A dynamic early warning system for earth heave during open caisson construction based on multi-parameter fusion

Abstract

The present invention relates to the field of civil engineering construction safety monitoring, and discloses a dynamic early warning system for earth heave during open caisson construction based on multi-parameter fusion, which includes a sensor array module, a data processing module, a dynamic parameter correlation analysis module, an adaptive covariance optimization module, a risk index generation module and an early warning output module; real-time data is collected through multi-source sensors to generate a parameter vector; the dynamic parameter correlation analysis module updates the dynamic matrix online to characterize the real-time correlation among earth pressure, displacement and water level; the adaptive covariance optimization module dynamically adjusts the covariance matrix to suppress noise interference; the risk index generation module calculates weights by fusing the dynamic matrix, the covariance inverse matrix and the parameter vector, and corrects the risk index in combination with the vibration state of construction machinery; the early warning output module triggers hierarchical alarms and feeds back to form a closed-loop optimization link. The present invention realizes real-time accurate early warning and adaptive prevention and control of earth heave risks.

Description

Technical Field

[0001] The present invention relates to the technical field of civil engineering construction safety monitoring, and specifically to a dynamic early warning system for earth heave during open caisson construction based on multi-parameter fusion. Background Art

[0002] As a core construction method for projects such as bridge foundations, underground pump stations, and tunnel shafts, open caisson construction achieves controlled sinking by overcoming soil frictional resistance through the self-weight of the caisson shaft or external loading. However, under complex geological conditions (such as soft soil, quicksand layer, water-rich strata) or construction disturbances (sudden loading, eccentric excavation), the pressure imbalance inside and outside the caisson wall is likely to trigger the earth heave phenomenon - the soil around the caisson instantly surges into the caisson, leading to caisson shaft inclination, ground settlement, and even collapse accidents. Traditional monitoring technologies have two major limitations:

[0003] Single-parameter threshold alarm mechanism: It relies on static thresholds of single parameters such as displacement and earth pressure to trigger early warnings, and cannot capture the multi-parameter coupling effects in the causes of earth heave (such as the synergistic effect of sudden water level drop and sudden earth pressure change), resulting in high false alarm rates and missed alarm rates.

[0004] Fixed-weight empirical model: Some improved technologies attempt to fuse parameters such as displacement, earth pressure, and water level, but use a fixed-weight allocation model preset before construction. Its weight values rely on manual experience and cannot dynamically respond to changes in geological conditions during construction (such as attenuation of cohesion, fluctuation of water content) and the influence of mechanical disturbances, resulting in significant early warning lag and difficulty in timely preventing and controlling the risk of earth heave.

[0005] The above defects make it difficult for the existing technology to meet the real-time and accurate early warning requirements for the risk of earth heave in complex construction environments. Therefore, there is an urgent need for an intelligent early warning system that can dynamically perceive the multi-parameter correlation, adaptively adjust the weight allocation, and suppress interference to improve the timeliness and reliability of earth heave prevention and control. Summary of the Invention

[0006] Aiming at the deficiencies of the existing technology, the present invention provides a dynamic early warning system for earth heave during open caisson construction based on multi-parameter fusion, which solves the problems of early warning lag, high misjudgment rate, and inability to dynamically adapt to geological conditions and construction disturbances caused by the traditional open caisson construction earth heave early warning method relying on single-parameter threshold alarm and fixed-weight empirical model.

[0007] To achieve the above objectives, the present invention is realized through the following technical solutions: A dynamic early warning system for earth heave during open caisson construction based on multi-parameter fusion, comprising:

[0008] A sensor array module for real-time collecting the earth pressure change rate, displacement change rate, water level change rate, and soil property parameters during open caisson construction;

[0009] A data processing module that normalizes the collected data and generates a normalized parameter vector;

[0010] A dynamic parameter correlation analysis module that online updates the dynamic matrix based on the normalized parameter vector and the change rate of soil property parameters, and characterizes the real-time correlation of earth pressure, displacement, and water level;

[0011] An adaptive covariance optimization module that dynamically adjusts the covariance matrix in combination with the risk gradient to suppress noise interference;

[0012] A risk index generation module that calculates weights based on the dynamic matrix and the covariance matrix, fuses the normalized parameter vector to generate a heaving soil risk index, and corrects the risk in combination with the state of construction machinery;

[0013] An early warning output module that triggers a graded alarm signal according to the corrected risk index, and feeds back data to the dynamic parameter correlation analysis module and the adaptive covariance optimization module for closed-loop optimization.

