Safety risk pre-control method and system for deep foundation pit construction based on cybernetics and dual system theory

Abstract

The present invention discloses a deep foundation pit construction safety risk pre-control method and system based on cybernetics and dual-system theory. Based on sensor expansion, the system performs risk identification and risk monitoring on deep foundation pit construction safety, clarifies risk points, and performs pre-control judgment and pre-control rating on the identified and monitored deep foundation pit construction safety risks. Finally, based on actuators and DST expansion, the system performs pre-control decision-making and pre-control execution on the pre-control risks of step 2. The present invention first systematically identifies and monitors root risks and state risks, which helps to timely discover and prevent potential risks, thereby improving the operational stability of the project. By monitoring various state risks, the system can timely identify and respond to potential safety hazards, ensuring the safety of construction personnel and the surrounding environment, and also playing a positive role in project cost control.

Description

Technical Field

[0001] The present invention belongs to the field of deep foundation pit construction risk management, and in particular relates to a deep foundation pit construction safety risk pre-control method and system based on cybernetics and dual-system theory. Background Art

[0002] Control theory, also known as cybernetic, was originally used to study the stability and control problems of mechanical systems. With the development of science and technology, control theory has gradually expanded to industrial control systems, automation systems, transportation and other fields. At the same time, control theory has also been proven to provide effective guidance for the safety management of complex systems ( & Rollenhagen, 2014). In control theory, a system's control loop is described as a closed-loop structure consisting of sensors, controllers, and actuators. Sensors measure the system's dynamic state, while controllers determine actuator actions based on the discrepancy between the sensor's measurements and the control standard. Ultimately, the actuators adjust the system's state. Furthermore, within this closed-loop structure, the system typically executes multiple tasks in parallel, creating feedback loops for each task. These feedback loops form a hierarchical structure based on their impact on the system (Powers, 1973). This structure is crucial for improving the performance of safety risk prevention and control in deep foundation pit projects. The introduction of control theory will enable real-time monitoring and adjustment of deep foundation pit construction status, enabling continuous feedback on safety risk prevention and control management under changing environmental conditions, thereby maintaining the steady progress of safe deep foundation pit construction.

[0003] The dual-system concept originated with William Jame (1996). After extensive research and discussion by numerous scholars, it was ultimately summarized into the now commonly used cognitive psychology theory, Dual System Theory (DST). DST posits that human behavior and decision-making are driven by two relatively independent cognitive systems: System 1 and System 2. System 1 is known as the intuitive heuristic system. This system relies on intuition and experience to solve problems, automatically generating solutions and making efficient decisions. System 2 is known as the rational analytical system. This system relies on further processing of information, possessing functions of reflection, control, and inhibition, expending more resources on problem-solving. Decisions made by System 2 often optimize those made by System 1 or complement decisions that System 1 was unable to complete. Dual-system theory provides a theoretical basis for explaining individual safety behavior and is also valuable for understanding safety management decision-making models. Relying solely on monitoring and early warning of engineering risks does not meet the needs of proactive safety risk prevention and control in deep foundation pit construction.

[0004] Deep foundation pit excavation, due to its high risk, has always been a key area of ​​focus in the construction industry. Currently, several technologies, based on control theory, have introduced DST to preemptively manage the safety risks of deep foundation pits. For example, patent CN110414785A discloses a risk identification method for deep foundation pit excavation, quantifying the risks involved and enabling timely response measures based on the risk level. For example, patent CN112115529B discloses a risk assessment method for ultra-deep foundation pit construction, addressing the technical issue of a significant discrepancy between assessed and actual risks due to a failure to account for the amplification effect of risk with increasing depth.

[0005] However, the above prior art still has the following deficiencies:

[0006] First, risk factors are insufficiently integrated. While the aforementioned methods identify and assess risk factors in deep foundation pit construction, they fail to rationally integrate the causal factors that lead to accidents into the risk management system, nor do they identify effective accident prevention measures in practice. This leads to a disconnect between risk management and actual construction safety, and prevents comprehensive risk control.

[0007] Secondly, there is a lack of clear guidance for safety risk management. Current safety oversight relies primarily on a combination of government regulations and corporate safety practices, but lacks clear guidelines for safety risk management within deep foundation pit projects. This makes it difficult for laws and regulations to fully account for the unique characteristics of each project, reducing their guiding value for on-site managers.

[0008] Furthermore, there is an over-reliance on subjective judgment. Risk source identification, risk rating, and risk control decisions still rely primarily on the experience and subjective judgment of safety managers or supervisors. Due to differences in managerial expertise, inexperience or human error can exacerbate safety hazards and, in severe cases, lead to further safety incidents.

[0009] Finally, purely automated risk management is difficult to achieve. Due to the complexity and unpredictability of deep excavation construction, it is challenging to comprehensively address all potential risks before a project begins. The dynamic nature of a construction site means it is impossible to predict and plan for all possible risk scenarios in advance, making a purely automated risk management approach less feasible in practical applications.

[0010] In summary, existing technologies for deep foundation pit construction risk management suffer from deficiencies such as insufficient integration, weak guidance, difficulty in proactive prevention, and over-reliance on subjective judgment. To ensure the safety and smooth completion of deep foundation pit construction, a semi-automated proactive safety risk pre-control method that balances monitoring data with expert experience is urgently needed. Summary of the Invention

[0011] Purpose of the invention: The purpose of the present invention is to solve the deficiencies in the prior art and to provide a method and system for pre-controlling the safety risks of deep foundation pit construction based on cybernetics and dual-system theory. In response to the current problems faced by the safety risk management of deep foundation pit projects, combined with the need for deep integration of information technology and production safety, the traditional safety management model that is mainly based on post-recovery is transformed into an active and spontaneous pre-safety management model to ensure the safety of deep foundation pit construction, continuity of construction period and smooth completion.

