Intelligent control system and process for subway station deep foundation pit dewatering

Abstract

This invention relates to the field of deep foundation pit engineering technology for subway stations, specifically an intelligent control system and process for dewatering deep foundation pits in subway stations. The system includes: a multi-source fusion sensing module, an edge computing and transmission module, an AI intelligent decision-making and prediction module, a hierarchical intelligent control module, a multi-level risk early warning module, a BIM visualization and remote management module, and a data storage and self-learning optimization module. This invention integrates multi-dimensional data such as water level, pore water pressure, settlement, displacement, flow rate, energy consumption, and pump operating conditions to achieve full-coverage coverage of the entire dewatering process. The monitoring accuracy reaches ±1mm, completely eliminating blind spots in traditional monitoring. It employs local real-time processing at the edge, with a command response latency of ≤100ms, far superior to traditional cloud-based processing modes, ensuring rapid risk management.

Description

Technical Field

[0001] This invention relates to the field of deep foundation pit engineering technology for subways, specifically to an intelligent control system and process for dewatering deep foundation pits in subway stations. Background Technology

[0002] Deep foundation pits for subway stations are mostly located in the core urban areas, and are generally characterized by large excavation depths, complex geological conditions, sensitive surrounding environments, and abundant groundwater. Dewatering is a core control point for the safety of foundation pit excavation. Traditional deep foundation pit dewatering methods rely on manual inspections, experience-based pump start / stop, fixed threshold control, and single-point monitoring, which have irreversible technical defects and can no longer meet the requirements of safe, green, and intelligent construction in modern subway engineering.

[0003] Monitoring methods are limited and data is isolated and lacks integration.

[0004] Monitoring based solely on a simple water level gauge inside the well, without simultaneously integrating multi-dimensional data such as pore water pressure, surrounding settlement, retaining wall displacement, flow rate, energy consumption, and pump operating conditions, cannot reflect the coupled response of precipitation, surrounding rock, and environment, and is prone to judgment bias and blind spots in control.

[0005] The control method is crude and has extremely poor precision.

[0006] Using manual start / stop of water pumps or fixed water level threshold control, without variable frequency stepless adjustment, precise zoning precipitation, and dynamic threshold adaptation functions, can easily lead to safety accidents such as excessive precipitation causing surrounding subsidence and cracking, or insufficient precipitation causing water to gush, surge, or rise at the bottom of the pit.

[0007] Lacking the ability to predict risks, passively responding to them

[0008] It can only monitor the current water level and cannot make advance predictions on future water level changes, subsidence trends, and water inrush risks. The early warning is delayed, the response time is insufficient, and the safety hazards are enormous.

[0009] The system is fragmented and lacks coordination and closed-loop control mechanisms.

[0010] The precipitation system, monitoring system, and construction system are independent of each other and have not formed a closed loop of monitoring-decision-control-feedback-optimization. Abnormal working conditions cannot be automatically linked and controlled, and rely on manual handling, with a response delay of ≥30 minutes.

[0011] High energy consumption, high operation and maintenance costs, and low level of intelligence.

[0012] The water pumps operate at full load for extended periods, lacking energy-saving control, fault self-diagnosis, and unattended operation functions. Manual inspections are intensive and inefficient, failing to meet the standardized and intelligent construction requirements of subway projects.

[0013] Poor adaptability to complex strata, lack of hierarchical control strategies

[0014] Faced with diverse geological formations such as soft soil, water-rich soil, sand and gravel, and karst, the lack of zoned, graded, and phased intelligent precipitation strategies and the uniform control model can easily lead to local precipitation imbalances and excessive environmental deformation.

[0015] While existing patents and technologies have proposed some ideas for intelligent control of precipitation, none of them have achieved a complete technical system of multi-source sensing, AI coupled prediction, hierarchical frequency conversion control, precipitation recharge linkage, and closed-loop self-learning. They cannot truly solve the core pain points of accuracy, safety, and intelligence in deep foundation pit dewatering of subway stations. Therefore, an intelligent control system and process for deep foundation pit dewatering of subway stations is proposed. Summary of the Invention

[0016] In view of this, the present invention provides an intelligent control system and process for dewatering deep foundation pits in subway stations, in order to solve or alleviate the technical problems existing in the prior art, and at least provide a beneficial option.

