A mud water shield loop system abnormal condition identification and early warning method

Abstract

The present application relates to the technical field of slurry shield ring flow anomaly identification, and discloses a slurry shield ring flow system abnormal condition identification and early warning method. It aims to solve the technical problems of low efficiency, poor accuracy and incomplete single fault identification in the prior art. The present application comprises the following steps: establishing a slurry shield ring flow system abnormal feature database; constructing a slurry shield ring flow system abnormal condition identification algorithm model; obtaining the current ring flow system parameters in real time and inputting them into the abnormal condition identification algorithm model; executing the corresponding treatment suggestion according to the risk level grading early warning information output by the model; judging whether the early warning information is eliminated, optimizing the ring flow system parameters, adjusting the tunneling speed parameters, and switching to the bypass mode or full system shutdown. The present application realizes comprehensive identification and grading early warning of multiple abnormal conditions, and improves the accuracy and comprehensiveness of abnormal condition identification.

Description

Technical Field

[0001] This invention relates to the field of slurry shield tunneling circulation anomaly identification technology, and in particular to a method for identifying and warning of abnormal conditions in slurry shield tunneling circulation systems. Background Technology

[0002] With the acceleration of urbanization and the continuous expansion of transportation infrastructure construction, slurry shield tunneling technology has been widely used in tunnel construction. Slurry shield tunneling technology has advantages such as high automation, labor saving, fast construction speed, and immunity to weather conditions. However, the monitoring of the operational status and fault early warning of its slurry circulation system still faces many challenges. Currently, common problems in slurry shield tunneling construction include pipe blockage, wear, and pressure loss. These problems can lead to shield machine malfunctions and shutdowns, seriously affecting construction progress and increasing construction costs.

[0003] Traditional shield tunneling monitoring and early warning systems only monitor sensor parameters of the slurry shield and cannot provide accurate early warnings of deformation during construction. Furthermore, while existing methods for detecting pipe blockages are highly efficient, they are also costly. Additionally, due to the limitations of the construction environment, it is difficult to directly observe and study the operational status of the slurry circulation system. These problems lead to safety hazards during shield tunneling, affecting both construction efficiency and safety.

[0004] The existing technology has the following drawbacks:

[0005] 1. Traditional shield tunneling monitoring and early warning systems only monitor the sensor parameters of slurry shields, which cannot quickly capture abnormal events involving multiple coupled parameters, leading to safety hazards during shield tunneling construction and affecting construction efficiency and safety.

[0006] 2. Due to the limitations of the construction environment, it is difficult to directly observe and study the operating status of the slurry circulation system, and it is impossible to accurately judge and warn of various abnormal events in the circulation system.

[0007] 3. In practical applications, existing shield tunneling circulation system blockage early warning methods rely solely on historical data and shield status information, which are insufficient to accurately determine circulation system blockage. They cannot make more accurate judgments and early warnings about the blockage status of the shield tunneling circulation system based on the shield tunneling construction parameters and historical data characteristics of different lines.

[0008] 4. Current methods for identifying abnormal conditions in shield tunneling circulation systems lack a comprehensive identification and graded early warning mechanism for multiple types of abnormal conditions, and cannot provide targeted handling suggestions for different types of abnormal events, thus affecting the timeliness and efficiency of construction.

[0009] To address the aforementioned issues, there is an urgent need for a method and system capable of real-time monitoring, accurate identification, and graded early warning of various abnormal conditions in the slurry shield tunneling circulation system. Summary of the Invention

[0010] In view of the above technical problems, this disclosure provides a method for identifying and warning of abnormal conditions in a slurry shield tunneling circulation system, which solves the technical problems of low efficiency, poor accuracy, and incomplete identification of single faults in the prior art.

[0011] The basic concept of this invention is to establish an abnormal feature database of the slurry shield tunneling circulation system and an abnormal condition identification algorithm model based on machine learning, based on massive historical data from multiple projects. By acquiring the current circulation system parameters in real time during the slurry circulation process and inputting them into the abnormal condition identification algorithm model, the model outputs graded early warning information according to the risk level and provides targeted handling suggestions, such as adjusting circulation system parameters, adjusting tunneling speed and other parameters, switching to bypass mode, and shutting down the entire system.

