A method for predicting gushing when a shield tunnel passes through a water-rich gravel layer

Abstract

The present invention relates to the technical field of shield construction, and specifically to a method for predicting gushing when a shield passes through a water-rich gravel layer, including data collection and preprocessing, including geological data collection, construction parameter collection, equipment performance data collection, and groundwater dynamic monitoring. The present invention constructs a gushing risk prediction system through multi-source data fusion and real-time monitoring, which significantly improves the safety and efficiency of shield construction passing through a water-rich gravel layer.

Description

Technical Field

[0001] The invention relates to the technical field of shield construction, in particular to a method for predicting gushing that occurs when a shield passes through a water-rich gravel layer. Background Art

[0002] Shield construction technology has been widely used in urban rail transit construction due to its advantages such as small impact on the surrounding environment. However, when passing through complex geological conditions such as water-rich gravel layers, it often faces many difficulties, and gushing is one of them, which seriously affects the construction progress and project safety.

[0003] The pores between particles in the water-rich gravel layer are large and the permeability is strong, so groundwater can easily flow through it. When a shield machine excavates in this stratum, groundwater will quickly flow into the soil bin and the inside of the shield machine, increasing the risk of eruptions. The cohesion between particles in the gravel layer is small, and the self-stabilizing ability of the soil is poor. During the shield excavation process, the excavation face is prone to instability, leading to soil collapse and a large influx of groundwater, which in turn triggers eruptions.

[0004] When a gushing occurs, a large amount of groundwater and debris will be ejected from the screw machine outlet or other parts, resulting in the inability to carry out normal construction. It will take a lot of time to clean up and deal with it, which will seriously affect the construction progress. At present, there is no complete gushing prediction system based on multi-source data fusion and multi-factor comprehensive analysis. It is unable to fully consider the impact of geological conditions, construction parameters, equipment performance and other factors on the gushing, resulting in low accuracy and reliability of the prediction results. Summary of the invention

[0005] In order to overcome the above technical problems, the purpose of the present invention is to provide a method for predicting gushing when a shield machine passes through a water-rich gravel layer, so as to solve the problem of the lack of an effective method for predicting gushing when a shield machine passes through a water-rich gravel layer mentioned in the above background technology.

[0006] To achieve the above-mentioned purpose, the present invention provides the following technical solutions: a method for predicting gushing when a shield tunnel passes through a water-rich gravel layer, comprising the following steps: S1: data collection and preprocessing; S2: gushing risk model construction; S3: real-time monitoring and early warning of gushing.

[0007] Preferably, the data collection and preprocessing in S1 include geological data collection, construction parameter collection, equipment performance data collection and groundwater dynamic monitoring. The address data collection utilizes geological exploration and three-dimensional modeling to obtain the particle distribution and permeability parameters of the gravel layer. The construction parameter collection collects key parameters such as shield advancement speed, excavation face pressure and soil bin pressure. The equipment performance data collection obtains the cutter head speed and torque shield machine performance indicators. The groundwater dynamic monitoring installs a piezometer to record the changes in groundwater level and flow rate in real time.

[0008] Preferably, the construction of the eruption risk model includes training data, input variables and output variables, the training data includes historical construction data and real-time field data, the input variables include current data characteristic data, the current data characteristic data includes formation permeability and formation water content, and the output variables are used to send the eruption data predicted by the eruption risk model to the construction data host.

[0009] Preferably, the real-time monitoring and early warning of gushing includes real-time monitoring, predictive calculation and early warning mechanisms, and the real-time monitoring and early warning of gushing is installed in a construction data host, and the construction data host includes a data processing module and a data storage module.

[0010] Preferably, the data processing module includes data cleaning, feature extraction and parameter optimization, the data cleaning is used to denoise, normalize and screen characteristic variables of construction and geological data collected in real time, the feature extraction is used to identify key indicators related to eruption, and the parameter optimization is used to pass the processed data to the eruption risk model.

[0011] Preferably, the data storage module is used to classify and store historical data and real-time data of the entire construction process.

[0012] Preferably, the early warning mechanism module includes a multi-level early warning function, and the early warning mechanism module includes operations such as pausing advancement, adjusting excavation parameters, grouting reinforcement or switching shield mode, while recording the early warning process and emergency treatment measures.

[0013] Preferably, the construction data host has self-learning and self-adaptation capabilities, performs iterative analysis on real-time data and historical data through artificial intelligence algorithms, automatically adjusts model weights and parameters, and enhances adaptability and prediction accuracy to eruption risks under different geological conditions.