[0017] The technical solution of the present invention is implemented as follows: an intelligent control system for dewatering deep foundation pits in subway stations, comprising: a multi-source fusion sensing module, an edge computing and transmission module, an AI intelligent decision-making and prediction module, a hierarchical intelligent control module, a multi-level risk early warning module, a BIM visualization and remote management module, and a data storage and self-learning optimization module.

[0018] The output of the multi-source fusion sensing module is connected to the edge computing and transmission module. The output of the edge computing and transmission module is connected to the AI ​​intelligent decision-making and prediction module. The output of the AI ​​intelligent decision-making and prediction module is connected to the hierarchical intelligent control module and the multi-level risk warning module. The hierarchical intelligent control module and the multi-level risk warning module both feed back to the multi-source fusion sensing module to form a closed loop. The BIM visualization and remote management module and the data storage and self-learning optimization module are bidirectionally connected to the above modules.

[0019] More preferably, the multi-source fusion sensing module includes an immersion water level sensor, a pore water pressure sensor, a hydrostatic level, a retaining structure displacement meter, an electromagnetic flow meter, an energy metering module, and a water pump status sensor.

[0020] The monitoring parameters include well water level, pore water pressure, settlement of surrounding surface / buildings, horizontal displacement of retaining structure, water flow rate, pump voltage / current / power / start / stop status, and pipeline pressure. The sampling frequency is adaptively adjustable from 1 to 5 Hz.

[0021] More preferably, the AI ​​intelligent decision-making and prediction module has a built-in LSTM-Attention water level-settlement coupling prediction model, which is used to predict the probability of water level, settlement and sudden surge risk in the next 1 to 24 hours;

[0022] The formula for calculating the dynamic early warning threshold is:

[0023] Dynamic threshold = baseline threshold × formation correction factor × excavation disturbance factor × water environment correction factor × environmental sensitivity factor;

[0024] The benchmark thresholds are determined based on specifications and design values, and each coefficient is dynamically iterated and optimized according to the geological formation, construction, and environment.

[0025] More preferably, the hierarchical intelligent control module includes a frequency converter control cabinet, a frequency converter water pump, an electric regulating valve, an intelligent recharge device, and an emergency drainage pump;

[0026] It supports four modes: zoned precise control, variable frequency stepless control, hierarchical linkage control, and precipitation-reinjection linkage. The pump frequency is continuously adjustable from 0 to 50 Hz.

[0027] More preferably, the multi-level risk warning module is divided into four levels: blue alert, yellow warning, orange alert, and red emergency alert;

[0028] The early warning system simultaneously triggers on-site audio-visual alarms, platform pop-ups, APP push notifications, and SMS notifications, and locates the risk area in three dimensions in the BIM model.

[0029] More preferably, the edge computing and transmission module adopts industrial Ethernet + 5G dual redundant communication, supports network outage resume and local caching, data processing latency ≤50ms, and end-to-end command response latency ≤100ms.

[0030] An intelligent control technology for dewatering deep foundation pits in subway stations includes the following steps:

[0031] S1 Field Survey and Deployment: Complete the deployment and network debugging of precipitation wells, observation wells, sensing equipment, control equipment, and edge gateways;

[0032] S2 system initialization: Input engineering geological and environmental parameters, set benchmark thresholds, and complete AI model pre-training and system integration.

[0033] S3 multi-source data real-time acquisition and spatiotemporal synchronization, acquiring full-element data at a frequency of 1-5Hz;

[0034] S4 edge data preprocessing: cleaning, noise reduction, fusion, extraction of water level rate, settling rate, and risk feature values;

[0035] S5AI Coupled Prediction and Dynamic Decision-Making: Predicting water level settlement, calculating dynamic thresholds, and generating zoned frequency conversion control commands;

[0036] S6-level intelligent precipitation control: Execute corresponding intensity control according to blue alert, yellow warning, orange alert, and red emergency alert;

[0037] S7 Multi-level Risk Warning and Linked Response: Triggers multi-level alarms and automatically executes risk suppression commands;

[0038] S8 Closed-Loop Feedback and Effect Verification: Monitor the control effect, correct deviations, and form a control closed loop;

[0039] S9 data import and self-learning iteration: data archiving, incremental model training, strategy optimization; intelligent well sealing and system shutdown after foundation pit completion.

[0040] More preferably, the hierarchical control strategy in step S6 is:

[0041] Blue alert: Maintain current operating parameters and continue monitoring;

[0042] Yellow alert: Slightly increase the frequency of water pumps, increase the number of pumps turned on, and strengthen monitoring density;

[0043] Orange alert: Rainfall intensity will be significantly increased; regional enhanced rainfall will be activated; and recharge compensation will be initiated if necessary.