[0012] It can make full use of massive historical data from multiple projects to establish an abnormal feature database and an abnormal condition identification algorithm model for the slurry shield tunneling circulation system. This enables real-time acquisition of current circulation system parameters and timely identification and early warning of abnormal conditions, thereby improving construction safety and efficiency, reducing energy consumption, and increasing tunneling efficiency.

[0013] According to one aspect of this disclosure, a method for identifying and providing early warning of abnormal conditions in a slurry shield tunneling circulation system is provided, comprising the following steps:

[0014] Step 1: Establish a database of abnormal characteristics of the slurry shield tunneling circulation system;

[0015] Step 2: Construct an algorithm model for identifying abnormal conditions in the slurry shield tunneling circulation system;

[0016] Step 3: Obtain the current circulation system parameters in real time and input them into the abnormal situation identification algorithm model;

[0017] Step 4: Based on the risk level classification and early warning information output by the model, implement the corresponding handling recommendations;

[0018] Step 5: Determine whether the warning message has been cleared. If yes, return to step 3; otherwise, proceed to step 6.

[0019] Step 6: Optimize the circulation system parameters, adjust the tunneling speed parameters, switch to bypass mode or shut down the entire system;

[0020] Step 7: Store the abnormal event and its handling results in the abnormal feature database, update the model training sample set, and periodically trigger incremental learning of the model.

[0021] In some embodiments of this disclosure, step 1 specifically includes the following steps:

[0022] Step 101: Collect massive amounts of historical data from multiple projects, including normal operation data and abnormal operation data. The normal operation data includes equipment operating status, tunneling parameters, and geological information. The abnormal operation data includes construction data, fault causes, and solutions when abnormal events occur. Data types include slurry inlet flow rate, slurry outlet flow rate, pressure at the top of the slurry chamber, liquid level in the air cushion chamber, speed and inlet / outlet pressure of the slurry inlet and outlet pumps, opening and closing status of valves in each pipeline, flow rate of each flushing pipeline, tunneling speed, cutterhead speed, temperature of each motor and flow rate of the sealing cooling water, and pressure of each air system.

[0023] Step 102: Preprocess and clean the historical data, perform feature screening and importance analysis, and extract key features. The key features specifically include classifying equipment by type, tunnel boring machine diameter, pipeline diameter, and tunneling mileage, and classifying personnel by operating experience, working hours, and past tunneling quality. The suitability of the segment division is verified through unsupervised clustering and supervised classification algorithms. The relevant data is then cleaned, visualized, standardized, and discretized for preprocessing.

[0024] Step 103: Establish an abnormal feature database, which includes normal features and abnormal features. The normal features include flow rate, pressure, and temperature parameters when the equipment is running normally, as well as tunneling status and geological status information. The abnormal features include flow rate fluctuations, pressure changes, and temperature anomaly indicators when abnormal events occur, as well as fault type and impact assessment indicators.

[0025] In some embodiments of this disclosure, step 2 specifically includes the following steps:

[0026] Step 201: Construct a first-level anomaly identification model. Based on a deep learning algorithm, a multi-layer end-to-end fully connected neural network structure is adopted. The input layer is a vector with a preset dimension, and the output is a vector representing the probability of each circulation anomaly event. This model is used to make a preliminary judgment on the operating status of the circulation system.

[0027] Step 202: Construct a second-level anomaly identification model. When the risk score output by the first-level anomaly identification model exceeds the preset trigger threshold, it is activated. Multi-source heterogeneous sensor data are fused and analyzed to identify typical anomalies such as pipeline blockage, slurry discharge stagnation, abnormal fluctuations in the liquid level of the air cushion chamber, and abnormal fluctuations in the speed and vibration of the slurry inlet pump and slurry outlet pump.

[0028] Step 203: Train the first-level anomaly detection model and the second-level anomaly detection model using an optimization algorithm, and optimize the weights and thresholds of the models;

[0029] Step 204: Verify the model performance using the test set. If the accuracy is lower than the preset threshold, return to step 201 to retrain the model.