[0012] Technical solution: The present invention provides a method for pre-controlling safety risks in deep foundation pit construction based on cybernetics and dual-system theory, comprising the following steps:

[0013] Step 1: Conduct risk identification (Se-RI) and risk monitoring (Se-RM) for deep foundation pit construction safety based on sensor expansion to identify risk points;

[0014] Step 1.1: Risk Identification (Se-RI) process. First, the safety risks of deep foundation pit construction are divided into root risk (RR) and state risk (SR). When identifying root risk (RR), the root causes that may lead to a series of state risks or accidents are explored, including the operating data of construction personnel, the operating conditions of mechanical equipment, and environmental monitoring indicators. When identifying state risk (SR), the focus is on the current physical state of the deep foundation pit, including pit displacement, crack development, and water level changes. After completing the identification of root risk and state risk (i.e., completing risk identification (Se-RI)), text mining and pattern recognition are used to extract key risk signals from the case library to identify and classify deep foundation pit construction safety risks.

[0015] Step 1.2: Risk Monitoring During the Se-RM process, identified root risks (RR) are first monitored through checklist inspections and surveillance cameras, while identified state risks (SR) are monitored through sensors. The resulting monitoring data is then transmitted in real time via wireless or wired networks to a central monitoring system for preliminary screening and preprocessing. Next, the multi-source heterogeneous risk source data preprocessing mechanism for deep foundation pit construction is used to preprocess multi-heterogeneous data, including manual inspection data, mechanical sensor data, and geological environmental monitoring data, to form a structured risk monitoring dataset.

[0016] Step 2: Based on the controller expansion, the deep foundation pit construction safety risks identified and monitored in step 1 are pre-controlled and judged Co-PJ and pre-control rating Co-PR;

[0017] Step 2.1: When pre-controlling and judging Co-PJ, the initial warning thresholds are first set based on the real-time monitoring data through the dynamic feedback mechanism on risk early warning standards (DFM-REWS) for deep foundation pit construction. basic Perform dynamic feedback adjustments. If the current stage risk detected in step 1 exceeds the dynamically adjusted warning threshold, the Co-PR rating will be automatically initiated.

[0018] Step 2.2: Pre-control rating (Co-PR) process: Risk points are first rated according to severity. Various risks are classified and prioritized through a hierarchical rating mechanism. Decision trees and risk matrices are used to guide the development of risk management strategies.

[0019] Step 3: Based on the actuator and DST extension, pre-control decision Ac-PD and pre-control execution Ac-PE are performed on the pre-control risk in step 2;

[0020] Step 3.1: When making preventive control decisions (Ac-PD), an intuitive heuristic execution system (IHES) and an analytical processing execution system (APES) are developed based on DST. IHES simulates human intuitive judgment to quickly screen and apply existing risk control strategies. The built-in heuristic rules and algorithms of IHES enable IHES to call upon effective risk assessments and decisions in the safety risk countermeasures database. When faced with risk points not recorded in the database, APES intervenes in the risk decision-making process. APES uses a slow system approach to analyze newly identified risk points in detail, clarify specific preventive control strategies and measures, and designate the departments and personnel involved in subsequent actions. APES is responsible for updating the risk prevention decision database to ensure that all preventive control measures are based on the latest practical experience.

[0021] Step 3.2: When Ac-PE is executed during pre-control, the pre-control decision output by Ac-PD will be converted into specific actions, and execution orders will be automatically issued to relevant departments and personnel to ensure that risk prevention measures are effectively implemented.

[0022] Furthermore, the deep foundation pit construction safety risk pre-control method based on cybernetics and dual-system theory is characterized in that the method of dynamically adjusting the risk warning threshold in step 2.1 is as follows:

[0023] Set the initial warning thresholds basic Together with the standardized monitoring data, it is continuously input into the Deep Foundation Pit Construction Risk Early Warning Standard Dynamic Feedback Mechanism DFM-REWS for prediction fitting;

[0024] Calculate the deviation between real-time data and predicted values ​​using data-driven models t , then, DFM-REWS evaluates the current performance P t , if the current performance P t Failure to achieve target performance P target , the difference ΔP between the two will be calculated t =P target -P t , and finally according to ΔP t The result of the feedback adjustment parameter β is used to dynamically update the warning threshold Threshold st+1 =Threshold st +β×ΔP t ;

[0025] Among them, et is obtained by comparing the actual monitoring value y t With the model f(x t θ t ) is determined by the expected value of the prediction, x t represents the input variable for real-time monitoring, θ t It is the parameter of the preset prediction model, which is calculated by the deviation e t The influence of P is quantified t , Threshold st is the risk warning threshold at the current stage, and Threshold st+1 is the updated threshold.

[0026] Furthermore, the details of the preliminary screening and pre-processing performed by the central monitoring system in step 1.2 are as follows:

[0027] Data Validation, which is the process of checking the collected raw data to identify and eliminate errors or outliers in the data collection, such as readings outside the expected range or data points that are not physically possible;

[0028] Data format standardization involves converting data from various sources into a unified format to ensure compatibility and consistency in subsequent processing. This includes standardizing timestamp formats, data encoding, and measurement units. Missing data handling involves identifying missing data and applying appropriate techniques to fill in or interpolate missing values ​​to prevent information gaps in subsequent analysis.

[0029] Through screening and preprocessing, we prepare for subsequent in-depth data analysis and structuring, ensuring that the data has been cleaned and standardized before entering the multi-source heterogeneous risk source data preprocessing mechanism, providing a reliable foundation for accurate risk assessment and management.

