Slurry shield slurry discharge pipeline stagnant discharge early warning method based on multi-source data fusion and dynamic risk evolution

Abstract

The invention discloses a slurry shield slurry discharge pipeline stagnant discharge early warning method based on multi-source data fusion and dynamic risk evolution, and the method comprises the steps: collecting multi-source monitoring data of a slurry discharge pipeline, and obtaining a comprehensive feature data set through multi-scale feature extraction; analyzing and calculating a time-varying weight based on the parameter coupling degree, constructing a risk evolution graph, and quantifying a risk evolution index; dynamically calculating a self-adaptive early warning threshold value in combination with the construction working condition; and performing multi-level early warning and risk traceability analysis according to the risk level. According to the method, the slurry discharging system is regarded as a nonlinear power system, the risk evolution trajectory is tracked through the phase-space reconstruction technology, the problems of unclear parameter coupling mechanism, risk evolution process deficiency, early warning threshold staticization and the like of a traditional method are solved, early accurate early warning and traceability diagnosis of the stagnant discharging risk are achieved, and the method is suitable for large-scale popularization and application. And the safety and efficiency of slurry shield construction are improved.

Description

Technical Field

[0001] This invention relates to the field of shield tunneling safety monitoring technology, and in particular to an early warning method for slurry discharge pipeline stagnation in slurry shield tunnels based on multi-source data fusion and dynamic risk evolution. Background Technology

[0002] As underground space development in my country, including urban subways, river-crossing tunnels, and integrated utility tunnels, extends to deeper and more complex strata, slurry shield tunneling machines (TBMs) are widely used due to their excellent adaptability to high water pressure and soft, uneven strata. In slurry shield tunneling, the slurry discharge pipe serves as the sole channel for discharging the slurry-excavation mixture, and its unobstructed flow directly affects the continuity of tunneling, the stability of the excavation face, and construction safety. However, engineering practice shows that slurry pipe blockage (i.e., a decrease in slurry flow velocity, sedimentation of slurry particles, and eventual blockage within the pipe) is one of the most common and dangerous failures in slurry shield tunneling.

[0003] At present, the following methods are mainly used for monitoring and early warning of slurry discharge pipeline stagnation in engineering: (1) Monitoring based on a single parameter threshold, such as monitoring whether the slurry discharge velocity is lower than the experience value or whether the slurry discharge pressure exceeds the set upper limit; (2) Comprehensive judgment based on human experience, relying on operators to make subjective assessments by observing changes in flow meters and pressure gauges, combined with pipeline vibration and sound; (3) Early warning model based on simple multi-parameter combination, such as considering a few parameters such as flow velocity and density at the same time.

[0004] However, these existing methods reveal the following deep-seated technical shortcomings when dealing with complex and ever-changing real-world engineering conditions: First, the nonlinear coupling mechanism between parameters has not been fully revealed. The solid-liquid two-phase flow in the slurry discharge pipeline is a complex dynamic system, with strong nonlinear interactions between parameters such as flow velocity, pressure, density, viscosity, and particle characteristics. Most existing methods simply superimpose or linearly combine these parameters, failing to quantify the dynamic coupling relationship between parameters and their time-varying contribution to the risk of sludge retention, resulting in poor adaptability of the early warning model when operating conditions change.

[0005] Secondly, the dynamic evolution of pipeline stagnation risk has not been effectively captured. Pipeline stagnation does not occur instantaneously, but is a dynamic process in which risk gradually accumulates and may eventually undergo a sudden change. Existing static threshold judgment methods can only provide a binary "yes / no" result, which cannot reflect the accumulation speed and evolution trend of risk. Therefore, they cannot achieve early warning, nor can they predict the timing of sudden changes in risk.

[0006] Third, the adaptive capacity to geological differences is insufficient. In different strata (such as cohesive mudstone, sand and gravel layers, and composite strata), the morphology, gradation, and interaction mechanisms of soil particles with mud vary significantly. Most existing early warning methods use fixed thresholds and lack the ability to dynamically adjust early warning standards according to stratum characteristics, leading to increased false alarm and missed alarm rates in specific strata.

[0007] Finally, the ability to identify early, subtle signs is weak. Before discharge occurs, the flow field within the pipeline will exhibit microscopic instability signs such as changes in flow velocity fluctuations and increased pressure pulsations. These nonlinear characteristics are difficult to extract effectively using traditional time-domain analysis methods, resulting in existing methods often providing delayed warnings of discharge and missing the optimal intervention window.

