability of the new tree structure.

[0059] Finally, the new dataset after incremental learning is used for 10-fold cross-validation. The optimized model parameters are overwritten in real time in the original storage file, and the decision rule set and visual risk assessment template are updated synchronously to achieve seamless adaptation of the model to the dynamic changes in the construction stage and continuously ensure prediction accuracy.

[0060] Example 2 This embodiment discloses a tunnel mechanical failure downtime impact prediction system based on CART decision tree; like Figure 2 As shown, the tunnel mechanical failure downtime impact prediction system based on CART decision tree includes: The data acquisition module is configured to: collect equipment operation data and construction environment parameters during drill-and-blast tunnel construction, and construct a multi-dimensional feature set; The feature matrix construction module is configured to: preprocess and filter the data in the multidimensional feature set to establish a fault-downtime correlation feature matrix; The prediction model building module is configured to: construct a prediction model of the impact of failure and shutdown based on the CART decision tree algorithm and using the failure-shutdown correlation feature matrix. The prediction result output module is configured to: input real-time collected equipment status data and construction parameters into the prediction model, and dynamically output the prediction result through decision tree rule matching; The model optimization module is configured to adaptively adjust and optimize the parameters of the prediction model based on the dynamic changes during the construction phase.

[0061] Example 3 The purpose of this embodiment is to provide a computer-readable storage medium.

[0062] A computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps in the tunnel mechanical failure downtime impact prediction method based on CART decision tree as described in Example 1.

[0063] Example 4 The purpose of this embodiment is to provide an electronic device.

[0064] An electronic device includes a memory, a processor, and a program stored in the memory and executable on the processor. When the processor executes the program, it implements the steps in the tunnel machinery failure downtime impact prediction method based on CART decision tree as described in Example 1.

[0065] The steps and methods involved in the apparatuses of Embodiments 2, 3, and 4 above correspond to those in Embodiment 1. For specific implementation details, please refer to the relevant description section of Embodiment 1. The term "computer-readable storage medium" should be understood as a single medium or multiple media including one or more instruction sets; it should also be understood as including any medium capable of storing, encoding, or carrying an instruction set for execution by a processor and enabling the processor to perform any of the methods in this invention.

[0066] Those skilled in the art will understand that the modules or steps of the present invention described above can be implemented using general-purpose computer devices. Optionally, they can be implemented using computer-executable program code, thereby allowing them to be stored in a storage device for execution by a computer device, or they can be fabricated as separate integrated circuit modules, or multiple modules or steps can be fabricated as a single integrated circuit module. The present invention is not limited to any particular combination of hardware and software.

[0067] While the specific embodiments of the present invention have been described above in conjunction with the accompanying drawings, this is not intended to limit the scope of protection of the present invention. Those skilled in the art should understand that various modifications or variations that can be made by those skilled in the art without creative effort based on the technical solutions of the present invention are still within the scope of protection of the present invention.

Claims

1. A method for predicting the impact of tunnel mechanical failure downtime based on CART decision tree, characterized in that, include: Collect equipment operation data and construction environment parameters during drill-and-blast tunnel construction, and construct a multi-dimensional feature set; The data in the multidimensional feature set are preprocessed and feature-filtered to establish a fault-downtime correlation feature matrix; Based on the CART decision tree algorithm, a fault-stoppage impact prediction model is constructed using the fault-stoppage correlation feature matrix. The real-time collected equipment status data and construction parameters are input into the prediction model, and the prediction results are dynamically output through decision tree rule matching. The parameters of the prediction model are adaptively adjusted and optimized based on the dynamic changes during the construction phase.

2. The method for predicting the impact of tunnel mechanical failure downtime based on CART decision tree as described in claim 1, characterized in that, The equipment operation data includes vibration, temperature, oil pressure and current data of the rock drilling rig and loader collected by sensors, maintenance records and fault downtime logs.

3. The method for predicting the impact of tunnel mechanical failure downtime based on CART decision tree as described in claim 1, characterized in that, The construction environment parameters include the surrounding rock grade, groundwater distribution, construction stage, temperature and humidity of the work surface, and geological risk level.

4. The method for predicting the impact of tunnel mechanical failure downtime based on CART decision tree as described in claim 1, characterized in that, The data in the multidimensional feature set are preprocessed and feature-filtered to establish a fault-downtime correlation feature matrix, including: Missing data are filled using multiple interpolation, outliers are detected and removed using box plots, and dimensional differences are eliminated by Z-score standardization. The preprocessed data were sorted by Pearson correlation coefficient and random forest feature importance to screen features that are strongly correlated with the impact of shutdown and construct a fault-shutdown correlation feature matrix.

5. The method for predicting the impact of tunnel mechanical failure downtime based on CART decision tree as described in claim 1, characterized in that, The CART decision tree algorithm-based model for predicting the impact of work stoppages using the fault-stoppage correlation feature matrix includes: The dataset is recursively partitioned using the Gini coefficient as the node splitting criterion; the cost complexity pruning is used to balance the model complexity, and the shutdown risk level is defined based on the leaf node sample distribution.

6. The method for predicting the impact of tunnel mechanical failure downtime based on CART decision tree as described in claim 1, characterized in that, The prediction results include the level of work stoppage risk, the expected duration of work stoppage, and the range of economic losses.

7. The method for predicting the impact of tunnel mechanical failure downtime based on CART decision tree as described in claim 1, characterized in that, Adaptive adjustment and optimization of the parameters of the prediction model, including: When the monthly new data and feature weight changes exceed the set threshold, incremental learning is triggered and the segmentation threshold is recalculated for nodes whose feature weight changes exceed the threshold. Retain the subtree structure in the historical decision-making path that has a verification accuracy greater than 85%.

8. A tunnel mechanical failure downtime impact prediction system based on CART decision tree, characterized in that, include: The data acquisition module is configured to: collect equipment operation data and construction environment parameters during drill-and-blast tunnel construction, and construct a multi-dimensional feature set; The feature matrix construction module is configured to: preprocess and filter the data in the multidimensional feature set to establish a fault-downtime correlation feature matrix; The prediction model building module is configured to: construct a prediction model of the impact of failure and shutdown based on the CART decision tree algorithm and using the failure-shutdown correlation feature matrix. The prediction result output module is configured to: input real-time collected equipment status data and construction parameters into the prediction model, and dynamically output the prediction result through decision tree rule matching; The model optimization module is configured to adaptively adjust and optimize the parameters of the prediction model based on the dynamic changes during the construction phase.

9. A computer-readable storage medium having a program stored thereon, characterized in that, When the program is executed by the processor, it implements the steps in the method for predicting the impact of tunnel mechanical failure downtime based on CART decision tree as described in any one of claims 1-7.

10. An electronic device comprising a memory, a processor, and a program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the steps in the method for predicting the impact of tunnel mechanical failure downtime based on CART decision tree as described in any one of claims 1-7.

Images