A construction monitoring system and method for geothermal drilling

Abstract

The present application relates to geothermal drilling monitoring management technical field, specifically for a kind of construction monitoring system and method for geothermal drilling;The method comprises: by constructing state variable set, using perturbation derivative method to quantify structure perturbation intensity and normalizing, define state transition tensor mapping behavior influence coefficient;In combination with the high-order derivative of perturbation trend function and behavior collaborative change matrix, early warning is carried out by generating offset risk factor fusion;Drilling area is divided into three-dimensional grid, local stress potential energy is calculated, risk distribution map is generated by dynamic sensitive zone adjustment weight and smoothing filtering;For medium-high risk area, risk potential energy centroid coordinates and gradient field inversion intervention vector are used, closed-loop dynamic regulation and control are carried out;Through multi-period risk potential energy tensor analysis, self-healing index and stable cone area matching degree are calculated, system self-healing ability is evaluated and feedback control strategy is fed back.The present application forms closed-loop monitoring and active control system of geothermal drilling nonlinear risk.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of geothermal drilling monitoring and management, and particularly relates to a construction monitoring system and method for geothermal drilling. BACKGROUND

[0002] Geothermal drilling construction faces extreme working conditions such as high temperature and high pressure, complex geological structure, etc. The nonlinear response mechanism of the underground structure to the construction behavior is difficult to accurately characterize in real time, especially the dynamic correlation of the multivariate coupling disturbance such as drilling pressure and mud parameters and the formation response needs quantitative analysis technology, and a closed-loop monitoring mechanism for risk early identification, multi-parameter collaborative regulation and system self-healing ability evaluation needs to be established to ensure construction safety.

[0003] The Chinese invention patent application with the publication number CN118212098A discloses a geological exploration drilling construction safety management system based on the Internet of Things, which collects construction data of the geological exploration drilling construction site, transmits the data to the control terminal using the Internet of Things technology to monitor, control and improve the safety of the geological exploration drilling engineering construction, and conducts regular training and detection to improve the construction safety and engineering quality. The signal correction module is used to correct the detection signal accurately, improve the accuracy of the detection signal, and improve the reliability of the detection result.

[0004] However, in view of the challenges of high temperature and high pressure, complex geological structure and multi-physical field coupling faced by geothermal drilling construction, the current technology has limitations in accurately quantifying the multi-parameter collaborative disturbance mechanism, the hidden risk evolution path and the three-dimensional space risk distribution, and it is urgent to develop a comprehensive monitoring system that integrates high-dimensional state modeling, nonlinear trend early warning and intelligent feedback regulation to improve the risk prediction and active control ability under complex working conditions. SUMMARY

[0005] The present application aims to solve the problems in the background art and proposes a construction monitoring system and method for geothermal drilling.

[0006] The technical solution of the present application is a construction monitoring method for geothermal drilling, which includes the following specific implementation steps:

[0007] S1, calculate the disturbance intensity based on the real-time collected drilling parameters and perform normalization processing, construct a time series state transition tensor through the outer product of the state vector and the disturbance vector, detect abnormal fluctuations using the Frobenius norm difference between consecutive tensors, and map the difference value to a behavior influence coefficient;

[0008] S2, calculate the disturbance velocity and acceleration based on the high-order difference analysis of the disturbance trend function, capture the multi-parameter co-motion displacement characteristics through the behavior variable co-motion change matrix and the time sliding window function, fuse the behavior influence coefficient, the disturbance acceleration and the co-motion change intensity to generate the displacement risk factor, and dynamically construct the risk sensitive band according to the historical data sliding window statistics;

[0009] S3, three-dimensional gridding of the drilling area, calculating the local stress potential energy based on the nonlinear coupling function of state parameters and disturbance characteristics, dynamically adjusting the weight coefficient according to the displacement risk factor, optimizing the potential energy distribution through spatial smoothing filtering, and finally dividing the risk level according to the threshold value;

[0010] S4, extract the medium and high risk grid and calculate the weighted potential energy centroid, inverse the intervention vector based on the risk gradient field and response sensitivity matrix, project it to the controllable parameter domain to update the control quantity, and set the observation window to delay track the potential energy change;

[0011] S5, calculate the self-healing index of each grid and the regional average recovery index according to the risk potential energy tensor, define the maximum risk decline rate to construct the stable cone area and calculate the trajectory falling proportion, combine the recovery index and the cone proportion to grade the system self-healing ability, and dynamically adjust the regulation strategy according to the level.