[0014] Preferably, the sensor array module includes:

[0015] Earth pressure sensors uniformly arranged along the circumferential direction of the caisson wall;

[0016] Displacement sensors installed at the top, middle, and bottom nodes of the well wall;

[0017] Underground water level monitors arranged around the caisson;

[0018] Soil property detectors buried in the construction area to obtain the internal friction angle, cohesion, and their change rates of the soil in real time.

[0019] Preferably, the dynamic matrix update method of the dynamic parameter correlation analysis module includes the following steps:

[0020] Receive the normalized parameter vector from the data preprocessing module;

[0021] Based on the recursive least squares method, online update the dynamic matrix :

[0022] ;

[0023] Where is the dynamic matrix at time characterizing the correlation of earth pressure, displacement, and water level; is the learning rate; is the Kalman gain matrix; is the prediction error vector; is the transpose symbol;

[0024] Combined with the cohesion change rate in the rate of change of soil property parameters and the water content change rate Modify the dynamic matrix update direction:

[0025] ;

[0026] Among them, 、 are the coupling coefficients of the cohesion and the water content change rate respectively, is the identity matrix, is the all-ones matrix.

[0027] Preferably, the initialization method of the dynamic matrix is:

[0028] Construct the initial dynamic matrix , indicating that the initial dynamic matrix is a 3-order real matrix, its diagonal elements are 1, and the off-diagonal elements , , are set according to the initial correlation of earth pressure, displacement, and water level;

[0029] The form of the initial matrix is:

[0030] ;

[0031] Among them, , and .

[0032] Preferably, the covariance matrix dynamic adjustment method of the adaptive covariance optimization module includes:

[0033] Receive the risk index gradient from the risk index generation module;

[0034] Adaptively adjust the forgetting factor according to the risk index gradient:

[0035] ;

[0036] Among them, is the initial forgetting factor, is the attenuation coefficient, is the risk index gradient;

[0037] Update the covariance matrix :

[0038] ;

[0039] Among them, is the normalized parameter vector, denote and the covariance matrix of denote the normalized parameter vector at the previous moment, denote the covariance matrix at the previous moment;

[0040] Output the updated covariance matrix to the risk index generation module.

[0041] Preferably, the calculation method of the risk index gradient is:

[0042] ;

[0043] wherein, and are the risk indices at the current moment and the previous moment respectively;

[0044] When , trigger the enhanced update of the covariance matrix:

[0045] ;

[0046] wherein, is the noise threshold, is the noise suppression coefficient, is the identity matrix.

[0047] Preferably, the calculation method of the heaving soil risk index of the risk index generation module includes:

[0048] Receive the dynamic matrix from the dynamic parameter correlation analysis module, the covariance matrix from the adaptive covariance optimization module, and

[0049] the normalized parameter vector from the data processing module;

[0050] ;

[0051] wherein, is the inverse matrix of the covariance matrix, and the weight vector satisfies , denotes the -th component of the weight vector;

[0052] Generate the heaving soil risk index :

[0053] ;

[0054] Among them, is the regularization coefficient, represents the transpose of the weight vector ; represents the Frobenius norm of the matrix.

[0055] Preferably, the method for combining the construction machinery status by the risk index generation module to perform risk correction includes:

[0056] Receiving the construction machinery vibration status signal from the sensor array module, and when the vibration amplitude exceeds the threshold, it is determined as mechanical operation interference;

[0057] Correcting the risk index:

[0058] ;

[0059] Among them, represents the measured value of the mechanical vibration acceleration at time is the preset vibration threshold, is the correction coefficient;

[0060] Outputting the corrected risk index to the early warning output module.

[0061] Preferably, the method for triggering hierarchical alarms of the early warning output module includes:

[0062] Receiving the corrected risk index from the risk index generation module ;

[0063] Triggering hierarchical alarm signals according to the preset risk threshold :

[0064] Level 1 alarm: When the corrected risk index is greater than or equal to the preset risk threshold and less than 1.2 times the preset risk threshold it is triggered;

[0065] Level 2 alarm: When the corrected risk index is greater than or equal to 1.2 times the preset risk threshold it is triggered;

[0066] No alarm: When the corrected risk index is less than the preset risk threshold the no-alarm state is maintained;

[0067] When a level 1 or level 2 alarm is triggered, the normalized parameter vector at the alarm time, the corrected risk index , the dynamic matrix and covariance matrix Synchronously feedback to the dynamic parameter correlation analysis module and the adaptive covariance optimization module for updating the real-time parameters of the dynamic matrix and the covariance matrix.

[0068] Preferably, the closed-loop optimization method includes:

[0069] After receiving the feedback data, the dynamic parameter correlation analysis module updates the dynamic matrix through reinforcement learning:

[0070] ;

[0071] wherein, is the incremental matrix based on the difference between the historical data;

[0072] After receiving the feedback data, the adaptive covariance optimization module resets the forgetting factor of the covariance matrix:

[0073] ;

[0074] and based on recalculates to suppress the influence of outdated data;

[0075] The updated and flow back to the risk index generation module to form a closed-loop optimization link.