[0044] Red Emergency Alert: Full-load rainfall + activation of emergency pump units, stop excavation, seal off dangerous areas, and organize evacuation.

[0045] More preferably, in step S8, the closed-loop feedback adopts a small-amplitude deviation correction mode, with a single adjustment parameter change of ≤10%, to avoid oscillation and instability;

[0046] In step S9, the self-learning module is trained incrementally using actual water level and settlement data as labels to continuously optimize prediction accuracy and dynamic threshold.

[0047] Furthermore, the process is applicable to all types of deep foundation pits for subway stations, including silty soil, gravel, soft rock, hard rock, soft upper and hard lower, and water-rich fractured soil. It supports unattended intelligent operation throughout the construction process, with water level control accuracy of ±5mm, improved surrounding settlement control accuracy by more than 60%, and energy saving rate of ≥30%.

[0048] The embodiments of the present invention have the following advantages due to the adoption of the above technical solutions:

[0049] I. This invention integrates multi-dimensional data such as water level, pore water pressure, settlement, displacement, flow rate, energy consumption, and pump operating conditions to achieve full coverage of the entire precipitation process, with a monitoring accuracy of ±1mm. It completely eliminates blind spots in traditional monitoring and adopts local real-time processing at the edge, with a command response delay of ≤100ms, which is far superior to the traditional cloud processing mode, ensuring rapid risk handling.

[0050] Second, the LSTM-Attention coupling model of this invention accurately predicts water level and settlement in the next 24 hours, provides early warning of the risk of sudden surge and settlement exceeding the standard, and changes from passive response to active prevention and control. The dynamic threshold adapts to changes in strata, construction and environment, and the zoned frequency conversion stepless adjustment avoids excessive precipitation and insufficient precipitation, and improves the accuracy of settlement control in the surrounding area by more than 60%.

[0051] Third, this invention features a four-level early warning system (blue, yellow, orange, and red) with automatic linkage and multiple alarm modes. This reduces risk response time by 90%, lowers the accident rate by over 90%, and automatically initiates recharge compensation when the water level is too low or settlement exceeds the standard. This maximizes the protection of surrounding buildings, structures, and pipelines. The fully automatic closed-loop control eliminates the need for manual inspection and monitoring, saves over 30% on energy, reduces labor costs by 80%, and increases construction efficiency by 40%.

[0052] Fourth, the system of this invention continuously learns engineering data and constantly optimizes models and strategies. It is applicable to all types of complex strata, including soft soil, gravel, water-rich soil, fractured rock, and soft upper and hard lower strata. It features three-dimensional visualization, remote monitoring, and automatic reporting, thereby improving the standardization and intelligence of construction management.

[0053] The above overview is for illustrative purposes only and is not intended to be limiting in any way. In addition to the illustrative aspects, embodiments, and features described above, further aspects, embodiments, and features of the invention will become readily apparent from the accompanying drawings and the following detailed description. Attached Figure Description

[0054] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0055] Figure 1 This is a flowchart of the process of the present invention. Detailed Implementation

[0056] In the following description, only certain exemplary embodiments are briefly described. As those skilled in the art will recognize, the described embodiments can be modified in various ways without departing from the spirit or scope of the invention. Therefore, the drawings and description are considered to be exemplary in nature and not restrictive.

[0057] The embodiments of the present invention will now be described in detail with reference to the accompanying drawings.

[0058] like Figure 1As shown, this embodiment of the invention provides an intelligent control system for dewatering deep foundation pits in subway stations, including: a multi-source fusion sensing module, an edge computing and transmission module, an AI intelligent decision-making and prediction module, a hierarchical intelligent control module, a multi-level risk early warning module, a BIM visualization and remote management module, and a data storage and self-learning optimization module.

[0059] The output of the multi-source fusion sensing module is connected to the edge computing and transmission module. The output of the edge computing and transmission module is connected to the AI ​​intelligent decision-making and prediction module. The output of the AI ​​intelligent decision-making and prediction module is connected to the hierarchical intelligent control module and the multi-level risk warning module. The hierarchical intelligent control module and the multi-level risk warning module both feed back to the multi-source fusion sensing module to form a closed loop. The BIM visualization and remote management module and the data storage and self-learning optimization module are bidirectionally connected to the above modules.