[0012] Preferably, the construction process of the time sequence state transition tensor is:

[0013] The sensor network is laid out to collect real-time geothermal drilling construction state data and construct a structure state variable set;

[0014] Based on the absolute value of the first derivative of the structure state variable, the instantaneous disturbance intensity of each parameter is calculated, and the maximum disturbance value in the whole cycle is normalized to eliminate the dimension influence and form a dimensionless disturbance vector representing the relative disturbance velocity;

[0015] The state transition tensor is constructed by the outer product operation of the current state vector and the next time disturbance vector, and the cross-period coupling modeling of the construction behavior and dynamic response is completed.

[0016] Preferably, the generation process of the behavior influence coefficient is:

[0017] The tensor difference index representing the state transition trend fluctuation intensity is defined by calculating the Frobenius norm difference of adjacent state transition tensors;

[0018] The behavior influence coefficient is calculated by the ratio of the current tensor difference value to the sum of all historical tensor difference indexes up to the current time, which quantifies the global weight of the current construction behavior on the overall structure.

[0019] Preferably, the construction process of the risk sensitive band is:

[0020] The tensor difference index is defined as a perturbation trend function, and the change speed is calculated by first-order difference, and the acceleration is calculated by second-order difference;

[0021] A cooperative change matrix is constructed based on the perturbation vector sequence, and the absolute difference sum of the matrix elements of adjacent periods is calculated as a nonlinear shift strength index;

[0022] Fusion behavior influence coefficient, perturbation acceleration and cooperative change strength three high-order indexes, through the weight coefficient control each component weight, construct the comprehensive risk factor of nonlinear perturbation trend;

[0023] Based on the historical data of the risk factor, the mean and standard deviation are calculated in real time by using the sliding window, and the dynamic sensitive band is constructed, which is used as the adaptive threshold boundary for detecting nonlinear abnormal shift.

[0024] Preferably, the optimization process of optimizing the potential energy distribution by spatial smoothing filtering is as follows:

[0025] The drilling area is gridded in three-dimensional space to form a unit set G, which contains M spatial grid units;

[0026] For each grid unit, based on the state variable and the perturbation characteristic quantity, the local stress potential energy value is calculated by a nonlinear coupling function;

[0027] Combined with the risk factor and the dynamic sensitive band, the weight value of each parameter is dynamically adjusted by the risk response amplification coefficient and the indicator function. When the risk factor exceeds the sensitive band, the weight amplification mechanism is activated, and the potential energy function parameter is adaptively optimized in a risk-driven manner;

[0028] Based on the spatial conduction weight of adjacent unit set, the local potential energy of the grid unit is weighted and filtered to eliminate isolated noise and enhance spatial continuity, and the smoothed risk potential energy value is generated.

[0029] Preferably, the nonlinear coupling function introduces a normalization adjustment factor to unify the dimension and risk sensitivity, uses the power sensitivity index of the state value for nonlinear weighting, and dynamically amplifies the risk evolution sensitivity through the perturbation item exponential growth factor, and finally constructs a potential expression accurately representing the local risk trend.

[0030] Preferably, the implementation process of setting the observation window to track the potential energy change is as follows:

[0031] Extract the medium and high risk grid set, and calculate the spatial centroid coordinate vector based on the weighted risk potential energy value of each unit;

[0032] Based on the set of parameters to be controlled, the three-dimensional risk gradient field is calculated by using the central difference method, the response sensitivity matrix is constructed, and the control intervention vector is inverted, and the intervention amplitude is adjusted by the global control coefficient;

[0033] Projecting the target intervention vector to the controllable parameter constraint space to obtain feasible adjustment amount, and updating the control parameter for safe regulation accordingly;

[0034] Setting the observation window length, calculating the average change rate of risk potential in the delay window, and periodically recalculating the response sensitivity matrix and intervention vector to form a dynamic closed-loop regulation chain.

[0035] Preferably, the evaluation process of system self-healing ability is:

[0036] A1, collect the risk potential sequence of all grid cells from k0 to k T Risk potential sequence of all grid cells in continuous period, construct high-order tensor representing the overall risk evolution of the region;

[0037] A2, calculate the risk potential change rate of each grid cell as the self-healing index, quantify the local risk recovery trend, and based on the cell subset G H Calculate the average value of all self-healing indexes to generate the regional average recovery index representing the overall recovery trend of the system;

[0038] A3, construct a conical stable region based on the maximum risk expectation decline rate, and evaluate the stability of system self-healing trend by calculating the falling proportion of actual risk trajectory in the time sequence cone area;

[0039] A4, according to the average recovery index and the proportion of risk trajectory in the stable cone area, divide the system self-healing ability into four levels.

