Portable mud logging instrument based on temperature calibration

Abstract

This invention relates to the field of geothermal exploration technology, specifically disclosing a portable logging tool based on temperature calibration. The tool includes a data acquisition module and a receiver wired to the data acquisition module. The data acquisition module includes a temperature probe, a gas detection device, and a lithology identification device. The receiver incorporates a preprocessing module, a formation real-time temperature calibration module, a machine learning real-time temperature prediction module, a drilling process parameter optimization decision module, and a closed-loop feedback and model adaptive update module. Through the wired connection between the temperature probe and the receiver, real-time temperature testing can be conducted on-site, and data can be recorded promptly, reducing temperature discrepancies from manual measurements. Compared to previous logging tools, this invention can be used for logging geothermal temperature measurement wells, hydrological wells, and ground-source heat pump exploration wells, enabling the real-time acquisition of more complete geological data. This provides crucial information for subsequent geothermal reservoir / aquifer evaluation, well design, system optimization, and safe construction.

Description

Technical Field

[0001] This application relates to the field of geothermal exploration technology, and specifically discloses a portable logging instrument based on temperature calibration. Background Technology

[0002] In the geothermal resource exploration phase, obtaining accurate and continuous formation temperature data is fundamental to accurately assessing resource reserves and development value. Currently, on-site temperature measurement operations for geothermal wells mainly involve lowering temperature measuring tools directly into the well after drilling is completed and the wellbore is in an open-hole state.

[0003] While this type of temperature measurement method can obtain the well temperature at a certain depth, after drilling is completed, the wellbore is directly connected to the surface atmosphere, and there is a continuous heat exchange between the fluid inside the well and the outside world. As a result, the measured downhole temperature is actually the result of the mixture of formation temperature and low surface temperature disturbance, which cannot accurately represent the true temperature of the original underground strata, and the accuracy of temperature measurement is difficult to guarantee.

[0004] Moreover, each temperature measurement requires professional technicians to carry temperature measuring instruments to the drilling site, which is often located in remote areas far from residential areas. In addition, before the measurement, the sealed wellhead must be reopened by the engineers. The whole process consumes a lot of manpower, material resources and time, resulting in high operating costs. Since each measurement requires the temperature measuring tool to be lowered and lowered, only discrete point measurements can be performed in time, making continuous real-time monitoring impossible. In particular, when the closed wellhead is frequently opened to obtain temperature changes over a longer period, the thermal balance downhole is repeatedly disrupted, which introduces greater measurement errors and further degrades the data quality.

[0005] It is evident that current geothermal exploration drilling temperature measurement suffers from problems such as inaccurate temperature data, large on-site workload, high cost, and discontinuous temperature data. There is an urgent need for a logging instrument that can acquire the true formation temperature of geothermal wells continuously and automatically over a long period of time, while significantly reducing manpower consumption and human interference, to meet the current needs of geothermal exploration. Summary of the Invention

[0006] The purpose of this invention is to provide a logging instrument that can obtain the true formation temperature of geothermal wells continuously and automatically over a long period of time, while significantly reducing manpower consumption and human interference, so as to meet the current needs of geothermal exploration and solve the related technical problems mentioned in the background art.

[0007] To achieve the above objectives, the basic solution of the present invention provides a portable logging instrument based on temperature calibration, including a data acquisition module and a receiver wiredly connected to the data acquisition module. The data acquisition module includes a temperature probe, a gas detection device, and a lithology identification device. The receiver has the following built-in components: The preprocessing module preprocesses the real-time temperature data and outputs a standardized temperature data stream. The formation true temperature calibration module receives a standardized temperature data stream, uses the wellbore-formation transient heat transfer model and real-time temperature measurement data to assimilate the data, and dynamically calibrates the formation true temperature and its uncertainty information. The machine learning real-time temperature prediction module uses the calibrated actual formation temperature as a benchmark and outputs the predicted values ​​of bottom hole circulation temperature and wellbore temperature distribution under different drilling conditions through a hybrid prediction engine. The drilling process parameter optimization decision module solves for the optimal drilling parameters within the feasible region jointly defined by temperature constraints and well control safety constraints, based on predicted values ​​and uncertainty information. The closed-loop feedback and model adaptive update module monitors the temperature response deviation after drilling parameter adjustment, and triggers the formation real temperature calibration module and the machine learning real-time temperature prediction module to perform adaptive updates when the deviation exceeds the limit.