[0008] In summary, developing an early warning method for slurry drainage pipe stagnation that can quantify parameter coupling relationships, track risk evolution processes, adapt to geological changes, and capture early signs has become a pressing technical challenge for the safe and efficient construction of slurry shield tunnels. This invention is proposed against this backdrop. Summary of the Invention

[0009] The purpose of this invention is to provide an early warning method for slurry discharge pipeline stagnation in slurry shield tunneling based on multi-source data fusion and dynamic risk evolution, in order to solve the aforementioned problems existing in the prior art.

[0010] The technical solution, a method for early warning of slurry discharge stagnation in slurry shield tunneling pipes based on multi-source data fusion and dynamic risk evolution, includes the following steps: Multi-source monitoring data of the slurry discharge pipeline were collected, and multi-scale feature extraction was performed to obtain a comprehensive feature dataset; Based on the comprehensive feature dataset, parameter coupling degree analysis is performed and time-varying weights are calculated to obtain the parameter weight set; Based on the parameter weight set and the comprehensive feature dataset, a risk evolution map is constructed to obtain a risk evolution index set; Based on the risk evolution index set and construction condition data, dynamic early warning thresholds are calculated to obtain adaptive early warning thresholds. Based on the risk evolution indicator set and adaptive early warning threshold, multi-level early warning judgment and risk source analysis are carried out, and early warning results and handling suggestions are output.

[0011] Beneficial effects: This invention effectively solves the problems of delayed early warning, numerous false alarms and missed alarms, and reliance on experience in traditional methods, providing a reliable technical means to ensure the safety of slurry shield tunneling and improve tunneling efficiency. Attached Figure Description

[0012] Figure 1 This is an overall flowchart of an embodiment of the present invention.

[0013] Figure 2This is a flowchart of multi-source parameter acquisition and multi-scale feature extraction in an embodiment of the present invention.

[0014] Figure 3 This is a flowchart of parameter coupling analysis and adaptive weight allocation in an embodiment of the present invention.

[0015] Figure 4 This is a flowchart of the construction of the lag risk evolution map according to an embodiment of the present invention.

[0016] Figure 5 This is a flowchart illustrating the dynamic adjustment of the adaptive early warning threshold in an embodiment of the present invention.

[0017] Figure 6 This is a flowchart of the multi-level early warning and source tracing analysis in an embodiment of the present invention. Detailed Implementation

[0018] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0019] This invention elevates the early warning system for slurry discharge pipeline stagnation in slurry shield tunneling from static threshold judgment to the level of dynamic risk evolution tracking and quantification. Treating the slurry discharge system as a nonlinear dynamic system, it uses phase space reconstruction technology to fuse multiple monitoring parameters into a traceable risk state space, and calculates dynamic indices such as the Lyapunov exponent to quantify system stability. This invention protects the core idea of ​​this method of converting physical parameters to a risk state space and utilizing dynamic indices for early warning.

[0020] The time-varying coupling degree analysis and adaptive weight allocation model between parameters does not use fixed weights or simple linear superposition of parameters. Instead, it calculates the nonlinear coupling relationship between parameters (partial correlation coefficient, principal component model) in real time and dynamically allocates weights based on coupling degree, sensitivity, and coefficient of variation. This dynamic, data-driven relationship mining and weight allocation algorithm ensures that the early warning model can adapt to different operating conditions.

[0021] An adaptive early warning threshold dynamic adjustment mechanism integrating formation identification and operating condition correction: The early warning threshold is not a fixed value, but a function that dynamically changes with formation type, construction conditions, and equipment wear. This multi-level, adaptive threshold calculation model is protected to improve the method's universality and accuracy.

[0022] A multi-level early warning and risk tracing method based on contribution analysis: While issuing an early warning, it can locate the main risk contribution parameters and provide targeted handling suggestions, achieving a leap from alarm to diagnosis. This integrated early warning and risk tracing functionality is implemented by protecting the underlying logic.