[0040] Preferably, step A3 includes:

[0041] Define the maximum expected deceleration rate of regional risk potential as:

[0042] ;

[0043] Construct a time sequence cone area: ;

[0044] Where, represents the maximum risk potential decline speed; represents the stable cone area of the i-th grid cell; represents the smoothed risk potential value of grid cell g i at the k-th time; represents the smoothed risk potential value of grid cell g i at the k+1-th time; H represents the evaluation region grid set, that is, the grid subset included in the risk recovery evaluation; represents the smoothed risk potential value of grid cell g i at the k0-th time;

[0045] The proportion of the trajectory in the cone:

[0046] Wherein, The proportion of the actual risk trajectory falling into the cone region.

[0047] The technical scheme of the present application: a construction monitoring system for geothermal drilling, which is used to execute the above-mentioned construction monitoring method for geothermal drilling, comprising:

[0048] The operation state modeling module acquires structural state data of the drilling site based on the sensor network arranged in the geothermal drilling, constructs a multi-dimensional operation state space, and generates an operation state transition tensor combining time series, behavior labels and response variables, identifies the influence law of operation behavior on construction dynamics, and outputs the tensor sequence and behavior influence coefficient.

[0049] The nonlinear disturbance feature extraction module extracts potential nonlinear deviation features through a high-order disturbance identification algorithm, calculates the disturbance direction, intensity and instability tendency of the risk trajectory, and outputs the deviation risk factor and dynamic risk sensitive band.

[0050] The deviation trend cone region modeling and classification module takes the nonlinear disturbance analysis result as input, automatically constructs a stable cone region model for different operation units and time intervals, calculates the local stress potential energy based on the nonlinear coupling function of state parameters and disturbance features, dynamically adjusts the weight coefficient according to the deviation risk factor, optimizes the potential energy distribution through spatial smoothing filtering, and divides the risk level according to the threshold.

[0051] The control intervention and adjustment module constructs an intervention control vector according to the risk trajectory and the deviation from the cone region, combines controllable process parameters to generate an optimal control strategy and execute parameter adjustment, and realizes active control of risk and dynamic constraint of nonlinear deviation.

[0052] The system self-healing evaluation module tracks and backtracks the system response after each round of intervention, calculates the regional average recovery index and the proportion in the cone, and evaluates the system self-healing ability according to the above, and combines the set level criterion to grade the self-healing, providing feedback support for the next round of intervention strategy.

[0053] Compared with the prior art, the above technical scheme of the present application has the following beneficial technical effects:

[0054] ​The application designs a construction monitoring system and method for geothermal drilling, accurately depicts the dynamic correlation between construction behavior and underground structure response by constructing state transition tensor, realizes multi-parameter disturbance coupling analysis, and significantly improves the abnormal identification sensitivity under complex working conditions; the offset risk factor is generated by fusing high-order derivative and collaborative change matrix, combined with the self-adaptive sensitive band mechanism, the early accurate capture of nonlinear risk is realized; based on the local stress potential energy mapping and dynamic weight adjustment technology of three-dimensional grid, the abstract risk is quantified as a spatial distribution model, solving the problem of fuzzy risk area positioning of geothermal drilling; the intelligent intervention vector is generated by risk gradient field inversion and controllable field projection, forming a monitoring-warning-regulation closed loop control, effectively inhibiting the evolution of nonlinear risk; the self-healing ability cone zone evaluation model is introduced, the system recovery trend and stability are quantified, providing a scientific basis for dynamically optimizing the control strategy, and finally realizing the coordinated improvement of the safety and efficiency of geothermal drilling construction. BRIEF DESCRIPTION OF DRAWINGS

[0055] Figure 1 A method flowchart of a construction monitoring method for geothermal drilling is proposed for the application;

[0056] Figure 2 A system architecture diagram of a construction monitoring system for geothermal drilling is proposed for the application. DETAILED DESCRIPTION

[0057] Embodiment one, as shown in the figure, the application proposes a construction monitoring method for geothermal drilling, including the following specific implementation steps: Figure 1

[0058] S1, a mathematical model for depicting the dynamic correlation between drilling behavior and underground structure response is constructed, high-dimensional state transition tensor construction + time sequence response interpolation rule is adopted, the underground structure disturbance mapping path driven by construction behavior is extracted from the measured data, and complete physical semantic support is provided for subsequent risk analysis and control strategy generation, and the specific implementation process is as follows:

[0059] S11, a sensor network is laid out on the construction site, and geothermal drilling construction state data (including but not limited to drill string axial stress, drilling pressure, rotating speed, mud density, annular pressure difference, wellhead temperature, well depth position) is collected in real time, forming a structure state variable set S(t);

[0060] S12, disturbance derivative method is adopted, and the structure disturbance intensity D i (t) is defined:

[0061] ;

[0062] Among them, represents the disturbance intensity, that is, the speed of change of each variable with time; ​S represents the state variable i (t), i. e., the rate of change; n represents the total number of collected structural state variables;

[0063] All perturbation quantities are standardized to obtain dimensionless perturbation vectors:

[0064] ;