[0030] Furthermore, in step 2.2, the Co-PR, the collected root risk and status risk data are first analyzed, and a fuzzy logic algorithm is used to determine the nature and severity of the risk. The risk data is then fed into a dynamic risk rating model, which ranks and categorizes the risks according to preset rating criteria and thresholds to ensure that the highest priority risks receive a timely response. The specific steps are as follows:

[0031] First, we define the membership function evaluation indicators: probability of occurrence (Probability, P) and severity (Severity, S). Both indicators are divided into three fuzzy sets: low, medium, and high, and the membership function is defined as follows:

[0032] For the probability of occurrence P:

[0033] Low (R 1* ):

[0034] Middle (R 2* ):

[0035] High (R 3* ):

[0036] For hazard level S:

[0037] Low (R *1 ):

[0038] Middle (R *2 ):

[0039] High (R *3 ):

[0040] Next, define the risk prevention level rules. Combining the fuzzy sets of probability of occurrence and degree of harm, define the following rules to evaluate the final risk prevention level:

[0041] Rule 1: If the risk level classification result is R 11 、R 12 、R 21 , then the risk control level is 1 (L1); Rule 2: If the risk level classification result is R 13 、R 22 、R 31 , then the risk control level is 2 (L21); Rule 3: If the risk level classification result is R 23 、R 32 , then the risk control level is 3 (L3);

[0042] Rule 4: If the risk level classification result is R 33 , then the risk control level is 4 (L4);

[0043] Finally, reasoning and defuzzification are performed, using the Mamdani inference method for fuzzy reasoning. For each rule, the output risk level is also expressed using fuzzy sets. Defuzzification uses the centroid method to calculate the final value of the risk level.

[0044] In addition, a performance feedback mechanism is set up in Co-PR to ensure that the rating model can be continuously optimized based on new risk monitoring data and historical response effects.

[0045] The present invention also discloses a system for implementing a deep foundation pit construction safety risk pre-control method based on cybernetics and dual-system theory, comprising a deep foundation pit construction safety risk identification and monitoring module based on sensor expansion, a deep foundation pit construction safety risk pre-control judgment and rating module based on controller expansion, and a deep foundation pit construction safety risk pre-control decision-making and execution module based on actuator and DST expansion.

[0046] The sensor-based deep foundation pit construction safety risk identification and monitoring module includes a risk identification (Se-RI) submodule and a risk monitoring (Se-RM) submodule. The risk identification submodule Se-RI identifies deep foundation pit construction safety risks and divides them into root risks (RR) and status risks (SR). The risk monitoring submodule Se-RM monitors the identified risks, and the monitoring data is transmitted to the central monitoring system in real time, which screens and pre-processes them to form a structured risk monitoring data set.

[0047] The controller-based extended deep foundation pit construction safety risk pre-control judgment and rating module includes a pre-control judgment submodule Co-PJ (Pre-control Judgment) and a pre-control rating submodule Co-PR (Pre-control Rating); the pre-control judgment submodule Co-PJ dynamically adjusts the initial threshold based on real-time monitoring data through the deep foundation pit construction risk early warning standard dynamic feedback mechanism DFM-REWS. If the current stage risk exceeds the adjusted threshold, the pre-control rating submodule Co-PR will be automatically activated; the pre-control rating submodule Co-PR rates risk points according to severity, classifies and prioritizes various risks through a hierarchical rating mechanism, and adopts a decision tree and risk matrix to guide the formulation of risk management strategies;

[0048] The deep foundation pit construction safety risk pre-control decision-making and execution module based on the actuator and DST expansion includes a pre-control decision sub-module Ac-PD (Pre-control Decision) and a pre-control execution sub-module Ac-PE (Pre-control Execution); in the pre-control decision sub-module Ac-PD, an intuitive heuristic execution system IHES and an analysis and processing execution system APES are developed based on DST. IHES simulates human intuitive judgment to quickly screen and apply existing risk control strategies. When faced with risk points not recorded in the database, APES will intervene in the risk decision-making process; in the pre-control execution sub-module Ac-PE, the pre-control decision output by Ac-PD will be converted into specific actions, and execution orders will be automatically issued to relevant departments and personnel to ensure that risk prevention measures are effectively implemented.

[0049] Beneficial Effects: This invention can significantly improve the scientific and systematic nature of risk management, thereby effectively avoiding the aforementioned problems. Through the implementation of this method, risk identification and monitoring are more comprehensive and systematic, pre-control judgment and rating are more dynamic and accurate, and pre-control decision-making and execution are more timely and efficient, ultimately achieving dynamic management and semi-automatic control of deep foundation pit construction risks. Compared with existing technologies, this invention has the following advantages:

[0050] 1. Significantly Improved Risk Management Efficiency: Through a semi-automated, proactive safety risk management approach based on monitoring data and expert experience, risk management response time is reduced from hours to minutes, significantly improving management efficiency and enabling real-time risk monitoring and immediate response. 2. Significant Economic Benefits: Preliminary implementation data indicates that this invention can reduce the occurrence of risk incidents by over 15%, correspondingly reducing the direct and indirect costs associated with these incidents. Project cost savings are expected to reach 5% of total costs, significantly lower than traditional risk management methods.

[0051] 3. Significant social effects: More precise risk management methods improve construction site safety, reduce work-related accidents, thereby reducing social insurance and medical expenses, and increasing public confidence in construction safety.