[0076] The present invention provides a dynamic warning system for earth heaving during open caisson construction based on multi-parameter fusion. It has the following beneficial effects:

[0077] 1. Through multi-dimensional real-time acquisition of earth pressure, displacement, water level and soil property parameters, and combining dynamic matrix correlation analysis and covariance optimization, the present invention realizes the collaborative perception of soil mechanical behavior, hydrological conditions and mechanical disturbance during open caisson construction. Compared with the traditional single-parameter monitoring method, it can capture the potential causes of earth heaving risk more comprehensively, and significantly reduce the probability of missed reports and false alarms.

[0078] 2. The dynamic parameter correlation analysis module of the present invention uses the recursive least squares method to update the dynamic matrix online, and reflects the change of the correlation strength between earth pressure, displacement and water level parameters in real time. This mechanism can automatically adapt to the soil response characteristics of different construction stages, avoid the prediction deviation caused by the change of geological conditions in the traditional static model, and improve the timeliness of risk warning.

[0079] 3. The adaptive covariance optimization module of the present invention dynamically adjusts the forgetting factor of the covariance matrix to distinguish normal data fluctuations from abnormal noise interference, and dynamically corrects the contribution degree of historical data in combination with the risk gradient. This design effectively suppresses the influence of interference such as sensor noise and mechanical vibration on the calculation of the risk index, ensuring the stability of weight allocation and risk quantification.

[0080] 4. By introducing the vibration state signal of construction machinery, the system of the present invention can distinguish the real soil heave risk from mechanical operation interference, and dynamically scale the risk index using a correction coefficient. At the same time, the corrected risk index is fed back to the dynamic parameter correlation analysis module to suppress the abnormal update of the matrix caused by mechanical interference, forming a closed-loop control logic and enhancing the anti-interference ability of the system in complex construction environments.

[0081] 5. The warning output module triggers hierarchical alarms according to the corrected risk index, and synchronously feeds back the key parameters at the alarm moment to the correlation module. This mechanism enables the system to continuously optimize the dynamic matrix update rule and covariance adjustment strategy based on historical alarm data, realizing the upgrade from passive warning to active learning and enhancing the adaptability of long-term construction risk prevention and control. BRIEF DESCRIPTION OF THE DRAWINGS

[0082] Figure 1 It is a schematic diagram of the system architecture of the present invention. DETAILED DESCRIPTION OF THE INVENTION

[0083] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the drawings of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0084] Please refer to the attached Figure 1 , the present invention provides a dynamic warning system for soil heave in caisson construction based on multi-parameter fusion. The system realizes accurate warning of soil heave risk by collecting earth pressure, displacement, water level and soil parameters in real time, and combining dynamic correlation analysis and closed-loop optimization mechanism.

[0085] As Figure 1 shown, the system includes a sensor array module, a data processing module, a dynamic parameter correlation analysis module, an adaptive covariance optimization module, a risk index generation module and a warning output module. The following are the specific implementation details of each module:

[0086] Sensor Array Module

[0087] This module collaboratively collects data on earth pressure, displacement, water level, and soil parameters through multiple types of sensors, establishing a mapping relationship between physical quantities and digital signals, providing a raw data basis for subsequent parameter correlation analysis and risk calculation.

[0088] The sensor array module includes an earth pressure change rate acquisition unit, a displacement change rate acquisition unit, a water level change rate acquisition unit, and a soil property parameter acquisition unit. The earth pressure change rate acquisition unit exemplarily uses fiber Bragg grating pressure sensors, which are arranged equidistantly along the circumferential direction of the caisson cutting edge, with an installation spacing preferably of 2 meters and a sampling frequency set to 10 Hz. By measuring the pressure fluctuations at the contact surface between the cutting edge and the soil, the earth pressure change rate is obtained in real time. . In this unit, the data of adjacent sensors are processed by spatial weighted averaging, and the calculation formula is:

[0089] ;

[0090] where represents the measured value of the th sensor at moment, is the sampling interval, and is the total number of circumferential sensors. This calculation can eliminate the measurement deviation caused by local soil inhomogeneity and provide stable input for the dynamic parameter correlation analysis module.