[0060] In one embodiment, the multi-source fusion sensing module includes an immersion water level sensor, a pore water pressure sensor, a hydrostatic level, a retaining structure displacement meter, an electromagnetic flow meter, an energy metering module, and a water pump status sensor.

[0061] The monitoring parameters include well water level, pore water pressure, settlement of surrounding surface / buildings, horizontal displacement of retaining structure, water flow rate, pump voltage / current / power / start / stop status, and pipeline pressure. The sampling frequency is adaptively adjustable from 1 to 5 Hz.

[0062] In one embodiment, the AI ​​intelligent decision-making and prediction module incorporates an LSTM-Attention water level-settlement coupling prediction model to predict the probability of water level, settlement, and sudden surge risk in the next 1 to 24 hours.

[0063] The formula for calculating the dynamic early warning threshold is:

[0064] Dynamic threshold = baseline threshold × formation correction factor × excavation disturbance factor × water environment correction factor × environmental sensitivity factor;

[0065] The benchmark thresholds are determined based on specifications and design values, and each coefficient is dynamically iterated and optimized according to the geological formation, construction, and environment.

[0066] In one embodiment, the hierarchical intelligent control module includes a frequency converter control cabinet, a frequency converter water pump, an electric regulating valve, an intelligent recharge device, and an emergency drainage pump.

[0067] It supports four modes: zoned precise control, variable frequency stepless control, hierarchical linkage control, and precipitation-reinjection linkage. The pump frequency is continuously adjustable from 0 to 50 Hz.

[0068] In one embodiment, the multi-level risk warning module is divided into four levels: blue alert, yellow warning, orange alert, and red emergency alert.

[0069] The early warning system simultaneously triggers on-site audio-visual alarms, platform pop-ups, APP push notifications, and SMS notifications, and locates the risk area in three dimensions in the BIM model.

[0070] In one embodiment, the edge computing and transmission module adopts industrial Ethernet + 5G dual redundant communication, supports network outage resume and local caching, data processing latency ≤50ms, and end-to-end command response latency ≤100ms.

[0071] An intelligent control technology for dewatering deep foundation pits in subway stations includes the following steps:

[0072] S1 Field Survey and Deployment: Complete the deployment and network debugging of precipitation wells, observation wells, sensing equipment, control equipment, and edge gateways;

[0073] S2 system initialization: Input engineering geological and environmental parameters, set benchmark thresholds, and complete AI model pre-training and system integration.

[0074] S3 multi-source data real-time acquisition and spatiotemporal synchronization, acquiring full-element data at a frequency of 1-5Hz;

[0075] S4 edge data preprocessing: cleaning, noise reduction, fusion, extraction of water level rate, settling rate, and risk feature values;

[0076] S5AI Coupled Prediction and Dynamic Decision-Making: Predicting water level settlement, calculating dynamic thresholds, and generating zoned frequency conversion control commands;

[0077] S6-level intelligent precipitation control: Execute corresponding intensity control according to blue alert, yellow warning, orange alert, and red emergency alert;

[0078] S7 Multi-level Risk Warning and Linked Response: Triggers multi-level alarms and automatically executes risk suppression commands;

[0079] S8 Closed-Loop Feedback and Effect Verification: Monitor the control effect, correct deviations, and form a control closed loop;

[0080] S9 data import and self-learning iteration: data archiving, incremental model training, strategy optimization; intelligent well sealing and system shutdown after foundation pit completion.

[0081] In one embodiment, the hierarchical control strategy in step S6 is as follows:

[0082] Blue alert: Maintain current operating parameters and continue monitoring;

[0083] Yellow alert: Slightly increase the frequency of water pumps, increase the number of pumps turned on, and strengthen monitoring density;

[0084] Orange alert: Rainfall intensity will be significantly increased; regional enhanced rainfall will be activated; and recharge compensation will be initiated if necessary.

[0085] Red Emergency Alert: Full-load rainfall + activation of emergency pump units, stop excavation, seal off dangerous areas, and organize evacuation.

[0086] In one embodiment, the closed-loop feedback in step S8 adopts a small-amplitude deviation correction mode, with a single adjustment parameter change of ≤10%, to avoid oscillation and instability.

[0087] In step S9, the self-learning module is trained incrementally using actual water level and settlement data as labels to continuously optimize prediction accuracy and dynamic threshold.

[0088] In one embodiment, the process is applicable to all types of deep foundation pits for subway stations, including silty soil, gravel, soft rock, hard rock, soft upper and hard lower, and water-rich fractured soil. It supports unattended intelligent operation throughout the construction process, with water level control accuracy of ±5mm, improved surrounding settlement control accuracy by more than 60%, and energy saving rate of ≥30%.