[0052] Furthermore, compared to traditional methods, this invention's technological and operational innovations eliminate the reliance on individual managers' experience and subjective judgment, enabling risk management to be conducted through a systematic, data-driven approach, thereby achieving higher safety management standards and lowering the probability of risk. Therefore, the technical effects of this invention have significant social significance and economic value, and are particularly suitable for high-risk, technically demanding deep foundation pit construction projects. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] Figure 1 Schematic diagram of the entire pre-control method of the present invention;

[0054] Figure 2 This is a diagram for identifying root risks and status risks in deep foundation pit construction according to the present invention;

[0055] Figure 3 Schematic diagram of the preprocessing mechanism of multi-source heterogeneous risk source data in deep foundation pit construction in the present invention;

[0056] Figure 4 Schematic diagram of the dynamic feedback mechanism of the deep foundation pit construction risk early warning standard in the present invention;

[0057] Figure 5 This is a schematic diagram of the operation logic of the deep foundation pit construction safety risk pre-control rating function of the present invention;

[0058] Figure 6 This is a logical diagram of the deep foundation pit construction safety risk pre-control decision-making based on the dual-system theory in the present invention;

[0059] Figure 7 It is a schematic diagram of the overall framework of the present invention;

[0060] Figure 8 Schematic diagram of the implementation process of the embodiment;

[0061] Figure 9 Schematic diagram of surface subsidence warning rate in the embodiment. DETAILED DESCRIPTION

[0062] The technical solution of the present invention is described in detail below, but the protection scope of the present invention is not limited to the embodiments.

[0063] Deep foundation pit construction, due to its complexity and high risk, has always been a key area of ​​concern in the construction industry. While existing risk management methods have improved, they often struggle to achieve proactive risk management due to issues such as insufficient integration of risk factors, unclear safety management guidance, and overreliance on subjective judgment. This leads to high accident rates and high risk management costs. These issues not only extend project durations and increase costs, but can also cause significant quality and personal safety issues, resulting in unpredictable economic losses.

[0064] To solve existing technical problems, such as Figure 1 and Figure 7 As shown, the deep foundation pit construction safety risk pre-control method based on cybernetics and dual-system theory of the present invention includes the following steps:

[0065] Step 1: Conduct risk identification (Se-RI) and risk monitoring (Se-RM) for deep foundation pit construction safety based on sensor expansion to identify risk points;

[0066] Step 1.1: Risk Identification (Se-RI) process. First, the safety risks of deep foundation pit construction are divided into root risk (RR) and state risk (SR). When identifying root risk (RR), the root causes that may lead to a series of state risks or accidents are explored, including the operating data of construction personnel, the operating status of mechanical equipment, and environmental monitoring indicators. When identifying state risk (SR), the focus is on the current physical state of the deep foundation pit, including pit displacement, crack development, and water level changes. After completing the identification of root risk and state risk (after completing the risk identification Se-RI), text mining and pattern recognition are used to extract key risk signals from the case library to identify and classify the safety risks of deep foundation pit construction.

[0067] Step 1.2: Risk Monitoring During the Se-RM process, identified root risks (RR) are first monitored through checklist inspections and surveillance cameras, while identified condition risks (SR) are monitored through sensors. The resulting monitoring data is then transmitted in real time via wireless or wired networks to a central monitoring system for preliminary screening and preprocessing. Next, the deep foundation pit construction multi-source heterogeneous risk source data preprocessing mechanism preprocesses manual inspection data, mechanical sensor data, and geological environment monitoring data to form a structured risk monitoring dataset.

[0068] Step 2: Based on the controller expansion, the deep foundation pit construction safety risks identified and monitored in step 1 are pre-controlled and judged Co-PJ and pre-control rating Co-PR;

[0069] Step 2.1: When pre-controlling and judging Co-PJ, the initial warning standard thresholds Thresholds are firstly set through the deep foundation pit construction risk warning standard dynamic feedback mechanism DFM-REWS. basic Perform dynamic feedback adjustments. If the current stage risk detected in step 1 exceeds the dynamically adjusted warning threshold, the Co-PR rating will be automatically initiated.

[0070] Step 2.2: Pre-control rating (Co-PR) process: First, risk points are rated according to severity. A hierarchical rating mechanism is used to categorize and prioritize various risks. A decision tree and risk matrix are used to guide the development of risk management strategies.

[0071] Step 3: Based on the actuator and DST extension, pre-control decision Ac-PD and pre-control execution Ac-PE are performed on the pre-control risk in step 2;

[0072] Step 3.1: When making preventive control decisions (Ac-PD), develop an intuitive heuristic execution system (IHES) and an analytical processing execution system (APES) based on DST. IHES simulates human intuitive judgment to quickly screen and apply existing risk control strategies. When faced with risk points not recorded in the database, APES will intervene in the risk decision-making process, conduct a detailed analysis of the newly identified risk points, clarify specific preventive control strategies and measures, and designate the departments and personnel involved. APES will also be responsible for updating the risk prevention decision database to ensure that all preventive control measures are based on the latest practical experience.

[0073] Step 3.2: When Ac-PE is executed during pre-control, the pre-control decision output by Ac-PD will be converted into specific actions and automatically issued to relevant departments and personnel to ensure that risk prevention measures are effectively implemented, such as Figure 6 shown.

[0074] like Figure 2 As shown, this embodiment identifies and categorizes the deep foundation pit construction safety risks obtained after risk identification Se-RI. The root risk RR is differentiated based on the pre-control factors and pre-control objects, resulting in personnel, machinery, materials, methods, and environment, etc. For example, the equipment status of a submersible pump includes its maintenance status, and submersible pumps are further divided into mud preparation pumps, etc. The status risk SR is differentiated based on the pre-control scope and pre-control object, resulting in settlement risk (surface settlement, pile top settlement), deformation risk (lateral deformation of soil, etc.), displacement risk, and others, etc.; ultimately, the risk scope, risk location, risk frequency, and risk warning indicators are obtained.