[0023] Specifically, such as Figure 1 As shown, a method for early warning of slurry discharge stagnation in slurry shield tunneling pipelines based on multi-source data fusion and dynamic risk evolution includes: S1. Multi-source parameter acquisition and multi-scale feature extraction, such as Figure 2 As shown, it includes the following steps: S1.1: Collect real-time data from multiple sources. In a preferred embodiment, four types of raw data are collected in real time through a sensor network deployed on the tunnel boring machine and the slurry discharge pipe: fluid dynamic parameters (slurry velocity v(t), pressure P(t), density ρ, viscosity μ), particle characteristic parameters (particle density ρp, median particle size d50, non-uniformity coefficient Cu), pipe structure parameters (pipe inner diameter D, bend ratio R / D, pipe inclination angle θ), and construction status parameters (propulsion speed Vp, cutterhead torque T).

[0024] S1.2: Perform data cleaning and standardization preprocessing. Further, the collected raw data (especially high-frequency data v(t) and P(t)) are subjected to moving average filtering to remove noise interference; and all parameters are normalized to eliminate the influence of dimensions and obtain a standardized parameter sequence.

[0025] S1.3: Perform time-domain feature extraction. Further, for each standardized parameter sequence, within a set time window, calculate its mean, variance, skewness, and kurtosis to obtain a time-domain feature vector describing the central tendency, dispersion, and distribution pattern of the data.

[0026] S1.4: Perform frequency domain feature extraction. Further, perform a fast Fourier transform on the high-frequency acquired flow velocity v(t) and pressure P(t) signals to convert them from the time domain to the frequency domain, calculate their dominant frequency components and the energy proportion of each frequency band, and obtain frequency domain feature vectors that reflect the flow periodicity and vibration characteristics.

[0027] S1.5: Perform nonlinear feature extraction. Further, phase space reconstruction technology is used to map one-dimensional time series parameters (such as flow velocity v(t)) to a high-dimensional phase space. In this space, the Lyapunov exponent, which characterizes the chaotic properties of the system, and the approximate entropy, which characterizes the sequence complexity, are calculated, and finally, the nonlinear feature vector is obtained.

[0028] S1.6: Generate a comprehensive feature set. Further, the time-domain, frequency-domain, and nonlinear feature vectors obtained in the above steps are concatenated and fused to form a multi-scale comprehensive feature dataset for subsequent risk analysis.

[0029] S2, Parameter Coupling Analysis and Adaptive Weight Allocation, such as Figure 3 As shown, it includes the following steps: S2.1: Constructing a parameter coupling degree matrix. In a preferred embodiment, based on a multi-scale comprehensive feature dataset, the partial correlation coefficient between any two parameters is calculated to eliminate the indirect influence of other parameters, accurately measure the direct correlation strength between parameters, and form a coupling degree matrix C that reflects the nonlinear coupling relationship between parameters.

[0030] S2.2: Perform principal component analysis to extract coupling patterns. Further, perform principal component analysis on the coupling degree matrix C to identify the main coupling patterns that can explain the cooperative variation of most parameters, thus obtaining a set of principal component coupling patterns.

[0031] S2.3: Calculate the time-varying weights of the parameters. Further, within the sliding time window, comprehensively analyze the sensitivity S_i(t) (the degree of influence on the risk indicator) of each parameter, its comprehensive coupling degree C_i(t) in the coupling matrix C, and its own coefficient of variation V_i(t). Calculate using the weighting formula w_i(t) = α·S_i(t) + β·C_i(t) + γ·V_i(t), ultimately outputting a set of time-varying parameter weights.

[0032] S3, Construction of the evolution map of delayed discharge risk, such as Figure 4 As shown, it includes the following steps: S3.1: Reconstructing the risk state phase space. In a preferred embodiment, state variables are selected from key parameters (such as flow rate, pressure, and viscosity), and the one-dimensional time series is embedded into a high-dimensional phase space using the delayed coordinate method to construct a risk state phase space that can describe the dynamic characteristics of the system.

[0033] S3.2: Tracking the evolution trajectory of risk. Further, in the risk state phase space, the system state at each moment is represented as a point, and these points are connected in chronological order to form the system risk evolution trajectory.

[0034] S3.3: Calculate the dynamic stability index. Further, along the system risk evolution trajectory, calculate the maximum Lyapunov exponent λ(t) at each point. This exponent is used to quantify the system's sensitivity to initial conditions, i.e., the degree of chaos or stability of the system, to obtain the Lyapunov exponent sequence.