[0065] wherein, represents the dimensionless perturbation vector, used to describe the relative perturbation speed of the system on each variable; represents the maximum value of the i-th perturbation quantity in the whole construction period (used for normalization);

[0066] S13, based on the sequence of perturbation characteristic quantities, defines the state transition tensor k in the time interval [t k+1 ]: ;

[0067] ;

[0068] wherein, represents the state transition tensor, depicting the influence pattern of the current state on the perturbation change at the next time; represents the perturbation vector at the next time, representing the future behavior change trend; represents the state variable vector at the current time; represents the tensor outer product, producing an n x n matrix; t k and t k+1 represent the current and next time sampling points, respectively;

[0069] Accordingly: the tensor sequence is formed by splicing multiple time points, used to observe the behavior evolution path;

[0070] S14, by comparing the change trend between consecutive tensors, the tensor difference index is defined:

[0071] ;

[0072] wherein, represents the state transition tensor at time k; represents the Frobenius norm, i. e., the overall difference between matrices; represents the fluctuation intensity of the state transition trend in the current time period, used to detect abnormal perturbations or sudden behaviors;

[0073] S15, the tensor difference index is mapped to the behavior influence coefficient:

[0074] ;

[0075] wherein, represents the current behavior impact ratio, that is, the importance of the behavior to the structure in the entire construction period; m represents the length of the tensor time series during the construction, that is, the number of time segments divided from the construction starting point to the current time t k .

[0076] S2, by constructing a disturbance trend function and its high-order derivative, fusing behavior collaborative change characteristics, extracting a nonlinear deviation risk factor, and performing early warning and dynamic monitoring of geothermal drilling construction behavior, the specific implementation process is as follows:

[0077] S21, define the disturbance trend function : , and perform high-order difference analysis on the disturbance trend function , and define:

[0078] First-order derivative (disturbance change rate): ;

[0079] Second-order derivative (trend / acceleration of disturbance change):

[0080] ;

[0081] wherein, represents the disturbance trend function value at time k, which measures the overall system response change energy level in the current time period; represents the first-order difference of the disturbance trend, that is, the disturbance speed, which is used to represent the speed of disturbance trend growth or decline, and to judge whether the disturbance trend is stable or intensified; represents the second-order difference of the disturbance trend, that is, the disturbance acceleration; and represent the disturbance trend function values at times k-1 and k-2, respectively;

[0082] S22, use the disturbance vector sequence to construct a behavior variable collaborative change matrix:

[0083] ;

[0084] In order to judge whether the collaborative motion has a nonlinear deviation, a time sliding window collaborative change function is introduced:

[0085] ;

[0086] wherein, represents the disturbance collaborative matrix; represents the disturbance vector (behavior disturbance feature vector) at time t; represents the collaborative change intensity index of the kth time period; represents the co-movement strength of behavior variable i and variable j in the kth time period; represents the co-movement strength of behavior variable i and variable j in the k+1th time period;

[0087] S23, fuse the above high-order indicators to define the shift risk factor :

[0088] ;

[0089] wherein, represents the shift risk factor (fusion indicator of nonlinear disturbance trend); , and represent the weighting coefficients, which control the weight of each component in the risk indicator; represents the behavior influence coefficient in the kth time period;

[0090] S24, adaptively construct the sensitive band according to the operation history: ;

[0091] wherein, represents the dynamic sensitive band of shift risk; and respectively represent the mean and standard deviation of the shift risk factor in the sliding window before time t.

[0092] S3, by constructing a three-dimensional construction grid, combining a nonlinear mapping function and dynamically adjusting the weight, reconstructing the local stress potential energy, and then generating a smooth risk distribution mapping, realizing the accurate identification and dynamic management of the potential risk area of geothermal drilling, the specific implementation process is as follows:

[0093] S31, divide the drilling area into a three-dimensional grid unit set according to the spatial dimension:

[0094] ;

[0095] wherein, G represents the grid division set of the three-dimensional construction area, containing a total of M spatial units; g i represents the i th grid unit of the division;

[0096] S32, for each unit g i , based on the state variable S(t) and the disturbance characteristic quantity in step S1, define the local stress potential energy function: ;

[0097] wherein, represents the local stress potential energy value in the grid unit g i at time point k; represents the global weight of the jth parameter in the stress potential energy in the original state; represents the value of the jth physical state parameter in the grid cell g i at the kth moment; represents the disturbance value of the jth disturbance characteristic parameter at the kth moment and in the cell g i ; represents the nonlinear coupling function of the jth state parameter and its disturbance characteristic, reflecting the contribution of a single-dimensional parameter to the local potential energy;