[0075] like Figure 3 As shown, in the risk monitoring Se-RM process of this embodiment, different strategies are used to complete risk data monitoring and collection for root risk RR and state risk SR. Root risk RR monitoring is mainly implemented through checklist inspections and surveillance cameras. The inspection checklist based on the Safety Risk Evolution Mechanism of deep foundation pit construction can directly evaluate the safety status of the construction site and promptly identify potential dangers caused by improper operation of construction personnel, equipment failure or changes in environmental factors; surveillance cameras provide continuous visual monitoring and provide continuous risk behavior and environmental changes. State risk SR monitoring relies on high-precision sensor technology, such as displacement sensors, crack monitors and water level sensors, to track the physical status of deep foundation pits in real time and ensure continuous monitoring of these key indicators.

[0076] Here, the multivariate heterogeneous data obtained above are finally preprocessed to form a structured risk monitoring data set, which includes text information, coordinate positioning, static images, dynamic videos, and continuous numerical data.

[0077] The method of dynamically adjusting the risk warning threshold in step 2.1 of this embodiment is as follows: Figure 4 As shown, the initial warning standard threshold of the monitoring data at the current time t is input, a prediction fit is performed, the loss deviation of the prediction fitting result is calculated, and performance evaluation and performance deviation calculation are performed, and then the warning threshold is dynamically adjusted; the details are as follows:

[0078] Set the initial warning threshold Thresholds basic Together with the standardized monitoring data, it is continuously input into the Deep Foundation Pit Construction Risk Early Warning Standard Dynamic Feedback Mechanism DFM-REWS for prediction fitting;

[0079] Calculate the deviation between real-time monitoring data and risk prediction value using data-driven model t , then, DFM-REWS evaluates the current performance P t , if the current performance P t Failure to achieve target performance P target , the difference ΔP between the two will be calculated t =P target -P t , and finally according to ΔP t The result of the feedback adjustment parameter β is used to dynamically update the warning threshold Threshold st+1 =Threshold st +β×ΔP t ;

[0080] Among them, e t By comparing the actual monitoring value y t With the model f(x t θ t ) is determined by the expected value of the prediction, x t Represents the monitoring data at the current time t, θ t It is the parameter of the preset prediction model, which is calculated by the deviation e t The influence of P is quantified t , Threshold st is the risk warning threshold at the current stage, and Threshold st+1 is the updated threshold.

[0081] In step 2.2 of the Co-PR pre-control rating of this embodiment, the collected root risk and status risk data are first preliminarily analyzed, and a fuzzy logic algorithm is used to determine the nature and severity of the risk. Then, the risk data is fed into a dynamic risk rating model, which ranks and classifies the risks according to the preset rating criteria and thresholds to ensure that the highest priority risks receive a timely response, such as Figure 5 The specific steps are as follows:

[0082] First, define the membership function evaluation indicators: probability of occurrence P and severity of harm S. Both indicators are divided into three fuzzy sets: low, medium, and high, and the membership function form is defined as follows:

[0083] For the probability of occurrence P:

[0084] Low (R 1* ):

[0085] Middle (R 2* ):

[0086] High (R 3* ):

[0087] For hazard level S:

[0088] Low (R *1 ):

[0089] Middle (R *2 ):

[0090] High (R *3 ):

[0091] Next, define the risk prevention level rules. Combining the fuzzy sets of probability of occurrence and degree of harm, define the following rules to evaluate the final risk prevention level:

[0092] Rule 1: If the risk level classification result is R 11 、R 12 、R 21 , then the risk control level is 1 (L1);

[0093] Rule 2: If the risk level classification result is R 13 、R 22 、R 31 , then the risk control level is 2 (L21);

[0094] Rule 3: If the risk level classification result is R 23 、R 32, then the risk control level is 3 (L3); Rule 4: If the risk level classification result is R 33 , then the risk control level is 4 (L4);

[0095] Finally, reasoning and defuzzification are performed, using the Mamdani inference method for fuzzy reasoning; for each rule, the output risk level is also expressed using fuzzy sets. Defuzzification uses the centroid method to calculate the final value of the risk level.

[0096] like Figure 7 As shown, the deep foundation pit construction safety risk pre-control system of this embodiment based on cybernetics and dual-system theory includes a deep foundation pit construction safety risk identification and monitoring module based on sensor expansion, a deep foundation pit construction safety risk pre-control judgment and rating module based on controller expansion, and a deep foundation pit construction safety risk pre-control decision-making and execution module based on actuator and DST expansion. The deep foundation pit construction safety risk identification and monitoring module based on sensor expansion is provided with a risk identification Se-RI submodule and a risk monitoring Se-RM submodule; the risk identification submodule Se-RI identifies deep foundation pit construction safety risks and divides them into root risk RR and state risk SR; the risk monitoring submodule Se-RM monitors the identified risks, and the monitoring data is transmitted in real time to the central monitoring system, which screens and pre-processes them, and finally forms a structured risk monitoring data set.

[0097] The controller-expanded deep foundation pit construction safety risk pre-control judgment and rating module includes a pre-control judgment sub-module Co-PJ and a pre-control rating sub-module Co-PR. The pre-control judgment sub-module Co-PJ dynamically adjusts the initial threshold based on real-time monitoring data through the deep foundation pit construction risk early warning standard dynamic feedback mechanism DFM-REWS. If the risk in the current stage exceeds the adjusted threshold, the pre-control rating sub-module Co-PR will be automatically started. The pre-control rating sub-module Co-PR rates risk points according to severity, classifies and prioritizes various risks through a hierarchical rating mechanism, and uses a decision tree and risk matrix to guide the formulation of risk management strategies.