[0091] The displacement change rate acquisition unit is realized by combining a laser rangefinder and an inclinometer. The laser rangefinder is vertically installed at the center of the caisson top, and the inclinometer is circumferentially arranged on the caisson sidewall, with a sampling frequency set to 5 Hz. By fusing the vertical displacement change and the inclination change , the comprehensive displacement change rate is calculated:

[0092] ;

[0093] where is the height of the caisson, is calculated from the adjacent sampling difference of the laser rangefinder, and is the change amount of the inclinometer measurement value. This formula uniformly quantifies the vertical sinking and tilting displacement, and can accurately characterize the overall displacement trend of the caisson, providing multi-dimensional displacement parameters for the risk index generation module.

[0094] The water level change rate acquisition unit is realized by a piezoresistive liquid level gauge. Two groups of sensors are arranged on both the inside and outside of the caisson, and the sampling period is set to 10 seconds. The water level difference is calculated as:

[0095] ;

[0096] where and respectively the water level heights inside and outside the well at a certain moment. The water level change rate is further obtained through differential calculation:

[0097] ;

[0098] This parameter can reflect the influence of groundwater flow velocity on soil stability. After normalization, it is input into the data processing module and participates in the weight calculation synergistically with soil pressure and displacement parameters.

[0099] The soil property parameter acquisition unit includes resistivity probes and moisture content sensors, which are arranged in a grid pattern 1 meter below the cutting edge, and the soil cohesion is collected every 30 minutes and moisture content . The cohesion is calculated through a resistivity inversion model. The moisture content is directly measured by a dielectric constant sensor, and the data is corrected by a temperature compensation algorithm.

[0100] The above parameters affect the dynamic matrix update through the soil state equation in the dynamic parameter correlation analysis module, and the specific manifestations are:

[0101] ;

[0102] This design enables the change of soil parameters to adjust the correlation strength between parameters in real time, enhancing the adaptability of the system to geological condition changes.

[0103] Preferably, all sensor data is transmitted to the data processing module through a hybrid network of RS-485 bus and LoRa wireless communication. Among them, the RS-485 bus is used for the wired stable transmission of soil pressure, displacement, and water level sensors, and LoRa is used for the data backhaul of soil parameter sensors and mobile monitoring points in the wireless coverage blind area.

[0104] Data processing module

[0105] This module receives the original data stream of the sensor array module, and through multi-level signal processing and normalization conversion, generates a standardized parameter vector, providing an input basis for dynamic parameter correlation analysis and risk calculation.

[0106] The data processing module first preprocesses the original signals of soil pressure, displacement, and water level. For soil pressure data, a moving average filter is used to eliminate high-frequency noise, and the filter window length is exemplarily set to 5 seconds, and the calculation formula is:

[0107] ;

[0108] Among them, is the number of sampling points within the window. This filtering process can suppress short-term mechanical vibration interference and retain the low-frequency trend component reflecting soil stability.

[0109] The displacement data includes vertical displacement and inclination measurement values, and the comprehensive displacement change rate needs to be calculated through fusion. First, the data of the laser rangefinder is denoised by wavelet thresholding. The sym4 wavelet basis is selected for 3-layer decomposition, and the hard thresholding processing formula is:

[0110] ;

[0111] where is the wavelet coefficient of the th layer, the threshold , is the estimated value of the noise standard deviation. This processing can effectively separate the true displacement signal from the measurement noise, and is especially suitable for the denoising requirements of non-stationary signals.

[0112] The change rate is calculated based on the preprocessed data. The change rate of earth pressure is calculated by first-order difference:

[0113] ;

[0114] where is the sampling interval. The displacement change rate is obtained by the difference of the fused displacement quantity:

[0115] ;

[0116] The change rate of water level is calculated by the symmetric difference method:

[0117] ;

[0118] This method can reduce phase delay, accurately capture water level mutation events, and provide real-time guarantee for the risk index generation module.

[0119] The normalization processing unit converts the change rates of each parameter into dimensionless quantities. The earth pressure normalization formula is:

[0120] ;

[0121] where and are the historical mean and standard deviation of the earth pressure change rate respectively, which are updated once an hour. The mean and standard deviation are calculated by sliding window statistics. The window length is exemplarily set to 24 hours, and the calculation formula:

[0122] ;

[0123] Similarly, the normalized values of displacement and water level are calculated as and This normalization method can eliminate the differences in the dimensions of parameters, enabling multi-source data to participate in matrix operations on a unified scale.

[0124] Finally, the normalized parameter vector is constructed as:

[0125] ;

[0126] This vector is transmitted to the dynamic parameter correlation analysis module through a standardization interface.

[0127] The processing of soil property parameters includes the fusion calculation of cohesion and water content .

[0128] The measured value of the resistivity probe is converted to cohesion through an inversion model:

[0129] ;

[0130] where and are calibration coefficients, and is the temperature compensation term.