[0014] Compared with the prior art, the present invention has the following beneficial effects:

[0015] 1. This method comprehensively considers geological conditions, construction parameters, equipment performance and other factors, and accurately identifies key indicators related to eruption through multi-source data fusion technology and feature extraction, significantly improving the accuracy of the prediction model for eruption risk and avoiding the deviation that may be caused by prediction from a single data source;

[0016] 2. Through real-time monitoring and multi-level early warning mechanisms, this method can identify the risk of gushing in advance and notify the construction team to take measures such as adjusting excavation parameters, reducing advancement speed, and grouting reinforcement, effectively preventing the occurrence of gushing accidents and ensuring construction safety;

[0017] 3. By predicting the risk of gushing in advance and providing targeted adjustment suggestions, this method avoids construction interruptions and cleanup work caused by gushing accidents, reduces construction time loss, and significantly improves construction efficiency;

[0018] 4. This method adopts a modular design and is suitable for different geological conditions and construction environments. It can provide scientific and effective prediction and decision-making support for shield construction in complex strata such as water-rich gravel layers. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] Figure 1 It is a topological schematic diagram of the gushing prediction method of the present invention;

[0020] Figure 2 A topological schematic diagram of the eruption risk model of the present invention;

[0021] Figure 3 It is a topological schematic diagram of the construction panel of the present invention;

[0022] Figure 4 It is a topological schematic diagram of the data processing module of the present invention. DETAILED DESCRIPTION

[0023] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0024] See also Figure 1-4 , an embodiment provided by the present invention: a method for predicting gushing when a shield machine passes through a water-rich gravel layer, comprising the following steps:

[0025] S1: data collection and preprocessing;

[0026] S2: Construction of the eruption risk model;

[0027] S3: Real-time monitoring and early warning of eruptions.

[0028] Data collection and preprocessing in S1 include geological data collection, construction parameter collection, equipment performance data collection and groundwater dynamic monitoring. Address data collection uses geological exploration and three-dimensional modeling to obtain the distribution of gravel particles and permeability parameters. Construction parameter collection collects key parameters such as shield advancement speed, excavation surface pressure, and soil bin pressure. Equipment performance data collection obtains cutterhead speed and torque shield machine performance indicators. Groundwater dynamic monitoring installs piezometers to record groundwater level and flow rate changes in real time. The detailed structure of the underground gravel layer is obtained through high-precision drilling and geological radar detection, and the three-dimensional geological modeling technology is combined to accurately describe the characteristics of stratum changes, and further clarify the particle shape, arrangement and porosity of the gravel layer. In addition, the permeability test is used to evaluate the permeability coefficient and hydrological characteristics of the stratum, providing basic support for subsequent risk analysis. During the shield construction process, real-time monitoring devices are deployed to automatically record dynamic change parameters such as shield advancement speed, excavation surface pressure, and soil bin pressure. At the same time, the changing patterns of mud fluidity and mud-water pressure in the soil bin are collected to provide more dimensional input data for the eruption risk prediction model; multiple performance indicators of the shield machine operation are obtained in real time, including cutter head speed, cutter head torque, main drive motor power, and bearing temperature. In addition, historical records of equipment operation status are added to analyze the interaction between equipment performance and changes in geological conditions; multi-point piezometers and water level gauges are deployed in the construction area to continuously monitor groundwater level changes and flow rate dynamics. Combined with groundwater chemical analysis, the potential impact of water quality on soil stability is evaluated. In addition, by introducing groundwater flow simulation technology, the interactive effect between groundwater flow direction and shield advancement path is predicted to provide a scientific basis for construction parameter adjustment; through data fusion technology, the collected multi-source heterogeneous data is standardized, including denoising, missing value filling, and data consistency check. At the same time, statistical analysis and machine learning methods are combined to reduce the dimension and extract features of the data, screen out the key variables that have the greatest impact on the eruption risk, and lay the foundation for model construction.

[0029] The construction of the eruption risk model includes training data, input variables and output variables. The training data includes historical construction data and real-time field data. The input variables include current data characteristic data. The current data characteristic data includes formation permeability and formation water content. The output variables are used to send the eruption data predicted by the eruption risk model to the construction data host.