[0098] The deep foundation pit construction safety risk pre-control decision-making and execution module based on the actuator and DST expansion includes the pre-control decision-making sub-module Ac-PD and the pre-control execution sub-module Ac-PE; in the pre-control decision-making sub-module Ac-PD, the intuitive heuristic execution system IHES and the analysis and processing execution system APES are developed based on DST. IHES simulates human intuitive judgment to quickly screen and apply existing risk control strategies. When faced with risk points not recorded in the database, APES will intervene in the risk decision-making process; in the pre-control execution sub-module Ac-PE, the pre-control decisions output by Ac-PD will be converted into specific actions, and execution orders will be automatically issued to relevant departments and personnel to ensure that risk prevention measures are effectively implemented.

[0099] Example 1

[0100] As shown in FIG8 , here, based on the technical solution of the present invention, safety risk pre-control is performed on a deep foundation pit construction site. The specific process is as follows:

[0101] Step 1: Establish the data foundation for risk prevention. At the initial stage of risk prevention, the environment and data foundation for deep foundation pit construction risk management should be established, including the establishment of an initial risk database and relevant risk assessment standards.

[0102] Step 2: Conduct preliminary risk identification and real-time monitoring based on established risk prevention and control objectives. The risk identification and monitoring module conducts preliminary screening and classification of potential risks. This process includes real-time monitoring of the construction site and analysis of historical data to identify key risk factors that may affect construction safety and efficiency.

[0103] Step 3: Assess the controllability and risk level of identified risks. Initially identified risks are evaluated in depth using the Pre-Control Identification and Rating module to determine their controllability and specific risk levels.

[0104] Step 4: Execute risk prevention and control decisions and response strategies. Based on the controllability and level of the risk, the risk prevention and control decision and execution module determines whether to execute existing risk prevention and control decisions and response strategies, or to conduct further review and discussion of risk points for which no risk prevention and control strategies are currently available.

[0105] Step 5: Monitor the effectiveness of risk prevention measures and make feedback adjustments. After implementing risk prevention measures, continuously monitor their effectiveness and adjust risk prevention thresholds and store new risk prevention strategies based on actual conditions through feedback loop learning.

[0106] Example 2

[0107] The invention technology scheme was implemented in a pilot project of a tunnel construction project JLT covering a deep foundation pit project. The foundation pit of the main tunnel project is about 1.1 km long, 21 to 41 m wide, with a maximum excavation depth of 22 m. The excavation depth of the local intersection with the rail transit is 26 m. The foundation pit is designed to be divided into 22 sub-pits, and parallel flow operations are organized. The following takes the deep foundation pit project of the B02 structural block of the A sub-pit in a certain area of ​​the tunnel as an example.

[0108] Table 1 Overview of deep foundation pit

[0109]

[0110] according to Figure 2 The risk identification classification shown here identifies and extracts the root risks that may be encountered during the construction process of this example. These risks are organized into five aspects: personnel, machinery, materials, methods, and environment, forming a root risk table. This table includes the parent root risk code, parent root risk, root risk code, root risk, root risk manifestation code, and the specific manifestation of the root risk.

[0111] Table 2 lists some examples of possible root risks identified before the start of construction of this project.

[0112] Table 2 Example of root risk list

[0113]

[0114]

[0115] Identification of state risks involves organizing settlement, deformation, displacement, and other conditions into a state risk table. This table clearly defines the state risk code, state risk, state risk object code, and state risk object. The key to risk identification at this stage is the comprehensive application of multiple methods to identify, as far as possible, the root risks that lead to state risks and establish correlation rules between them.

[0116] Table 3 lists the possible status risks identified before the start of the case construction and their relationship with the root risks.

[0117] Table 3 Example of status risk list

[0118]

[0119]

[0120] Table 4 shows the monitoring object, monitoring range, measurement point sections, and spacing for typical condition risk monitoring equipment, such as example A-B02. The Se-RM submodule provides real-time visibility of the key configuration points and monitoring status of condition risk equipment. Based on the root risk table, condition risk table, and monitoring equipment, the Se-RM submodule continuously monitors and records risks throughout the deep foundation pit construction process.

[0121] Table 4A-B06 Examples of Condition Risk Monitoring Equipment

[0122]

[0123] When a risk situation that requires pre-control occurs, the deep foundation pit construction safety risk pre-control judgment and rating module "Co-PJ&Co-PR" based on the controller expansion and the deep foundation pit construction safety risk pre-control decision-making and execution module "Ac-PD&Ac-PE" based on the actuator and DST expansion will be called.

[0124] The following will take "the surface settlement risk around A-B06 exceeds the warning value" as an example to describe in detail how this embodiment performs pre-control for specific risks. In the "Co-PJ&Co-PR" module, the warning standards for surface settlement status risks and risk pre-control ratings under various scenarios are shown in Table 5. One day, when monitoring the surface settlement around A-B06, it was found that the settlement value reached 3.1mm and the cumulative settlement value reached 33.7mm (the environmental protection level of this structural block is level I, depth H = 21.8m). The risk monitoring situation this time met the warning standard, and the pre-control rating was L2. The Co-PR submodule generates a risk warning sheet coded as "20210405-SE-SU-YJ-01" based on the rating results. In the safety risk pre-control countermeasures database, automatic screening is performed based on the key search term "SE-SU-YJ" in the risk warning sheet code.

[0125] The screening results show that the following key preventive measures are in place for this risk situation:

[0126] 1. Stop excavation immediately.

[0127] 2. Contact the design and supervision units to identify the main causes of foundation pit deformation based on the association rules with specific root risks

[0128] 3. Take measures to control deformation, such as increasing the support axial force or increasing the number of supports. If necessary, water injection into the foundation pit should be used to offset the pressure of the surrounding soil.

[0129] 4. Redefine the inspection list and strengthen inspection efforts on relevant root risks.