[0131] The water content is measured by a dielectric constant sensor and corrected by the Topp formula.

[0132] The above parameters affect the matrix update process through the state equation in the dynamic parameter correlation analysis module.

[0133] Dynamic parameter correlation analysis module

[0134] This module updates the dynamic matrix online through the recursive least squares method, quantifies the real-time correlation intensity of earth pressure, displacement, and water level, and provides a dynamic weight basis for risk index calculation.

[0135] The core function of the dynamic parameter correlation analysis module is to capture the time-varying correlation characteristics of multiple parameters in the soil body system. To achieve this goal, first, define the dynamic matrix as a mathematical representation of the parameter correlation relationship. Its elements are set according to the prior data of the initial correlation of earth pressure (P), displacement (D), and water level (W). The specific rules are as follows:

[0136] Diagonal elements: , representing the dominant role of the historical values of the parameters themselves, which is in line with the inertial characteristics of soil parameters in the early stage of caisson construction.

[0137] Non-diagonal elements:

[0138] When there is a prior correlation, it is set according to the soil engineering investigation data or historical construction statistics results:

[0139] , is the parameter and 's initial correlation coefficient, with a value range of [-1, 1].

[0140] When there is no prior data, the default setting is:

[0141] , at this time degenerates into the identity matrix, reflecting the initial independence assumption of the parameters.

[0142] Among them, the sources of initialization data include:

[0143] Geological exploration data: Obtain the initial relationship between cohesion and water content through soil shear tests and indirectly deduce ;

[0144] Historical construction data: Extract the parameter covariance matrix from the database of similar projects and normalize it to the correlation coefficient;

[0145] Expert experience assignment: When quantitative data is lacking, it is qualitatively set by engineers according to the soil type.

[0146] This initialization strategy is based on the inertial characteristics of the soil body system. It is assumed that each parameter is mainly dominated by its own historical value in the initial state, which conforms to the physical reality that the parameters have relatively strong independence in the early stage of open caisson construction.

[0147] Aiming at the time-varying characteristics of the correlation relationship of soil parameters, the recursive least squares (RLS) algorithm is used to realize the online update of the matrix. This algorithm avoids the repeated storage of historical data through recursive calculation and meets the real-time requirements. The specific iterative process first calculates the prediction error vector:

[0148] ;

[0149] Among them is the normalized parameter vector output by the data processing module, is the sampling interval. The error vector reflects the prediction deviation of the current parameter correlation model and is used to drive the matrix correction.

[0150] The Kalman gain is calculated through the covariance inverse matrix as follows:

[0151] ;

[0152] Among them, the forgetting factor λ = 0.99 is used to balance the influence weights of historical data and new data and prevent excessive matrix fluctuations. The covariance inverse matrix adopts the rank-1 update rule:

[0153] ;

[0154] This update mechanism avoids matrix inversion operations through recursive calculations, significantly reducing the computational complexity.

[0155] The iterative correction formula for the dynamic matrix is:

[0156] ;

[0157] Learning rate controls the matrix update step size to prevent the model from becoming unstable due to excessive single - time errors. To ensure the physical interpretability of the matrix, the following constraint conditions are imposed:

[0158] 1. Autocorrelation constraint: The diagonal elements , ensuring the dominant position of the historical values of the parameters themselves;

[0159] 2. Direction consistency constraint: The signs of the non - diagonal elements remain the initial set directions to avoid sudden changes in the correlation relationship.

[0160] When the iteration result violates the constraint, the projection correction algorithm is activated:

[0161] ;

[0162] where is the Kronecker delta function, and this operation restricts the parameters to a reasonable range to maintain the stability of the model.

[0163] The soil property parameters affect the matrix update through the differential coupling term. The change rate of cohesion introduces a correction term:

[0164] ;

[0165] where α is the coupling coefficient, is the identity matrix. Similarly, the change rate of water content generates a correction amount through the all - ones matrix :

[0166] ;

[0167] The extended dynamic matrix update formula is:

[0168] ;

[0169] This design enables the changes in soil parameters to adjust the correlation weights in real - time, enhancing the adaptability of the system to dynamic changes in geological conditions.

[0170] The early warning feedback mechanism achieves closed-loop optimization by adjusting the learning rate. When receiving the signal from the early warning output module, the learning rate is dynamically adjusted to:

[0171] ;

[0172] where is the reference value. This mechanism compresses the learning rate when the risk rises, inhibits the interference of abnormal data on the correlation model, and at the same time retains the model's ability to track new trends.

[0173] Adaptive covariance optimization module

[0174] This module dynamically senses the change of the system risk state, establishes a closed-loop adjustment mechanism for the parameter covariance matrix, and provides noise suppression and statistical stability guarantee for the risk weight calculation.