[0030] In some embodiments of this disclosure, step 3 specifically includes the following steps:

[0031] Step 301: Collect real-time data using pressure sensors, flow sensors, and temperature sensors installed on the pipeline to obtain flow, pressure, and temperature parameters;

[0032] Step 302: Obtain the operating status information of the slurry shield tunnel, including slurry inlet flow rate, slurry outlet flow rate, pressure at the top of the slurry chamber, liquid level in the air cushion chamber, speed and inlet / outlet pressure of the slurry inlet and outlet pumps, opening and closing status of valves in each pipeline, and flow data of each flushing pipeline.

[0033] Step 303: Input the real-time data and operating status information into the abnormal situation identification algorithm model.

[0034] In some embodiments of this disclosure, step 4 specifically includes the following steps:

[0035] Step 401: Based on the risk score output by the anomaly identification model, construct a multi-source information fusion diagnostic model based on graph neural network and self-attention mechanism to achieve comprehensive identification of anomalies under multiple operating conditions and classify warning levels: low risk, medium risk, and high risk.

[0036] Step 402: When the risk score exceeds the preset threshold, generate corresponding early warning information, including the fault type, risk level and recommended handling measures;

[0037] Step 403: Based on the warning level, provide specific handling suggestions. When the risk is low, adjust the parameter range of a single circulation system; when the risk is medium, adjust the parameters of multiple circulation systems and reduce the tunneling speed; when the risk is high, switch to bypass mode or shut down the entire system.

[0038] In some embodiments of this disclosure, step 6 specifically includes the following steps:

[0039] Step 601: Adjust the operating parameters of the circulation system according to the cause of the abnormality, including the speed of the feed pump and discharge pump, the main feed flow rate, the main discharge flow rate, and the flow rate of the flushing pipeline;

[0040] Step 602: Optimize the tunneling module parameters, including tunneling speed and cutterhead rotation speed;

[0041] Step 603: Determine the switching strategy based on the fault type: When the fault type is pipeline blockage or slurry discharge stagnation, switch to bypass mode to flush the pipeline and then gradually resume tunneling; when the fault type is mechanical failure of the slurry pump or slurry discharge pump, execute a full system shutdown and determine the shutdown duration.

[0042] The beneficial effects of this invention are as follows:

[0043] It enables comprehensive identification and graded early warning of various abnormal conditions in the slurry shield tunneling circulation system, overcoming the limitations of traditional monitoring systems that only monitor a single parameter, and improving the accuracy and comprehensiveness of abnormal condition identification.

[0044] A database of abnormal features of circulation systems based on massive historical data from multiple projects was constructed, and multi-condition comprehensive identification was performed by combining real-time operating parameters.

[0045] Based on deep learning algorithms, it automatically generates hierarchical early warning information and integrates intelligent handling suggestions. The anomaly identification model has the ability to adaptively adjust across projects and can flexibly adapt to different construction routes.

[0046] 1. This invention establishes an abnormal feature database and an abnormal condition identification algorithm model for the slurry shield tunneling circulation system, thereby achieving comprehensive identification and hierarchical early warning of multiple abnormal conditions. This overcomes the limitation of traditional monitoring systems that only monitor a single parameter, and improves the accuracy and comprehensiveness of abnormal condition identification.

[0047] 2. This invention utilizes massive historical data from multiple projects, combined with real-time acquired circulation system parameters, to more accurately determine and identify the operating status of the circulation system, effectively solving the problem in existing technologies where relying solely on historical data and shield tunneling status information is insufficient to accurately determine anomalies;

[0048] 3. This invention achieves comprehensive monitoring and early warning of complex and ever-changing construction environments by automatically generating early warning information through an anomaly identification algorithm model and combining it with intelligent handling suggestions, thereby improving the safety and efficiency of construction.

[0049] 4. The abnormal condition identification and early warning method of the present invention can monitor and identify abnormal conditions of the circulation system in real time, providing construction personnel with timely early warning information and targeted handling suggestions, effectively avoiding shield tunneling failures and shutdowns caused by abnormalities, and improving construction progress and efficiency.