[0035] S3.4: Quantitative Risk Evolution Indicators. Further, the Lyapunov index sequence is integrated to calculate the risk accumulative degree A(t), characterizing the cumulative effect of risk; the second derivative of the risk accumulative degree A(t) is calculated to obtain the risk mutation degree M(t), characterizing the accelerated deterioration trend of risk; and the system resilience R(t) is calculated based on the change in risk accumulative degree, characterizing the system's self-recovery ability. Finally, these indicators are integrated into a risk evolution indicator set.

[0036] S4. Adaptive early warning threshold dynamic adjustment, such as Figure 5 As shown, it includes the following steps: S4.1: Determine the basic warning threshold. In a preferred embodiment, a basic warning threshold Π_base applicable to standard conditions is determined based on a large amount of historical normal operating condition data and laboratory tests.

[0037] S4.2: Calculate the formation correction term. Further, by analyzing parameters such as cutterhead torque T and propulsion speed Vp in real time, the formation identification algorithm is invoked to determine the current formation type (e.g., cohesive mudstone, sand and gravel), and the corresponding formation correction coefficient ΔΠ_geo is retrieved from the preset knowledge base.

[0038] S4.3: Calculate the working condition correction term. Further, based on current construction status parameters such as advance speed, cutterhead torque, and mud pump power, the working condition correction coefficient ΔΠ_oper is calculated using a pre-trained regression model.

[0039] S4.4: Calculate the aging correction term. Further, based on the cumulative tunneling mileage or time of the tunnel boring machine, and combined with the pipeline wear model, calculate the aging correction coefficient ΔΠ_time caused by factors such as increased pipeline roughness.

[0040] S4.5: Synthetic dynamic early warning threshold. Further, the above correction coefficients are added to the base threshold. Ultimately, dynamic early warning thresholds applicable to the current specific time and conditions are generated. .

[0041] S5, multi-level early warning and source tracing analysis, such as Figure 6 As shown, it includes the following steps: S5.1: Determine the risk level. In a preferred embodiment, the real-time risk evolution index obtained in S3 (such as the distance between the current state point and the stable attractor) is combined with the dynamic early warning threshold generated in S4. The system compares the risk levels and determines the current risk level (attention level, warning level, alert level, critical level) based on preset interval rules.

[0042] S5.2: Perform risk source analysis. Further, based on the time-varying parameter weight set and the real-time deviation of the parameters obtained in S2, calculate the contribution of each parameter to the current overall risk and identify a list of main risk source parameters.

[0043] S5.3: Output early warning and diagnostic results. Further, based on the risk level, trigger corresponding audible and visual alarms, interface prompts, or control commands. Simultaneously, combining the list of main risk source parameters, match and output targeted handling suggestions from the countermeasures knowledge base, completing the entire process from early warning to diagnosis.

[0044] The preferred embodiments of the present invention have been described in detail above. However, the present invention is not limited to the specific details in the above embodiments. Within the scope of the technical concept of the present invention, various equivalent transformations can be made to the technical solutions of the present invention, and these equivalent transformations all fall within the protection scope of the present invention.

Claims

1. A method for early warning of sludge discharge stagnation in slurry shield tunneling pipelines based on multi-source data fusion and dynamic risk evolution, characterized in that, include: Multi-source monitoring data of the slurry discharge pipeline were collected, and multi-scale feature extraction was performed to obtain a comprehensive feature dataset; Based on the comprehensive feature dataset, parameter coupling degree analysis is performed and time-varying weights are calculated to obtain the parameter weight set; Based on the parameter weight set and the comprehensive feature dataset, a risk evolution map is constructed to obtain a risk evolution index set; Based on the risk evolution index set and construction condition data, dynamic early warning thresholds are calculated to obtain adaptive early warning thresholds. Based on the risk evolution indicator set and adaptive early warning threshold, multi-level early warning judgment and risk source analysis are carried out, and early warning results and handling suggestions are output.

2. The method according to claim 1, characterized in that, The multi-source monitoring data collected from the slurry discharge pipeline is subjected to multi-scale feature extraction, including: Collect raw data on fluid dynamics parameters, particle characteristic parameters, pipeline structural parameters, and construction status parameters; The original data is preprocessed by moving average filtering and normalization to obtain a standardized parameter sequence; Temporal features are extracted from the standardized parameter sequence, and the mean, variance, skewness, and kurtosis are calculated to obtain the temporal feature vector. Fast Fourier transform is performed on the high-frequency acquired flow velocity and pressure signals to extract the main frequency and frequency band energy distribution, and obtain the frequency domain feature vector; The phase space reconstruction technique is used to calculate the Lyapunov exponent and approximate entropy to obtain the nonlinear eigenvectors. By fusing time-domain feature vectors, frequency-domain feature vectors, and nonlinear feature vectors, a comprehensive feature dataset is obtained.