[0130] Since the corresponding pre-control countermeasures already exist in the safety risk pre-control countermeasures database, the Ac-PD submodule directly generates a risk pre-control order numbered "YKD20210415-001" with a status of "pending execution". The pre-control order will directly push the specific pre-control countermeasures to the relevant departments through the Ac-PE submodule (this risk pre-control order will be pushed to the Engineering Department, Quality Department, and Safety Department). After each department implements the pre-control countermeasures, the status of the pre-control order will be updated to "executed". So far, this method has successfully completed the pre-control operation of "the surface settlement risk around the A-B06 foundation pit exceeds the warning value."

[0131] Table 5 Warning standards and risk control levels for surface settlement around deep foundation pits

[0132]

[0133] Example 3

[0134] To validate the management effectiveness of this technical solution, we analyzed operational data from a one-year implementation case. Based on this method, we identified 196 root risk points and 21 status risk points, and completed 556 risk prevention orders. A risk prevention decision database pre-stored 207 prevention strategies, with 34 new ones added during deep foundation pit construction. Next, we will use "surface settlement around foundation pits" as an example to demonstrate the management effectiveness of this method.

[0135] The monthly surface subsidence risk warning rate calculated based on the monitoring frequency and the number of warnings is shown in Figure 6. Risk, defect, and accident are an evolutionary path. Timely and appropriate preventive measures for state risks can greatly reduce the possibility of further evolution of risks. Preventive measures taken for the root risk of the state risk can effectively reduce the probability of the state risk getting out of control next time. Therefore, the risk warning rate can well reflect the effect of this method in reducing the probability of risk occurrence during the construction process. Figure 9 As can be seen from the trend line in the figure, the risk out-of-control ratio is generally on a downward trend.

[0136] Table 6 Key indicators for pre-control of surface settlement risk around foundation pits

[0137]

[0138] Overall, the present invention, through systematic identification and monitoring of root and condition risks, helps promptly detect and prevent potential risks, thereby improving project operational stability and bringing greater controllability and predictability. Secondly, as a data-driven management approach, the inspection of root risks and monitoring of condition risks provide the project management team with real-time, objective information on project status. This data-based decision support ensures that the management team can make decisions based on facts and data, rather than relying solely on experience or intuition, thereby improving the accuracy and credibility of decisions. Furthermore, the present invention is crucial for improving project safety. By monitoring various condition risks, such as foundation pit settlement, deflection, and deformation, potential safety hazards can be promptly identified and addressed, ensuring the safety of construction workers and the surrounding environment. This not only helps prevent accidents but also mitigates their impact when they do occur, significantly improving the overall safety level of the project. Finally, the present invention also plays a positive role in project cost control. Preliminary calculations show that the application of this method in a case study saved approximately 10% of the total cost. By reducing the occurrence of accidents and unexpected events, the resulting additional costs and project delays can be avoided, ensuring the economic benefits and sustainability of the project.

Claims

1. A method for pre-controlling safety risks in deep foundation pit construction based on cybernetics and dual-system theory, characterized in that: The following steps are involved: Step 1: Conduct risk identification (Se-RI) and risk monitoring (Se-RM) for deep foundation pit construction safety based on sensor expansion to identify risk points; Step 1.1: Risk Identification (Se-RI) process. First, deep foundation pit construction safety risks are divided into root risk (RR) and condition risk (SR). When identifying root risk (RR), the root causes that may lead to a series of condition risks or accidents are explored, including the operating data of construction personnel, the operating conditions of mechanical equipment, and environmental monitoring indicators. When identifying condition risk (SR), the focus is on the current physical state of the deep foundation pit, including pit displacement, crack development, and water level changes. After completing the identification of root risk and condition risk, text mining and pattern recognition are used to extract key risk signals from the case library to identify and classify deep foundation pit construction safety risks. Step 1.2: Risk Monitoring During the Se-RM process, first, the identified root risk RR is monitored through inventory inspections and surveillance cameras, and the identified state risk SR is monitored through sensors; The resulting monitoring data is then transmitted in real time via wireless or wired networks to a central monitoring system, which performs preliminary screening and preprocessing. Next, a multi-source heterogeneous risk source data preprocessing mechanism for deep foundation pit construction is used to preprocess manual inspection data, mechanical sensor data, and geological environment monitoring data to form a structured risk monitoring dataset. Step 2: Based on the controller expansion, the deep foundation pit construction safety risks identified and monitored in step 1 are pre-controlled and judged Co-PJ and pre-control rating Co-PR; Step 2.1: When pre-controlling and judging Co-PJ, the initial warning standard thresholds Thresholds are firstly set through the deep foundation pit construction risk warning standard dynamic feedback mechanism DFM-REWS. basic Perform dynamic feedback adjustments. If the current stage risk detected in step 1 exceeds the dynamically adjusted warning threshold, the Co-PR rating will be automatically initiated. Step 2.2: Pre-control rating (Co-PR) process: First, risk points are rated according to severity. A hierarchical rating mechanism is used to categorize and prioritize various risks. A decision tree and risk matrix are used to guide the development of risk management strategies. Step 3: Based on the actuator and DST extension, pre-control decision Ac-PD and pre-control execution Ac-PE are performed on the pre-control risk in step 2; Step 3.1: When making pre-control decisions (Ac-PD), develop an intuitive heuristic execution system (IHES) and an analytical processing execution system (APES) based on DST. IHES simulates human intuitive judgment to quickly screen and apply existing risk control strategies. When faced with risk points not recorded in the database, APES will intervene in the risk decision-making process, conduct a detailed analysis of the newly identified risk points, clarify specific prevention and control strategies and measures, and designate the departments and personnel involved in the follow-up. APES is also responsible for updating the risk prevention and control decision database to ensure that all prevention and control measures are based on the latest practical experience. Step 3.2: When Ac-PE is executed during pre-control, the pre-control decision output by Ac-PD will be converted into specific actions, and execution orders will be automatically issued to relevant departments and personnel to ensure that risk prevention measures are effectively implemented.