[0050] 5. The abnormal feature database and identification algorithm model established by this invention can be flexibly adjusted and optimized according to the construction parameters and characteristics of different lines, and has strong adaptability and practicality, which can meet the needs of different construction projects. Attached Figure Description

[0051] Figure 1 A flowchart illustrating the workflow for identifying and providing early warning of abnormal conditions in a slurry shield tunneling circulation system. Detailed Implementation

[0052] The preferred embodiments of the present invention will be described below with reference to the accompanying drawings. It should be understood that the preferred embodiments described herein are for illustration and explanation only and are not intended to limit the present invention.

[0053] Example 1

[0054] This example discloses a method for identifying and warning of abnormal conditions in a slurry shield tunneling circulation system. (See also...) Figure 1 It includes the following steps:

[0055] Step 1: Establish a database of abnormal characteristics of the slurry shield tunneling circulation system;

[0056] Step 2: Construct an algorithm model for identifying abnormal conditions in the slurry shield tunneling circulation system;

[0057] Step 3: Obtain the current circulation system parameters in real time and input them into the abnormal situation identification algorithm model;

[0058] Step 4: Based on the risk level classification and early warning information output by the model, implement the corresponding handling recommendations;

[0059] Step 5: Determine whether the warning message has been cleared. If yes, return to step 3; otherwise, proceed to step 6.

[0060] Step 6: Optimize the circulation system parameters, adjust the tunneling speed parameters, switch to bypass mode or shut down the entire system;

[0061] Step 7: Store the abnormal event and its handling results in the abnormal feature database, update the model training sample set, and periodically trigger incremental learning of the model.

[0062] Step 1 specifically includes the following steps:

[0063] Step 101: Collect massive amounts of historical data from multiple projects, including normal operation data and abnormal operation data. Normal operation data includes equipment operating status, tunneling parameters, and geological information; abnormal operation data includes construction data, fault causes, and solutions when abnormal events occur; data types include slurry inlet flow rate, slurry outlet flow rate, pressure at the top of the slurry chamber, liquid level in the air cushion chamber, speed and inlet / outlet pressure of the slurry inlet and outlet pumps, opening and closing status of valves in each pipeline, flow rate of each flushing pipeline, tunneling speed, cutterhead speed, temperature of each motor and flow rate of sealing cooling water flushing, and pressure of each air system.

[0064] Step 102: Preprocess and clean the historical data, perform feature screening and importance analysis, and extract key features. Key features include classifying equipment by type, tunnel boring machine diameter, pipeline diameter, and tunneling mileage, and classifying personnel by operating experience, working hours, and past tunneling quality. The suitability of the segment division is verified through unsupervised clustering and supervised classification algorithms. The relevant data is cleaned, visualized, standardized, and discretized for data preprocessing.

[0065] Step 103: Establish an abnormal feature database, which includes normal features and abnormal features. Normal features include flow rate, pressure, and temperature parameters when the equipment is running normally, as well as tunneling status and geological status information. Abnormal features include flow rate fluctuations, pressure changes, and temperature anomalies when abnormal events occur, as well as assessment indicators of fault type and impact degree.

[0066] Step 2 specifically includes the following steps:

[0067] Step 201: Construct a first-level anomaly detection model. Based on a deep learning algorithm, a multi-layer end-to-end fully connected neural network structure is adopted. The input layer length is a vector of a preset dimension, and the output is a vector representing the probability of each circulation anomaly event. This model is used to make a preliminary judgment on the operating status of the circulation system. The input layer length is 180, and the output is a 10-dimensional vector. The network structure of the model is designed, and the number of hidden layers is determined to be 10, with 256 hidden layer nodes.

[0068] Step 202: Construct a second-level anomaly identification model. When the risk score output by the first-level anomaly identification model exceeds the preset trigger threshold, it is activated. Multi-source heterogeneous sensor data are fused and analyzed to identify typical anomalies such as pipeline blockage, slurry discharge stagnation, abnormal fluctuations in the liquid level of the air cushion chamber, and abnormal fluctuations in the speed and vibration of the slurry inlet pump and slurry outlet pump.

[0069] Step 203: Train the first-level and second-level anomaly detection models using the Adam optimization algorithm, optimizing the model weights and thresholds; the learning rate is 0.001, the batch size is 64, and the training is performed 1000 times.

[0070] Step 204: Verify the model performance using the test set. If the accuracy is below 95%, return to step 201 to retrain the model.