[0089] In one embodiment, a smart dewatering project for a deep foundation pit at a standard subway station is applied.

[0090] 1. Project Overview

[0091] A standard underground two-level island platform station of a subway has a foundation pit length of 186m, a width of 20.5m, and an excavation depth of 16.2m. The geology is a water-rich composite stratum of silty clay and fine sand, with a groundwater level of 1.8m. The surrounding area is densely populated with old brick-concrete buildings and municipal pipelines, and the environmental sensitivity level is Level 1, with extremely high safety control requirements.

[0092] 2. System Deployment

[0093] Dewatering wells: 46 wells, spaced 15m apart, with a depth of 28m;

[0094] Observation wells: 12, located at the four corners and the middle of the foundation pit;

[0095] Sensing equipment: 46 sets of water level sensors, 8 sets of pore water pressure sensors, 16 sets of sedimentation sensors, 46 sets of flow sensors, and 46 sets of energy consumption sensors;

[0096] Control equipment: 4 variable frequency control cabinets, 46 variable frequency water pumps, 2 sets of recharge devices, and 2 sets of emergency pump sets;

[0097] Edge gateway: 2 units, for real-time local processing;

[0098] Platform: One BIM visualization cloud platform;

[0099] 3. Process Implementation Steps

[0100] Complete equipment installation and network debugging, and system initialization;

[0101] Set benchmark thresholds: water level drawdown 8-10m, daily settlement ≤0.5mm, settlement rate ≤0.3mm / h;

[0102] The system automatically collects data at a frequency of 1Hz and processes it in real time at the edge.

[0103] AI model predicts water level changes and settlement trends over the next 6 hours;

[0104] When the excavation reached a depth of 10m, the water level on the east side dropped slowly and the rate exceeded the standard, triggering a yellow alert.

[0105] Automatic control: The frequency of the water pumps in the eastern area is increased from 30Hz to 40Hz, and two more water pumps are activated in tandem.

[0106] The water level returned to normal after 30 minutes, the settling rate decreased, and the warning was lifted.

[0107] The entire process operates in a closed loop, without human intervention, and generates a precipitation report every 24 hours.

[0108] Once the foundation pit is completed, the system automatically executes the well sealing process, archives the data, and the model completes self-learning.

[0109] 4. Implementation Results

[0110] Water level control accuracy: ±5mm;

[0111] Settlement control value: ≤2.8mm, far below the standard limit;

[0112] Early warning accuracy rate: 99.2%;

[0113] Energy saving rate: 32%;

[0114] Unattended operation duration: 68 days;

[0115] There were no safety accidents or damage to the surrounding environment, and the economic and social benefits were significant.

[0116] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any person skilled in the art can easily conceive of various variations or substitutions within the technical scope disclosed in the present invention, and these should all be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. An intelligent control system for dewatering deep foundation pits in subway stations, characterized in that: include: Multi-source fusion sensing module, edge computing and transmission module, AI intelligent decision-making and prediction module, hierarchical intelligent control module, multi-level risk early warning module, BIM visualization and remote management module, data storage and self-learning optimization module; The output of the multi-source fusion sensing module is connected to the edge computing and transmission module. The output of the edge computing and transmission module is connected to the AI ​​intelligent decision-making and prediction module. The output of the AI ​​intelligent decision-making and prediction module is connected to the hierarchical intelligent control module and the multi-level risk warning module. The hierarchical intelligent control module and the multi-level risk warning module both feed back to the multi-source fusion sensing module to form a closed loop. The BIM visualization and remote management module and the data storage and self-learning optimization module are bidirectionally connected to the above modules.

2. The intelligent control system for dewatering deep foundation pits in subway stations according to claim 1, characterized in that: The multi-source fusion sensing module includes an immersion water level sensor, a pore water pressure sensor, a hydrostatic level, a retaining structure displacement meter, an electromagnetic flow meter, an energy metering module, and a water pump status sensor. The monitoring parameters include well water level, pore water pressure, settlement of surrounding surface / buildings, horizontal displacement of retaining structure, water flow rate, pump voltage / current / power / start / stop status, and pipeline pressure. The sampling frequency is adaptively adjustable from 1 to 5 Hz.