[0030] The real-time monitoring and early warning of gushing includes real-time monitoring, prediction calculation and early warning mechanism. The real-time monitoring and early warning of gushing is installed in the construction data host. The construction data host contains data processing module and data storage module. The real-time monitoring and early warning system of gushing takes the construction data host as the core, and realizes dynamic control of the risk of gushing during the construction process through highly integrated data collection, analysis and early warning functions. Real-time monitoring relies on high-precision sensors and Internet of Things technology to transmit the geological parameters, construction parameters and equipment performance data obtained during the shield advancement process to the host in real time to ensure the comprehensiveness and timeliness of data collection; the monitoring content covers multiple key indicators such as groundwater level fluctuations, excavation surface pressure changes and shield machine operation status, providing basic data support for prediction calculation. Predictive calculation uses an efficient and optimized gushing risk model to quickly analyze the real-time transmitted data, dynamically evaluate the gushing risk level based on the current geological and construction conditions, and judge the possible risk mode in combination with historical data trends. The early warning mechanism is embedded in the construction data host, providing accurate risk feedback in a multi-level early warning manner. When the predicted calculation results reach the set threshold, the system automatically triggers the early warning signal, including sound, visual alarms and operation suggestions. The construction data host can also generate targeted disposal suggestions in real time, such as suspending excavation, adjusting excavation parameters or initiating reinforcement measures. At the same time, it records detailed information on each early warning and response for subsequent construction optimization and decision support.

[0031] The data processing module includes data cleaning, feature extraction and parameter optimization. Data cleaning is used to denoise, normalize and screen feature variables of construction and geological data collected in real time. Feature extraction is used to identify key indicators related to eruption. Parameter optimization is used to pass the processed data to the eruption risk model. The data processing module integrates a variety of technologies to achieve comprehensive processing and optimization of raw data. First, data cleaning improves data quality through means such as denoising, outlier detection and data consistency verification. Denoising uses wavelet transform or Kalman filtering technology to eliminate noise caused by sensor interference. Outlier detection uses the isolation forest algorithm to screen out error data. At the same time, time and space synchronization verification is performed on data from different sources to ensure data consistency. Secondly, feature extraction identifies key indicators that have a significant impact on eruption risk prediction through association analysis, nonlinear feature mining and dynamic feature modeling. For example, Pearson correlation and mutual information are used to analyze variable relationships, and principal component analysis or time series modeling methods are used to extract and simplify dynamic features. Finally, parameter optimization further improves data utilization efficiency through feature weight allocation, parameter adjustment and multi-model input fusion. The model interpretation method is used to weight features and focus on variables with greater influence. The construction parameters are dynamically adjusted to optimize the data input format. At the same time, integrated learning technology is combined to enhance the overall performance of model prediction, ensuring that the eruption risk model can respond quickly and accurately to dynamic changes in the construction process.

[0032] The data storage module is used to classify and store historical data and real-time data of the entire construction process. The data storage module is an important part of the construction data host. It is specifically responsible for efficient and scientific classification and storage of historical data and real-time data generated during the entire construction process, providing a data basis for subsequent analysis and application. Through an intelligent data management system, this module stores multi-dimensional data such as geological conditions, construction parameters, equipment operating status, groundwater dynamics, etc. collected during the construction process according to attributes and uses, and supports data access and dynamic expansion in multiple formats to ensure data integrity and consistency. Real-time data is instantly written to the storage system through a high-speed data transmission channel and automatically associated with historical data so that the analysis process can quickly call the latest information; historical data is classified and archived according to labels such as time, geological area, and equipment model, which is convenient for retrospective analysis and model optimization. The data storage module is also equipped with data compression and encryption functions, which saves storage space while ensuring the security of construction data, and can set different access rights to ensure that the use of data is limited to the authorized scope. In addition, the module also has a regular backup function, which can quickly restore when data is lost or the system is abnormal, ensuring the continuous availability of construction records and analysis results, and providing strong support for risk prediction and model update during the construction process.

[0033] The early warning mechanism module includes multi-level early warning functions. The early warning mechanism module includes operations such as pausing advancement, adjusting excavation parameters, grouting reinforcement or switching shield mode, and records the early warning process and emergency response measures.

[0034] The construction data host has self-learning and adaptive capabilities. It uses artificial intelligence algorithms to iteratively analyze real-time data and historical data, automatically adjust model weights and parameters, and enhance adaptability and prediction accuracy to eruption risks under different geological conditions.

[0035] Working principle: First, in the early stage of construction, by deploying sensor networks and related monitoring equipment, geological exploration data are collected, including geological parameters such as the particle distribution, permeability and moisture content of the gravel layer, as well as construction parameters such as shield advancement speed, excavation face pressure and soil bin pressure, and equipment performance data such as cutter head speed and cutter head torque. In addition, by installing equipment such as piezometers, changes in groundwater level and flow rate are monitored in real time. After unified normalization, all data are transmitted to the construction data host for cleaning, removing outliers and noise, and screening out key characteristic variables.