[0175] The input data stream of the adaptive covariance optimization module includes the real-time risk index from the risk index generation module , and the normalized parameter vector output by the data processing module. By fusing the risk state and the parameter statistical characteristics, this module dynamically adjusts the covariance matrix update strategy, and the specific implementation process is as follows:

[0176] Step 1: Risk gradient calculation and state perception

[0177] After receiving the from the risk index generation module, first calculate the risk gradient to quantify the system state change rate:

[0178] ;

[0179] where and are the risk indices at the current moment and the previous moment respectively. This gradient value is transmitted to the dynamic parameter correlation analysis module through the feedback channel of the early warning output module, forming a cross-module state sharing mechanism. When , is the noise threshold; it is determined that the system is in a high-risk transient state, and the covariance matrix reinforcement update is triggered.

[0180] Step 2: Dynamic adjustment of forgetting factor

[0181] Based on the risk gradient, adaptively adjust the historical data weight coefficient :

[0182] ;

[0183] where is the initial forgetting factor, is the attenuation coefficient. When the risk gradient rises, Exponential decay to reduce the weight of the historical covariance matrix to make the system more sensitive to current parameter changes. This adjustment process forms a collaborative control with the learning rate of the dynamic parameter correlation analysis module to jointly suppress the interference of sudden disturbances on model updates.

[0184] Step 3: Iterative update of the covariance matrix

[0185] Update the covariance matrix using the normalized parameter vector :

[0186] ;

[0187] where is the normalized parameter vector, represents the covariance matrix between and represents the normalized parameter vector at the previous moment, represents the covariance matrix at the previous moment.

[0188] The real-time covariance term is calculated through a sliding window:

[0189] ;

[0190] where is the number of window samples, is the window mean. This calculation strategy matches the real-time requirement of the data processing module while retaining the temporal correlation of the parameters and avoiding the memory overhead caused by storing historical data.

[0191] Step 4: Noise suppression and matrix stability processing

[0192] When the risk gradient exceeds the limit, perform diagonal strengthening of the covariance matrix:

[0193] ;

[0194] where is the noise suppression coefficient, is the identity matrix. This operation suppresses the abnormal correlation calculation caused by sudden noise by enhancing the self-correlation weight of the parameters.

[0195] Step 5: Matrix verification and output

[0196] Perform symmetry and positive semi-definiteness verification on the updated covariance matrix:

[0197] Symmetry correction: For the asymmetric element , force symmetric processing:

[0198] ;

[0199] Semi - positive definite correction: truncating and projecting the negative eigenvalues:

[0200] ;

[0201] where is the eigenvalue diagonal matrix, is the eigenvector matrix. The corrected covariance matrix is output to the risk index generation module through the standardization interface, and its data structure is aligned with the matrix output of the dynamic parameter correlation analysis module to ensure the dimensional consistency of weight calculation.

[0202] Risk index generation module

[0203] This module constructs a closed - loop calculation model for the soil heave risk index by integrating the dynamic matrix correlation, parameter covariance characteristics, and the state of construction machinery, and realizes the multi - dimensional quantitative assessment of the soil instability risk.

[0204] The risk index generation module receives the dynamic matrix from the dynamic parameter correlation analysis module , the covariance matrix from the adaptive covariance optimization module , and the normalized parameter vector from the data processing module . The module processing flow is as follows:

[0205] Step 1: Weight vector calculation

[0206] First, calculate the parameter weight vector according to the dynamic matrix and the inverse covariance matrix:

[0207] ;

[0208] where the inverse covariance matrix is obtained through the reversibility guarantee algorithm:

[0209] ;

[0210] is the micro - perturbation coefficient, is the identity matrix. The weight vector needs to satisfy the normalization constraint:

[0211] ;

[0212] This constraint is achieved through post - processing scaling:

[0213] ;

[0214] The normalized weight vector For subsequent risk calculation.

[0215] Step 2: Generation of soil heaving risk index

[0216] Perform a linear combination of the weight vector and the parameter vector, and introduce a regularization term for the dynamic matrix variation:

[0217] ;

[0218] Where is the regularization coefficient, represents the weight vector transpose; represents the Frobenius norm of the matrix.

[0219] Step 3: Correction of construction machinery status

[0220] Receive the mechanical vibration acceleration signal from the sensor array module , when , it is determined that the mechanical operation causes significant disturbance to the soil mass. The risk index correction formula is:

[0221] ;

[0222] is the correction coefficient, and this value is exemplarily set according to engineering experience.

[0223] Step 4: Data encapsulation and output

[0224] The corrected risk index is encapsulated into a standardized data packet and output to the early warning output module.