[0071] Step 3 specifically includes the following steps:

[0072] Step 301: Collect real-time data using pressure sensors, flow sensors, and temperature sensors installed on the pipeline to obtain flow, pressure, and temperature parameters;

[0073] Step 302: Obtain the operating status information of the slurry shield tunnel, including slurry inlet flow rate, slurry outlet flow rate, pressure at the top of the slurry chamber, liquid level in the air cushion chamber, speed and inlet / outlet pressure of the slurry inlet and outlet pumps, opening and closing status of valves in each pipeline, and flow data of each flushing pipeline.

[0074] Step 303: Input the real-time data and running status information into the anomaly identification algorithm model in JSON format.

[0075] Step 4 specifically includes the following steps:

[0076] Step 401: Based on the risk score output by the anomaly identification model, construct a multi-source information fusion diagnostic model based on graph neural network and self-attention mechanism to achieve comprehensive identification of anomalies under multiple operating conditions and classify warning levels: low risk, medium risk, and high risk.

[0077] Step 402: When the risk score exceeds 0.8, generate corresponding early warning information, including the fault type, risk level, and recommended handling measures;

[0078] Step 403: Based on the warning level, provide specific handling suggestions. When the risk is low, adjust the parameter range of a single circulation system; when the risk is medium, adjust the parameters of multiple circulation systems and reduce the tunneling speed; when the risk is high, switch to bypass mode or shut down the entire system.

[0079] Step 6 specifically includes the following steps:

[0080] Step 601: Adjust the operating parameters of the circulation system according to the cause of the abnormality, including the speed of the feed pump and discharge pump, the main feed flow rate, the main discharge flow rate, and the flow rate of the flushing pipeline;

[0081] Step 602: Optimize the tunneling module parameters, including tunneling speed and cutterhead rotation speed; the adjustment ranges are 0.5-10.0 mm / min and 50-200 m³ / h, respectively.

[0082] Step 603: Determine the switching strategy based on the fault type: When the fault type is pipeline blockage or slurry discharge stagnation, switch to bypass mode to flush the pipeline and then gradually resume tunneling; when the fault type is mechanical failure of the slurry pump or slurry discharge pump, execute a full system shutdown and determine the shutdown duration.

[0083] Although some preferred embodiments of the invention have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments as well as all changes and modifications falling within the scope of the invention.

[0084] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this application and their equivalents, this invention also intends to include these modifications and variations.

Claims

1. A method for identifying and warning of abnormal conditions in a slurry shield tunneling circulation system, characterized in that, Includes the following steps: Step 1: Establish a database of abnormal characteristics of the slurry shield tunneling circulation system; Step 2: Construct an algorithm model for identifying abnormal conditions in the slurry shield tunneling circulation system; Step 3: Obtain the current circulation system parameters in real time and input them into the abnormal situation identification algorithm model; Step 4: Based on the risk level classification and early warning information output by the model, implement the corresponding handling recommendations; Step 5: Determine whether the warning message has been cleared. If yes, return to step 3; otherwise, proceed to step 6. Step 6: Optimize the circulation system parameters, adjust the tunneling speed parameters, switch to bypass mode or shut down the entire system; Step 7: Store the abnormal event and its handling results in the abnormal feature database, update the model training sample set, and periodically trigger incremental learning of the model.

2. The method for identifying and warning of abnormal conditions in a slurry shield tunneling circulation system as described in claim 1, characterized in that: Step 1 specifically includes the following steps: Step 101: Collect massive amounts of historical data from multiple projects, including normal operation data and abnormal operation data. The normal operation data includes equipment operating status, tunneling parameters, and geological information. The abnormal operation data includes construction data, fault causes, and solutions when abnormal events occur. Data types include slurry inlet flow rate, slurry outlet flow rate, pressure at the top of the slurry chamber, liquid level in the air cushion chamber, speed and inlet / outlet pressure of the slurry inlet and outlet pumps, opening and closing status of valves in each pipeline, flow rate of each flushing pipeline, tunneling speed, cutterhead speed, temperature of each motor and flow rate of the sealing cooling water, and pressure of each air system. Step 102: Preprocess and clean the historical data, perform feature screening and importance analysis, and extract key features. The key features specifically include classifying equipment by type, tunnel boring machine diameter, pipeline diameter, and tunneling mileage, and classifying personnel by operating experience, working hours, and past tunneling quality. The suitability of the segment division is verified through unsupervised clustering and supervised classification algorithms. The relevant data is then cleaned, visualized, standardized, and discretized for preprocessing. Step 103: Establish an abnormal feature database, which includes normal features and abnormal features. The normal features include flow rate, pressure, and temperature parameters when the equipment is running normally, as well as tunneling status and geological status information. The abnormal features include flow rate fluctuations, pressure changes, and temperature anomaly indicators when abnormal events occur, as well as fault type and impact assessment indicators.