3. The intelligent control system for dewatering deep foundation pits in subway stations according to claim 1, characterized in that: The AI ​​intelligent decision-making and prediction module has a built-in LSTM-Attention water level-settlement coupling prediction model, which is used to predict the probability of water level, settlement and sudden surge risk in the next 1 to 24 hours. The formula for calculating the dynamic early warning threshold is: Dynamic threshold = baseline threshold × formation correction factor × excavation disturbance factor × water environment correction factor × environmental sensitivity factor; The benchmark thresholds are determined based on specifications and design values, and each coefficient is dynamically iterated and optimized according to the geological formation, construction, and environment.

4. The intelligent control system for dewatering deep foundation pits in subway stations according to claim 1, characterized in that: The hierarchical intelligent control module includes a frequency converter control cabinet, a frequency converter water pump, an electric regulating valve, an intelligent recharge device, and an emergency drainage pump. It supports four modes: zoned precise control, variable frequency stepless control, hierarchical linkage control, and precipitation-reinjection linkage. The pump frequency is continuously adjustable from 0 to 50 Hz.

5. The intelligent control system for dewatering deep foundation pits in subway stations according to claim 1, characterized in that: The multi-level risk warning module is divided into four levels: blue alert, yellow warning, orange alert, and red emergency alert. The early warning system simultaneously triggers on-site audio-visual alarms, platform pop-ups, APP push notifications, and SMS notifications, and locates the risk area in three dimensions in the BIM model.

6. The intelligent control system for dewatering deep foundation pits in subway stations according to claim 1, characterized in that: The edge computing and transmission module adopts industrial Ethernet + 5G dual redundant communication, supports network interruption resume and local caching, data processing latency ≤50ms, and end-to-end command response latency ≤100ms.

7. An intelligent control technology for dewatering deep foundation pits in subway stations, coupled with an intelligent control system for dewatering deep foundation pits in subway stations as described in any one of claims 1-6, characterized in that: Includes the following steps: S1 Field Survey and Deployment: Complete the deployment and network debugging of precipitation wells, observation wells, sensing equipment, control equipment, and edge gateways; S2 system initialization: Input engineering geological and environmental parameters, set benchmark thresholds, and complete AI model pre-training and system integration. S3 multi-source data real-time acquisition and spatiotemporal synchronization, acquiring full-element data at a frequency of 1-5Hz; S4 edge data preprocessing: cleaning, noise reduction, fusion, extraction of water level rate, settling rate, and risk feature values; S5AI Coupled Prediction and Dynamic Decision-Making: Predicting water level settlement, calculating dynamic thresholds, and generating zoned frequency conversion control commands; S6-level intelligent precipitation control: Execute corresponding intensity control according to blue alert, yellow warning, orange alert, and red emergency alert; S7 Multi-level Risk Warning and Linked Response: Triggers multi-level alarms and automatically executes risk suppression commands; S8 Closed-Loop Feedback and Effect Verification: Monitor the control effect, correct deviations, and form a control closed loop; S9 data import and self-learning iteration: data archiving, incremental model training, strategy optimization; intelligent well sealing and system shutdown after foundation pit completion.

8. The intelligent control technology for dewatering deep foundation pits in subway stations according to claim 7, characterized in that: The hierarchical control strategy in step S6 is as follows: Blue alert: Maintain current operating parameters and continue monitoring; Yellow alert: Slightly increase the frequency of water pumps, increase the number of pumps turned on, and strengthen monitoring density; Orange alert: Rainfall intensity will be significantly increased; regional enhanced rainfall will be activated; and recharge compensation will be initiated if necessary. Red Emergency Alert: Full-load rainfall + activation of emergency pump units, stop excavation, seal off dangerous areas, and organize evacuation.

9. The intelligent control technology for dewatering deep foundation pits in subway stations according to claim 7, characterized in that: In step S8, the closed-loop feedback adopts a small-amplitude deviation correction mode, with a single adjustment parameter change of ≤10%, to avoid oscillation and instability. In step S9, the self-learning module is trained incrementally using actual water level and settlement data as labels to continuously optimize prediction accuracy and dynamic threshold.

10. The intelligent control technology for dewatering deep foundation pits in subway stations according to claim 7, characterized in that: The technology is applicable to all types of deep foundation pits for subway stations, including silty soil, gravel, soft rock, hard rock, soft upper and hard lower, and water-rich fractured soil. It supports unattended intelligent operation throughout the construction process, with water level control accuracy of ±5mm, improved surrounding settlement control accuracy by more than 60%, and energy saving rate of ≥30%.

Images