3. The method according to claim 1, characterized in that, The step of performing parameter coupling analysis and calculating time-varying weights based on a comprehensive feature dataset includes: Based on the comprehensive feature dataset, the partial correlation coefficients between parameters are calculated, and the parameter coupling matrix is ​​constructed. Principal component analysis is performed on the parameter coupling matrix to extract the principal component coupling patterns; Based on the principal component coupling mode, the sensitivity, coupling degree and coefficient of variation of each parameter are calculated; The weights of the time-varying parameters are calculated by weighting the sensitivity, coupling degree, and coefficient of variation, thus obtaining the parameter weight set.

4. The method according to claim 1, characterized in that, The construction of the risk evolution map based on the parameter weight set and the comprehensive feature dataset includes: Key parameters are selected as state variables from the comprehensive feature dataset, and the phase space is reconstructed using the delayed coordinate method; By tracing the motion trajectory of the system state points in the reconstructed phase space, the risk evolution trajectory can be obtained; The Lyapunov index sequence is calculated along the risk evolution trajectory to characterize the system stability; Based on the Lyapunov index sequence, risk accumulation, risk mutation and system resilience are calculated to obtain a set of risk evolution indicators.

5. The method according to claim 1, characterized in that, The calculation of dynamic early warning thresholds based on the risk evolution index set and construction condition data includes: The basic early warning threshold is determined based on historical normal operating condition data; Real-time analysis of cutterhead torque and feed speed; identification of current formation type; calculation of formation correction coefficient. Based on construction status parameters, the working condition correction coefficient is calculated through a regression model; Based on the cumulative tunneling mileage of the tunnel boring machine, the aging correction coefficient is calculated using the pipeline wear model. The adaptive early warning threshold is calculated by combining the basic early warning threshold, the formation correction coefficient, the operating condition correction coefficient, and the timeliness correction coefficient.

6. The method according to claim 1, characterized in that, The process of multi-level early warning judgment and risk source tracing analysis based on a risk evolution indicator set and adaptive early warning threshold includes: The risk evolution indicators are compared with the adaptive early warning threshold, and the risk level is determined according to the preset range. Based on the parameter weight set and the real-time parameter deviation, the contribution of each parameter to the overall risk is calculated. Identify key risk source parameters based on their contribution ranking; Based on the risk level and main risk source parameters, corresponding handling suggestions are matched from the knowledge base and output.

7. The method according to claim 2, characterized in that, The method of using phase space reconstruction to calculate the Lyapunov exponent and approximate entropy includes: The time delay is determined by mutual information, and the embedding dimension is determined by the spurious nearest neighbor method. Reconstruct the phase space based on time delay and embedding dimension; The average divergence rate of adjacent trajectories is calculated in the reconstructed phase space to obtain the Lyapunov exponent. Calculate the conditional probability of the sequence under pattern matching to obtain the approximate entropy.

8. The method according to claim 5, characterized in that, The real-time analysis of cutterhead torque and feed speed, and identification of the current formation type, includes: Collect timing data of cutterhead torque and feed speed; Calculate the fluctuation characteristics of the cutterhead torque and the stability index of the feed speed; Based on fluctuation characteristics and stability indicators, stratigraphic types are identified using a trained stratigraphic classification model. The corresponding correction coefficient is retrieved from the preset knowledge base based on the stratigraphic type.

9. The method according to claim 6, characterized in that, The step of comparing risk evolution indicators with adaptive early warning thresholds and determining the risk level based on a preset range includes: Four risk level ranges are defined: Attention Level, Warning Level, Alert Level, and Critical Level; When the risk indicator is at the attention level, only abnormal parameters are recorded; when it is at the warning level, an audible and visual alert is issued and parameter adjustments are recommended; when it is at the alarm level, construction parameters are forcibly intervened; when it is at the critical level, the machine is shut down immediately and the dredging procedure is initiated.

10. The method according to claim 1, characterized in that, It also includes a visual representation of the early warning results: The risk evolution trajectory is displayed in real time through a 3D interface; The parameter coupling relationship is shown in the form of a heatmap; Use trend charts to show the historical changes in warning levels; Highlight the parameters of the main risk sources and their contribution.

Images