3. The method for identifying and warning of abnormal conditions in a slurry shield tunneling circulation system as described in claim 1, characterized in that: Step 2 specifically includes the following steps: Step 201: Construct a first-level anomaly identification model. Based on a deep learning algorithm, a multi-layer end-to-end fully connected neural network structure is adopted. The input layer is a vector with a preset dimension, and the output is a vector representing the probability of each circulation anomaly event. This model is used to make a preliminary judgment on the operating status of the circulation system. Step 202: Construct a second-level anomaly identification model. When the risk score output by the first-level anomaly identification model exceeds the preset trigger threshold, it is activated. Multi-source heterogeneous sensor data are fused and analyzed to identify typical anomalies such as pipeline blockage, slurry discharge stagnation, abnormal fluctuations in the liquid level of the air cushion chamber, and abnormal fluctuations in the speed and vibration of the slurry inlet pump and slurry outlet pump. Step 203: Train the first-level anomaly detection model and the second-level anomaly detection model using an optimization algorithm, and optimize the weights and thresholds of the models; Step 204: Verify the model performance using the test set. If the accuracy is lower than the preset threshold, return to step 201 to retrain the model.

4. The method for identifying and warning of abnormal conditions in a slurry shield tunneling circulation system as described in claim 1, characterized in that: Step 3 specifically includes the following steps: Step 301: Collect real-time data using pressure sensors, flow sensors, and temperature sensors installed on the pipeline to obtain flow, pressure, and temperature parameters; Step 302: Obtain the operating status information of the slurry shield tunnel, including slurry inlet flow rate, slurry outlet flow rate, pressure at the top of the slurry chamber, liquid level in the air cushion chamber, speed and inlet / outlet pressure of the slurry inlet and outlet pumps, opening and closing status of valves in each pipeline, and flow data of each flushing pipeline. Step 303: Input the real-time data and operating status information into the abnormal situation identification algorithm model.

5. The method for identifying and warning of abnormal conditions in a slurry shield tunneling circulation system as described in claim 1, characterized in that: Step 4 specifically includes the following steps: Step 401: Based on the risk score output by the anomaly identification model, construct a multi-source information fusion diagnostic model based on graph neural network and self-attention mechanism to realize the comprehensive identification of anomalies under multiple working conditions and classify the warning levels: low risk, medium risk, and high risk. Step 402: When the risk score exceeds the preset threshold, generate corresponding early warning information, including the fault type, risk level and recommended handling measures; Step 403: Based on the warning level, provide specific handling suggestions. When the risk is low, adjust the parameter range of a single circulation system; when the risk is medium, adjust the parameters of multiple circulation systems and reduce the tunneling speed; when the risk is high, switch to bypass mode or shut down the entire system.

6. The method for identifying and warning of abnormal conditions in a slurry shield tunneling circulation system as described in claim 1, characterized in that: Step 6 specifically includes the following steps: Step 601: Adjust the operating parameters of the circulation system according to the cause of the abnormality, including the speed of the feed pump and discharge pump, the main feed flow rate, the main discharge flow rate, and the flow rate of the flushing pipeline; Step 602: Optimize the tunneling module parameters, including tunneling speed and cutterhead rotation speed; Step 603: Determine the switching strategy based on the fault type: When the fault type is pipeline blockage or slurry discharge stagnation, switch to bypass mode to flush the pipeline and then gradually resume tunneling; when the fault type is mechanical failure of the slurry pump or slurry discharge pump, execute a full system shutdown and determine the shutdown duration.

Images