2. The deep foundation pit construction safety risk pre-control method based on cybernetics and dual system theory according to claim 1 is characterized in that: The detailed method of the central monitoring system for preliminary screening and preprocessing in step 1.2 includes data verification, data format table conversion and missing data processing; Data validation refers to checking the collected raw data, identifying and eliminating errors or outliers in data collection; data format standardization refers to converting data from various sources into a unified format to ensure data compatibility and consistency in subsequent processing stages; missing data processing refers to identifying missing data and applying appropriate techniques to fill or interpolate.

3. The method for pre-controlling safety risks of deep foundation pit construction based on cybernetics and dual-system theory according to claim 1 is characterized in that: The method for dynamically adjusting the risk warning threshold in step 2.1 is as follows: Set the initial warning standard thresholds basic Together with the monitoring data, it is continuously input into the Deep Foundation Pit Construction Risk Early Warning Standard Dynamic Feedback Mechanism DFM-REWS for prediction fitting; Calculate the deviation between real-time monitoring data and predicted risk value using data-driven models t , then, DFM-REWS evaluates the current performance P t , if the current performance P t Failure to achieve target performance P target , the difference ΔP between the two will be calculated t =P target -P t , and finally according to ΔP t The result of the feedback adjustment parameter β is used to dynamically update the warning threshold Threshold st+1 =Threshold st +β×ΔP t ; Among them, e t By comparing the actual monitoring value y t With the model f(x t θ t ) is determined by the expected value of the prediction, x t represents the input variable for real-time monitoring, θ t It is the parameter of the preset prediction model, which is calculated by the deviation e t The influence of P is quantified t , Threshold st is the risk warning threshold at the current stage, and Threshold st+1 is the updated threshold.

4. The method for pre-controlling safety risks of deep foundation pit construction based on cybernetics and dual-system theory according to claim 1 is characterized in that: In step 2.2, the Co-PR process involves first conducting a preliminary analysis of the collected root risk and status risk data, using a fuzzy logic algorithm to determine the nature and severity of the risk. The risk data is then fed into a dynamic risk rating model, which ranks and categorizes risks based on pre-set rating criteria and thresholds to ensure that the highest priority risks receive a timely response. The specific steps are as follows: First, the membership function evaluation indicators are defined: probability of occurrence P and severity of harm S. Both indicators are divided into three fuzzy sets: low, medium and high, and the membership function is defined as follows: For the probability of occurrence P: Low (R 1* ): Middle (R 2* ): High (R 3* ): For hazard level S: Low (R *1 ): Middle (R *2 ): High (R *3 ): Next, define the risk prevention level rules. Combining the fuzzy sets of probability of occurrence and degree of harm, define the following rules to evaluate the final risk prevention level: Rule 1: If the risk level classification result is R 11 、R 12 、R 21 , then the risk prevention level is 1 (L1); Rule 2: If the risk level classification result is R 13 、R 22 、R 31 , then the risk control level is 2 (L21); Rule 3: If the risk level classification result is R 23 、R 32 , then the risk control level is 3 (L3); Rule 4: If the risk level classification result is R 33 , then the risk control level is 4 (L4); Finally, reasoning and defuzzification are performed, and fuzzy reasoning is performed using the Mamdani reasoning method; For each rule, the output risk level is also expressed using fuzzy sets, and defuzzification uses the centroid method to calculate the final value of the risk level.

5. A system for implementing the deep foundation pit construction safety risk pre-control method based on cybernetics and dual-system theory as described in any one of claims 1 to 4, characterized in that: It includes a deep foundation pit construction safety risk identification and monitoring module based on sensor expansion, a deep foundation pit construction safety risk pre-control judgment and rating module based on controller expansion, and a deep foundation pit construction safety risk pre-control decision-making and execution module based on actuator and DST expansion; The sensor-based deep foundation pit construction safety risk identification and monitoring module is equipped with a risk identification Se-RI submodule and a risk monitoring Se-RM submodule; The risk identification submodule Se-RI identifies the safety risks of deep foundation pit construction and divides them into root risk RR and state risk SR; The risk monitoring submodule Se-RM monitors the identified risks, and the monitoring data is transmitted to the central monitoring system in real time. The central monitoring system screens and pre-processes the data, and finally forms a structured risk monitoring data set. The controller-based extended deep foundation pit construction safety risk pre-control discrimination and rating module includes a pre-control discrimination submodule Co-PJ and a pre-control rating submodule Co-PR; The pre-control discrimination submodule Co-PJ uses the deep foundation pit construction risk early warning standard dynamic feedback mechanism DFM-REWS to dynamically adjust the initial threshold based on real-time monitoring data. If the current stage risk exceeds the adjusted threshold, the pre-control rating submodule Co-PR will be automatically activated; The pre-control rating submodule Co-PR rates risk points according to severity, classifies and prioritizes various risks through a hierarchical rating mechanism, and uses a decision tree and risk matrix to guide the formulation of risk management strategies; The deep foundation pit construction safety risk pre-control decision-making and execution module based on the actuator and DST expansion includes a pre-control decision-making submodule Ac-PD and a pre-control execution submodule Ac-PE. In the pre-control decision-making submodule Ac-PD, an intuitive heuristic execution system IHES and an analytical processing execution system APES are developed based on DST. IHES simulates human intuitive judgment to quickly screen and apply existing risk control strategies. When faced with risk points not recorded in the database, APES will intervene in the risk decision-making process. In the pre-control execution submodule Ac-PE, the pre-control decisions output by Ac-PD will be converted into specific actions, and execution orders will be automatically issued to relevant departments and personnel to ensure that risk prevention measures are effectively implemented.

Images