[0098] It should be noted that the nonlinear coupling function is used to describe the coupling relationship between the state value of each monitoring parameter and its corresponding disturbance intensity, which comprehensively considers the current physical state of the parameter, the disturbance trend, and the response sensitivity of the parameter in abnormal evolution. Through the power enhancement of the state parameter and the exponential weighting fusion of the disturbance factor, a nonlinear response gain mechanism is formed, which amplifies the influence of variables highly sensitive to risk evolution, suppresses the interference effect of slowly changing or stable variables, and builds a stress potential energy expression that can accurately reflect the local risk evolution trend: ;

[0099] wherein, represents the normalized adjustment factor of the jth state parameter, which is trained according to historical construction data and is used for dimension unification and risk sensitivity alignment of different dimensional state values; represents the power sensitivity index of the state value x, which is set according to the nonlinear trend of system response, and controls the nonlinear weighting of the contribution of state value change to potential energy, indicating that the state value is more sensitive to large state values; then the low value state is more "dangerous"; represents the risk sensitivity coefficient (exponential growth factor) of the disturbance term y, which is fitted through historical data and is used to adjust the amplification degree of the disturbance to the risk, and determines whether the risk disturbance triggers "risk acceleration". If the disturbance value continues to increase, will increase exponentially, rapidly increasing the potential energy;

[0100] S33, considering the offset risk factor and the dynamic sensitive band of the offset risk output by step S2 , dynamically adjusting the weight in the local stress potential energy function: ;

[0101] wherein, represents the weight value of the jth parameter after risk adjustment at time k; represents the risk response amplification coefficient, which is used to adjust the response strength of the risk factor to the weight; represents an indicator function, which is Beyond the current time point sensitive zone 1, otherwise 0;

[0102] S34, introduce the local potential energy of each unit into smoothing filter according to spatial continuity:

[0103] ;

[0104] wherein, represents the smoothed risk potential energy value of the grid unit g i at the kth moment; N(g i ) represents the set of adjacent units of the grid unit g i ; represents the spatial transmission influence weight of the unit g j on the unit g i ;

[0105] S35, define a set of risk classification threshold values according to the statistical characteristics of the global potential energy: ;

[0106] And give each unit a risk level: ;

[0107] wherein, represents the upper limit of low risk; represents the lower limit of high risk; represents the risk level of the grid unit g i at the kth moment.

[0108] S4, by constructing the risk potential energy centroid and parameter response matrix, generate the inverse intervention vector and project it to the controllable domain, realize the intelligent dynamic adjustment of the nonlinear risk of geothermal drilling, the specific implementation process is as follows:

[0109] S41, extract the grid set whose risk level is high at the current moment:

[0110] ;

[0111] Calculate the weighted risk potential energy centroid position vector:

[0112] ;

[0113] wherein, represents the set of grid units whose risk level is “medium” and above; represents the weighted centroid coordinate vector of the risk potential energy at the kth moment; represents the spatial coordinates of the grid unit g i ;

[0114] S42, define the set of operation parameters to be controlled U: ;

[0115] The gradient at discrete points is calculated using the central difference method, i.e., the risk gradient field is calculated. :

[0116] ;

[0117] Where up represents the p-th parameter to be controlled; P represents the number of parameters to be controlled; Represents the grid cell g i The risk potential gradient vector at time k; These represent relative to the current grid g. i The risk potential energy value of the right neighboring cell in the x direction; Represents relative to the current grid g i The risk potential energy value of the left neighboring cell in the x direction; and These represent relative to the current grid g. i The risk potential energy values ​​of the upper and lower adjacent units in the y direction; and These represent relative to the current grid g. i The risk potential energy value of the adjacent grid cells in the z direction (i.e., well depth); , and These represent the side length of the mesh cell in the x-direction, the side length of the mesh in the y-direction, and the height of the mesh in the direction perpendicular to the well depth (z-direction), respectively.

[0118] Based on the risk gradient field in step S3 ( Based on the current physical state of the drilling system, construct the response sensitivity matrix of the control parameters to each risk direction:

[0119] ;

[0120] By projecting the response sensitivity matrix onto the gradient direction of the risk potential energy distribution, the intervention vector that minimizes the potential energy can be derived: ;

[0121] in, This represents the response sensitivity matrix, characterizing the degree of influence of each parameter on the potential energy; This represents the control parameter adjustment vector (initial intervention amount). This represents the overall control coefficient, which controls the magnitude of intervention.

[0122] S43, Target intervention vector Project to the allowed control subspace:

[0123] ;

[0124] Final control parameter update: ;

[0125] where, is the projected control intervention vector after projection to the feasible operation space; is the control variable allowable space, i.e., the constraint set of all operation parameters; is the projection operator that restricts the parameter adjustment vector to the range ;

[0126] S44, set the observation window , delay track the trend of key structure parameters:

[0127] ;

[0128] where, is the control delay observation window length; is the grid g i The average risk potential change rate in the delay window; is the risk potential change value of the adjacent time;

[0129] Accordingly: periodic re-computation and , form a dynamic closed-loop intervention chain.