[0225] Early warning output module

[0226] This module constructs a closed-loop control system for risk early warning and model self-update through a hierarchical risk threshold determination and multi-module collaborative optimization mechanism, and realizes real-time monitoring and adaptive adjustment of the construction safety status.

[0227] The early warning output module receives the corrected risk index output by the risk index generation module , and obtains the of the dynamic parameter correlation analysis module, the of the adaptive covariance optimization module, and the of the data processing module. The module processing flow and interaction mechanism are as follows:

[0228] Step 1: Trigger hierarchical alarm

[0229] Based on the preset risk threshold trigger a three-level alarm strategy:

[0230] 1. No alarm status: When the system maintains normal operation and only records data;

[0231] 2. First-level alarm: When the audible and visual alarm is activated with a yellow warning;

[0232] 3. Second-level alarm: When an orange alarm is triggered.

[0233] Step 2: Alarm data encapsulation

[0234] When an alarm is triggered, a four-dimensional data packet is encapsulated and fed back to the dynamic parameter correlation analysis module and the adaptive covariance optimization module:

[0235] ;

[0236] Step 3: Closed-loop optimization execution

[0237] After the dynamic parameter correlation analysis module receives it updates the dynamic matrix:

[0238] ;

[0239] Among them, the incremental matrix is calculated through the historical data difference:

[0240] ;

[0241] Among them, is the sliding window length. The function maps the risk index to the interval [-1, 1] to suppress excessive correction amounts.

[0242] After the adaptive covariance optimization module receives the data, it resets the forgetting factor:

[0243] ;

[0244] At the same time, based on it recalculates the covariance matrix:

[0245] ;

[0246] This operation reduces the weight of historical data and enhances the representation of the current state.

[0247] Step 4: Optimized parameter feedback

[0248] The updated and It flows back to the risk index generation module through a standardized interface, participates in the weight calculation of the next cycle, forms a closed-loop feedback link from risk warning to model parameter update, and improves the adaptability of the system to the dynamic changes of the construction environment.

[0249] Generally speaking, the present invention collects the change rate data of earth pressure, displacement, water level and soil parameters in real time through a sensor array, generates a parameter vector after normalization by a data processing module, combines the recursive least squares method of a dynamic parameter correlation analysis module to online update a dynamic matrix reflecting the real-time correlation between parameters, and the risk gradient-driven noise suppression mechanism of an adaptive covariance optimization module to generate a heaving soil risk index integrating dynamic weights and covariance inverse matrices, and dynamically corrects it based on the vibration state of construction machinery; triggers hierarchical alarms through an early warning output module, and at the same time feeds back the corrected risk index and key matrix to the dynamic parameter correlation module and the covariance optimization module to form a full-process adaptive early warning system, realizing the collaborative perception of sudden earth pressure changes, abnormal displacements and water level fluctuations and the precise prevention and control of heaving soil risks.

[0250] Although the embodiments of the present invention have been shown and described, for those of ordinary skill in the art, it can be understood that various changes, modifications, substitutions and variations can be made to these embodiments without departing from the principle and spirit of the present invention, and the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A dynamic early warning system for earth heave during open caisson construction based on multi-parameter fusion, characterized in that, Including: A sensor array module for real-time collection of the rate of change of earth pressure, displacement rate, water level change rate, and soil property parameters during open caisson construction; A data processing module for normalizing the collected data and generating a normalized parameter vector; A dynamic parameter correlation analysis module, based on the normalized parameter vector and the change rate of soil property parameters, updates the dynamic matrix online through the recursive least squares method to characterize the real-time correlation of earth pressure, displacement, and water level; An adaptive covariance optimization module that dynamically adjusts the covariance matrix in combination with the risk gradient to suppress noise interference; A risk index generation module that calculates weights based on the dynamic matrix and the covariance matrix, fuses the normalized parameter vector to generate a heaving soil risk index, and performs risk correction in combination with the construction machinery status; An early warning output module that triggers a graded alarm signal according to the corrected risk index, and feeds back data to the dynamic parameter correlation analysis module and the adaptive covariance optimization module for closed-loop optimization.

2. The dynamic early warning system for earth heave during open caisson construction based on multi-parameter fusion according to claim 1, characterized in that, The sensor array module includes: Earth pressure sensors evenly arranged circumferentially along the open caisson wall; Displacement sensors installed at the top, middle, and bottom nodes of the well wall; Underground water level monitors arranged around the open caisson; Soil property detectors buried in the construction area to obtain the internal friction angle, cohesion, and their change rates of the soil in real time.