[0036] After data processing is completed, the system uses the extracted geological features and construction parameters as input variables and inputs them into the eruption risk model for calculation. The model is trained based on historical construction data and real-time on-site data, and uses machine learning algorithms to predict the level of eruption risk. Based on the calculation results, the system will display the real-time level of the current eruption risk on the construction data host and send the prediction results to the construction panel.

[0037] When the risk level of eruption is low, the system allows normal construction and only records relevant data for analysis. When the risk level of eruption reaches medium or high, the system will trigger the early warning mechanism and notify the construction team in real time. The early warning mechanism provides targeted operational suggestions, such as reducing the advancement speed, adjusting the excavation face pressure or grouting reinforcement. If the risk of eruption continues to increase, the system will recommend suspending the advancement and implementing more stringent emergency treatment measures.

[0038] During the entire construction process, the system continuously transmits the real-time collected data to the eruption risk model, which uses the self-learning function to dynamically adjust the weights and parameters to adapt to the real-time changing geological conditions and construction parameters. All construction data and eruption risk prediction results will be classified and stored in the data storage module as a reference for subsequent construction optimization and model updates.

[0039] Ultimately, through real-time monitoring and prediction, early warning and adjustment, and the system's adaptive capabilities, the construction team was able to respond to possible gushing risks in advance and ensure the safety and efficiency of shield tunneling in water-rich gravel layers.

[0040] It will be apparent to those skilled in the art that the invention is not limited to the details of the exemplary embodiments described above and that the invention can be implemented in other specific forms without departing from the spirit or essential features of the invention. Therefore, the embodiments should be considered exemplary and non-limiting in all respects, and the scope of the invention is defined by the appended claims rather than the foregoing description, and it is intended that all variations falling within the meaning and scope of the equivalent elements of the claims be included in the invention. Any reference numeral in a claim should not be considered as limiting the claim to which it relates.

Claims

1. A method for predicting gushing when a shield machine passes through a water-rich gravel layer, characterized in that: The following steps are involved: S1: data collection and preprocessing; S2: Construction of the eruption risk model; S3: Real-time monitoring and early warning of eruptions.

2. The method for predicting gushing when a shield machine passes through a water-rich gravel layer according to claim 1 is characterized in that: The data collection and preprocessing in S1 include geological data collection, construction parameter collection, equipment performance data collection and groundwater dynamic monitoring. The address data collection uses geological exploration and three-dimensional modeling to obtain the particle distribution and permeability parameters of the gravel layer. The construction parameter collection collects key parameters such as shield advancement speed, excavation face pressure, and soil bin pressure. The equipment performance data collection obtains the cutter head speed and torque shield machine performance indicators. The groundwater dynamic monitoring installs a piezometer to record the groundwater level and flow rate changes in real time.

3. The method for predicting gushing when a shield machine passes through a water-rich gravel layer according to claim 1 is characterized by: The construction of the eruption risk model includes training data, input variables and output variables. The training data includes historical construction data and real-time field data. The input variables include current data characteristic data, and the current data characteristic data includes formation permeability and formation water content. The output variables are used to send the eruption data predicted by the eruption risk model to the construction data host.

4. The method for predicting gushing when a shield machine passes through a water-rich gravel layer according to claim 1 is characterized by: The real-time monitoring and early warning of gushing includes real-time monitoring, prediction calculation and early warning mechanism. The real-time monitoring and early warning of gushing is installed in a construction data host. The construction data host includes a data processing module and a data storage module.

5. The method for predicting gushing when a shield machine passes through a water-rich gravel layer according to claim 4 is characterized in that: The data processing module includes data cleaning, feature extraction and parameter optimization. The data cleaning is used to denoise, normalize and screen characteristic variables of construction and geological data collected in real time. The feature extraction is used to identify key indicators related to eruption. The parameter optimization is used to pass the processed data to the eruption risk model.

6. The method for predicting gushing when a shield machine passes through a water-rich gravel layer according to claim 4 is characterized in that: The data storage module is used to classify and store historical data and real-time data of the entire construction process.

7. The method for predicting gushing when a shield machine passes through a water-rich gravel layer according to claim 4 is characterized by: The early warning mechanism module includes a multi-level early warning function, and the early warning mechanism module includes operations such as pausing advancement, adjusting excavation parameters, grouting reinforcement or switching shield mode, and records the early warning process and emergency treatment measures.

8. The method for predicting gushing when a shield machine passes through a water-rich gravel layer according to claim 4 is characterized by: The construction data host has self-learning and self-adaptation capabilities. It uses artificial intelligence algorithms to iteratively analyze real-time data and historical data, automatically adjust model weights and parameters, and enhance adaptability and prediction accuracy to eruption risks under different geological conditions.

Images