[0130] S5, build a multi-period risk potential evolution model to quantify the matching degree of system recovery trend and stability cone, realize the hierarchical evaluation and control strategy feedback of geothermal well structure self-healing ability, its specific implementation process is as follows:

[0131] S51, collect the regional risk potential tensor in the continuous period from the k0 time to the current time k T : ;

[0132] where, is the multi-period risk potential tensor, i.e., the high-order tensor composed of all risk potential sequence data from time k0 to k T ; N g is the total number of grid cells, i.e., the total number of spatial grids divided in the geothermal well area;

[0133] S52, for each grid cell g i , calculate its risk potential self-healing index:

[0134] ;

[0135] Based on the RI values of all units, calculate the regional average recovery index:

[0136] ;

[0137] wherein, represents the self-healing index, i.e. the risk recovery trend of the ith grid cell, the larger the value, the more obvious the risk reduction; represents the average recovery index, i.e. the average value of the self-healing indexes of the grid in the whole region, used to evaluate the overall self-healing trend of the system; H represents the evaluation region grid set, i.e. the subset of grids included in the risk recovery evaluation, which in the embodiment is set as the region with significant risk activities;

[0138] S53, to further evaluate whether the system has a stable self-healing trend, a stable cone tolerance zone is introduced, specifically:

[0139] The maximum expected deceleration rate of the regional risk potential is defined as:

[0140] ;

[0141] The time sequence cone zone is constructed: ;

[0142] wherein, represents the maximum risk potential decline rate, i.e. the maximum rate of risk reduction within the evaluation period, used to construct the lower limit of the stable trend; represents the stable cone zone of the ith grid cell, i.e. the conical region used to determine the "ideal self-healing trend", the lower boundary of which is defined as the maximum risk expected decline trend;

[0143] The proportion of the trajectory within the cone is calculated: ;

[0144] wherein, represents the proportion of the actual risk trajectory falling within the cone zone, reflecting the real self-healing trend and stability of the system;

[0145] S54, according to the recovery index and the trajectory proportion within the cone , the self-healing ability of the system is defined, as shown in Table 1:

[0146] Table 1 Self-healing ability grading table

[0147]

[0148] wherein, and represent the self-healing grading thresholds;

[0149] S55, according to the evaluated self-healing grade:

[0150] If it is A / B grade: the system has good recovery, continue to execute the current control strategy;

[0151] If C: suggest adjusting the control weight coefficient , or introducing auxiliary control parameters (such as mud property adjustment);

[0152] If D: determine the structural non-stable evolution trend, suggest suspending the operation, changing the process, implementing structural pre-support, etc. Artificial intervention.

[0153] In Example Two, as Figure 2 shown, the present application proposes a construction monitoring system for geothermal drilling, which is used to execute the construction monitoring method for geothermal drilling proposed in Example One, and includes an operation state modeling module, a nonlinear disturbance feature extraction module, an offset trend cone modeling and classification module, a control intervention and adjustment module, and a system self-healing evaluation module.

[0154] The operation state modeling module acquires structural state data of the drilling site based on the sensor network arranged in the geothermal drilling, constructs a multi-dimensional operation state space, and generates an operation state transition tensor in combination with time series, behavior labels and response variables to identify the influence law of operation behavior on construction dynamics; the tensor sequence and behavior influence coefficient are outputted to provide a structural expression basis for subsequent disturbance extraction;

[0155] The nonlinear disturbance feature extraction module receives the tensor structure and behavior influence factor, extracts potential nonlinear offset features through a high-order disturbance identification algorithm; calculates the disturbance direction, intensity and instability tendency of the risk trajectory, and outputs the offset risk factor and dynamic risk sensitive band to reveal the implicit instability trend in the drilling process;

[0156] The offset trend cone modeling and classification module automatically constructs a stable cone model for different operation units and time intervals based on the nonlinear disturbance analysis results; compares each actual risk trajectory with the cone relationship, calculates quantitative indexes such as trajectory cone proportion and trend matching degree, realizes the structural classification of potential offset trend, and provides boundary criteria for control intervention;

[0157] The control intervention and adjustment module constructs an intervention and control vector based on the risk trajectory and cone deviation, in combination with controllable process parameters (including but not limited to pump pressure, drilling speed, flow rate); generates an optimal control strategy based on the disturbance-response mapping function and executes parameter adjustment to realize active control of risk and dynamic constraint of nonlinear offset; forms a disturbance-identification-adjustment closed loop system;

[0158] The system self-healing evaluation module tracks and analyzes the system response after each round of intervention, calculates the average risk recovery rate and cone proportion, and evaluates the system self-healing ability based on the set level criteria (A~D); provides feedback support for the next round of intervention strategy, realizes intelligent evolutionary adaptive optimization.