3. The dynamic early warning system for earth heaving during open caisson construction based on multi-parameter fusion according to claim 1, wherein The dynamic matrix update method of the dynamic parameter correlation analysis module includes the following steps: Receiving the normalized parameter vector from the data preprocessing module; Online update of the dynamic matrix based on the recursive least squares method is performed as follows: ； Among them, is the dynamic matrix at time , representing the correlation of earth pressure, displacement and water level; is the learning rate; is the Kalman gain matrix; is the prediction error vector; is the transpose symbol; Combined with the rate of change of cohesion in the soil property parameter change rate and the rate of change of water content Modify the dynamic matrix update direction: ； Among them, , are the coupling coefficients of cohesion and water content change rate respectively, is the identity matrix, is the all-one matrix.

4. The dynamic early warning system for earth heave during open caisson construction based on multi-parameter fusion according to claim 3, characterized in that, The initialization method of the dynamic matrix is: Construct the initial dynamic matrix , denote the initial dynamic matrix as a 3-order real matrix, whose diagonal elements are 1 and the non-diagonal elements are set according to the initial correlations of earth pressure, displacement and water level.

5. The dynamic warning system for earth heave during open caisson construction based on multi-parameter fusion according to claim 4, wherein The covariance matrix dynamic adjustment method of the adaptive covariance optimization module includes: Receiving the risk index gradient from the risk index generation module; Adaptive adjustment of forgetting factor according to risk index gradient : ； Among them, is the initial forgetting factor, is the attenuation coefficient, is the risk index gradient; Update covariance matrix : ； Among them, is the normalized parameter vector, denotes the covariance matrix of and represents the normalized parameter vector at the previous moment, represents the covariance matrix at the previous moment; Output the updated covariance matrix to the risk index generation module.

6. The dynamic warning system for earth heave during open caisson construction based on multi-parameter fusion according to claim 5, characterized in that, The calculation method of the risk index gradient is: ； Among them, and are the risk indices at the current moment and the previous moment respectively; When occurs, trigger the enhanced update of the covariance matrix: ； wherein, is the noise threshold,[ is the noise suppression coefficient,[ is the identity matrix.[ 7. The dynamic warning system for earth heaving during open caisson construction based on multi-parameter fusion according to claim 6, characterized in that The calculation method of the heaving soil risk index of the risk index generation module includes: Receive the dynamic matrix from the dynamic parameter correlation analysis module , the covariance matrix of the adaptive covariance optimization module , and the normalized parameter vector of the data processing module ; Calculate the weight vector : ； wherein, is the inverse matrix of the covariance matrix, and the weight vector satisfies , represents the weight vector the -th component; Generate the risk index of soil heave : ； Among them, is the regularization coefficient, represents the transpose of the weight vector ; represents the Frobenius norm of the matrix.

8. The dynamic warning system for earth heave during open caisson construction based on multi-parameter fusion according to claim 7, wherein The method of combining the construction machinery status for risk correction by the risk index generation module includes: Receiving the construction machinery vibration status signal from the sensor array module, and when the vibration amplitude exceeds the threshold, it is determined as mechanical operation interference; Correcting the risk index: ； Among them, represents the measured value of mechanical vibration acceleration at a moment, is a preset vibration threshold, is a correction factor; Output the corrected risk index to the early warning output module.

9. The dynamic early warning system for earth heave during open caisson construction based on multi-parameter fusion according to claim 8, wherein, The graded alarm triggering method of the early warning output module includes: Receive the corrected risk index from the risk index generation module ; According to a preset risk threshold Trigger a graded alarm signal: Level 1 Alarm: When the corrected risk index is greater than or equal to the preset risk threshold and less than 1.2 times the preset risk threshold it is triggered; Secondary alarm: When the corrected risk index is greater than or equal to 1.2 times of the preset risk threshold it is triggered; No alarm: When the corrected risk index is less than the preset risk threshold the no-alarm state is maintained; When a first-level or second-level alarm is triggered, the normalized parameter vector at the alarm moment , the corrected risk index , the dynamic matrix and the covariance matrix are synchronously fed back to the dynamic parameter correlation analysis module and the adaptive covariance optimization module for updating the real-time parameters of the dynamic matrix and the covariance matrix.

10. The dynamic early warning system for soil heave during open caisson construction based on multi-parameter fusion according to claim 9, characterized in that, The closed-loop optimization method includes: After receiving the feedback data, the dynamic parameter correlation analysis module updates the dynamic matrix through reinforcement learning: ； Among them, is the base and the incremental matrix of the difference from historical data; After receiving the feedback data, the adaptive covariance optimization module resets the forgetting factor of the covariance matrix: ； and based on recalculate to suppress the impact of obsolete data; Updated and flow back to the risk index generation module to form a closed-loop optimization link.

Images