[0159] The embodiments of the present application are described in detail above with reference to the accompanying drawings, but the present application is not limited to the embodiments. Various changes that one skilled in the art would make within the scope of the knowledge of the art without departing from the spirit of the present application are possible.

Claims

1. A method for construction monitoring of geothermal drilling, characterized in that, The specific implementation steps include the following: S1, based on the real-time collected drilling parameters, the disturbance intensity is calculated and normalized, the state vector and the disturbance vector are used to construct the time sequence state transition tensor, the abnormal fluctuation is detected by using the Frobenius norm difference between continuous tensors, and the difference value is mapped to the behavior influence coefficient; S2, based on the high-order difference analysis of the disturbance trend function, the disturbance velocity and acceleration are calculated, the behavior variable cooperative change matrix and the time sliding window function are used to capture the multi-parameter cooperative deviation characteristics, the behavior influence coefficient, the disturbance acceleration and the cooperative change intensity are fused to generate the deviation risk factor, and the risk sensitive band is dynamically constructed according to the historical data sliding window statistics; S3, the drilling area is three-dimensionally gridded, the local stress potential energy is calculated based on the nonlinear coupling function of the state parameters and the disturbance characteristics, the weight coefficient is dynamically adjusted according to the deviation risk factor, the potential energy distribution is optimized through spatial smoothing filtering, and finally the risk level is divided according to the threshold value; S4, the medium and high risk grids are extracted and the weighted potential energy centroid is calculated, the intervention vector is inverted based on the risk gradient field and the response sensitivity matrix, the control amount is updated after being projected to the controllable parameter domain, and the observation window is set to delay track the potential energy change; S5, according to the risk potential energy tensor, the self-healing index of each grid and the regional average recovery index are calculated, the stable cone area is constructed by defining the maximum risk decline rate and calculating the trajectory falling proportion, the self-healing ability of the system is evaluated by combining the recovery index and the cone proportion, and the regulation strategy is dynamically adjusted according to the level.

2. The method for monitoring the construction of a geothermal well according to claim 1, wherein, The construction process of the time sequence state transition tensor is as follows: The sensor network is laid out to collect real-time geothermal drilling construction state data, and a set of structural state variables is constructed; Based on the absolute value of the first derivative of the structural state variable, the instantaneous disturbance intensity of each parameter is calculated, and the normalized processing is performed through the maximum disturbance value in the whole cycle, so that the dimensionless disturbance vector representing the relative disturbance velocity is formed after eliminating the dimension influence; The state transition tensor is constructed by the outer product operation of the current state vector and the next time disturbance vector, and the cross-period coupling modeling of the construction behavior and the dynamic response is completed.

3. The method for monitoring the construction of a geothermal well according to claim 2, wherein, The generation process of the behavior influence coefficient is as follows: The Frobenius norm difference of the adjacent state transition tensors is calculated, and the tensor difference index representing the state transition trend fluctuation intensity is defined; The behavior influence coefficient is calculated by the ratio of the current tensor difference value to the sum of all historical tensor difference indexes up to the current time, which quantifies the global weight of the influence of the current construction behavior on the overall structure.

4. The method for construction monitoring of geothermal drilling according to claim 3, wherein, The construction process of the risk sensitive band is as follows: The tensor difference index is defined as the disturbance trend function, the change velocity is calculated by the first-order difference, and the acceleration is calculated by the second-order difference; The cooperative change matrix is constructed based on the disturbance vector sequence, and the sum of the absolute differences of the adjacent period matrix elements is calculated as the nonlinear deviation intensity index; The behavior influence coefficient, the disturbance acceleration and the cooperative change intensity are fused to construct the comprehensive deviation risk factor representing the nonlinear disturbance trend, and the weight of each component is controlled by the weighted coefficient. Based on the historical data of the offset risk factor, the mean and standard deviation are calculated in real time using a sliding window to construct a dynamic sensitive band, which is used as the adaptive threshold boundary for detecting nonlinear abnormal offset.

5. The method for construction monitoring of geothermal drilling according to claim 4, wherein, The optimization process of potential energy distribution by spatial smoothing filtering is as follows: The drilling area is three-dimensionally gridded to form a unit set G, which contains M spatial grid units. For each grid unit, the local stress potential value is calculated based on the state variable and disturbance characteristic quantity through a nonlinear coupling function. Combined with the offset risk factor and the dynamic sensitive band, the parameter weight values are dynamically adjusted through the risk response amplification coefficient and the indicator function. When the risk factor exceeds the sensitive band, the weight amplification mechanism is activated, and the risk-driven potential function parameter adaptive optimization is performed. Based on the spatial conduction weight of the adjacent unit set, the local potential energy of the grid unit is weighted and filtered to eliminate isolated noise and enhance spatial continuity, generating a smoothed risk potential value.

6. The method for construction monitoring of geothermal drilling according to claim 5, wherein, The nonlinear coupling function introduces a normalization adjustment factor to unify the dimension and risk sensitivity. The power sensitivity index of the state value is used for nonlinear weighting, and the disturbance term exponential growth factor is used to dynamically amplify the risk evolution sensitivity, finally constructing a potential expression that accurately represents the local risk trend.

7. The method for construction monitoring of geothermal drilling according to claim 6, wherein, The implementation process of setting the observation window to track the potential change is as follows: Extract the medium and high risk grid set, and calculate the spatial centroid coordinate vector based on the risk potential value of each unit. Based on the set of parameters to be controlled, the three-dimensional risk gradient field is calculated using the central difference method, and the response sensitivity matrix is constructed to invert the control intervention vector. The intervention amplitude is adjusted by the global control coefficient. The target intervention vector is projected into the controllable parameter constraint space to obtain the feasible adjustment amount, and the control parameters are updated accordingly for safety control. Set the observation window length, calculate the average change rate of risk potential in the delay window, and periodically recalculate the response sensitivity matrix and intervention vector to form a dynamic closed-loop control chain.

8. The method for construction monitoring of geothermal drilling according to claim 7, wherein, The evaluation process of system self-healing ability is as follows: A1, collect from k0 to k T The risk potential sequence of all grid cells in the continuous period, constructs a high-order tensor representing the overall risk evolution of the region. A2, for each grid cell, compute its risk potential change rate as a self-healing index, quantifying the local risk recovery trend, and based on the subset of cells G of the significant area of risk activity H compute the average of all self-healing indices, generating a regional average recovery index characterizing the overall recovery trend of the system; A3, based on the maximum risk expectation decline rate, a conical stable area is constructed, and the stability of the system self-healing trend is evaluated by calculating the proportion of the actual risk trajectory falling within the time sequence cone; A4, according to the average recovery index and the proportion of the risk trajectory in the stable cone, the system self-healing ability is divided into four levels.

9. The method for construction monitoring of geothermal drilling according to claim 8, wherein, Step A3 includes: Define the maximum expected deceleration rate of the regional risk potential energy as: ； Constructing timing cones: ; wherein, represents the maximum risk potential descent speed; represents the stability cone area of the i-th grid cell; represents the smoothed risk potential value of the grid cell g i at the k-th time instant; represents the smoothed risk potential value of the grid cell g i at the k+1-th time instant; G H represents the grid set of the evaluation area, i.e. the subset of grid cells that are included in the risk resilience evaluation; represents the smoothed risk potential value of the grid cell g i at the k0-th time instant; Calculate the proportion of the trajectory in the cone: ； wherein, represents the proportion of actual risk trajectories falling into the cone region; N g represents the total number of grid cells, that is, the total number of spatial grids divided in the geothermal drilling area.

10. A construction monitoring system for geothermal drilling, which is used to perform the construction monitoring method for geothermal drilling according to any one of claims 1 to 9, characterized in that, Including: The operation state modeling module acquires structural state data from the sensor network deployed in the geothermal well, constructs a multi-dimensional operation state space, and generates operation state transition tensors combined with time series, behavior labels, and response variables. The influence of operation behavior on construction dynamics is identified, and the tensor sequence and behavior influence coefficient are output. The nonlinear disturbance feature extraction module extracts potential nonlinear offset features through high-order disturbance identification algorithms, calculates the disturbance direction, intensity, and instability tendency of the risk trajectory, and outputs the offset risk factor and dynamic risk sensitive band. The offset trend cone region modeling and classification module takes the nonlinear disturbance analysis results as input, automatically constructs a stable cone region model for different operation units and time intervals, calculates the local stress potential energy based on the nonlinear coupling function of state parameters and disturbance characteristics, dynamically adjusts the weight coefficient according to the offset risk factor, optimizes the potential energy distribution through spatial smoothing filtering, and divides the risk level according to the threshold; The control intervention and adjustment module constructs an intervention control vector according to the risk trajectory and cone region deviation, generates an optimal control strategy and executes parameter adjustment, realizes active control of risk and dynamic constraint of nonlinear offset, and realizes active control of risk and dynamic constraint of nonlinear offset; The system self-healing evaluation module tracks and analyzes the system response after each round of intervention, calculates the regional average recovery index and the proportion in the cone, evaluates the system self-healing ability, and classifies the self-healing according to the set level criterion, providing feedback support for the next round of intervention strategy.

Images