[0008] Furthermore, the formation true temperature calibration module includes: The wellbore-formation transient heat transfer model unit establishes a coupled forward model consisting of the drilling fluid energy conservation equation and the formation heat conduction equation, describing the dynamic response of the temperature field of the wellbore and the formation under given engineering parameters and formation thermal properties. The layered temperature correction inversion unit constructs a cost function with a smooth regularization term based on the residual between the measured temperature and the model simulated temperature, and obtains the original formation temperature profile and thermal conductivity parameters of each layer through inversion. The Kalman filter real-time assimilation calibration unit defines a state vector containing formation temperature node values ​​and thermal conductivity parameters. Through iterative prediction and update steps, the real-time temperature observations are dynamically fused with the coupled forward model to output the calibrated true formation temperature distribution and uncertainty confidence interval.

[0009] Furthermore, in the ensemble Kalman filter real-time assimilation calibration unit: An initial set containing multiple members is generated based on the output of the hierarchical temperature correction inversion unit. The initial value of each member is formed by superimposing statistical perturbations on the inversion result. When new construction parameters are obtained during drilling, the coupled forward model is used to integrate each set member from the analysis state of the previous moment to the current moment, and the model error term is added to obtain the prediction set. Upon receiving a new real-time temperature observation vector, the Kalman gain is calculated based on the statistical background error covariance matrix of the prediction set, and the Kalman gain and observation perturbation are used to update each set member to obtain the analysis set. The mean of the analysis set is used as the calibrated formation temperature distribution and thermophysical parameters, and the statistical dispersion of the analysis set is used to output the calibrated temperature confidence interval for each depth point.

[0010] Furthermore, the machine learning real-time temperature prediction module includes: The physical information neural network sub-model takes depth and time as network inputs, embeds the wellbore heat transfer control equation as a physical residual term into the loss function, and outputs the predicted wellbore temperature distribution that satisfies the physical constraints. The temporal neural network sub-model learns the temporal mapping relationship between construction parameters, calibration geothermal gradient and historical bottom hole circulation temperature, and outputs the predicted value of future bottom hole circulation temperature. The working condition discrimination and switching unit automatically switches between the physical information neural network sub-model, the temporal neural network sub-model, or the lightweight gradient boosting tree model for prediction based on the rate of change of drilling parameters. The online drilling heat map query unit, based on an offline pre-calculated parameter-temperature mapping table, can query the cooling potential of different parameter combinations in real time.

[0011] Furthermore, the loss function of the physical information neural network sub-model is composed of a weighted sum of a data matching error term and a physical residual term. The physical residual term is the mean square value of the residual of the wellbore drilling fluid energy conservation equation, and the calibrated formation temperature output by the formation real temperature calibration module is used as the boundary condition.

[0012] Furthermore, the drilling process parameter optimization decision module includes: The multi-objective optimization solution unit constructs a multi-objective optimization model with cooling deviation, pump power saving and mechanical drilling speed improvement as optimization objectives and temperature safety constraints and wellbore pressure balance constraints as constraints, and solves for the optimal drilling parameters; The high-temperature anomaly identification and prevention unit identifies the high-temperature anomaly zone ahead based on the calibrated formation temperature gradient profile, estimates the drilling time to the anomaly section, and outputs preventive cooling strategies in advance.

[0013] Furthermore, the optimization model constructed by the multi-objective optimization solution unit uses mud discharge, inlet temperature, rotational speed and drilling pressure as decision variables; The temperature safety constraints include a predicted bottomhole circulation temperature plus a safety margin determined by the calibration uncertainty confidence interval that is less than or equal to the upper limit of the downhole tool temperature resistance. The wellbore pressure balance constraint includes the bottom hole circulating pressure being between the formation pore pressure and the fracture pressure, and the bottom hole circulating pressure calculation incorporates the correction for drilling fluid density due to temperature.

[0014] Furthermore, the closed-loop feedback and model adaptive update module includes: The response-prediction deviation monitoring unit continuously compares the deviation between the actual response value of the bottom hole circulation temperature after drilling parameter adjustment and the predicted value output by the machine learning real-time temperature prediction module. The model mismatch determination and triggering unit determines that the model is mismatched when the deviation continuously exceeds a preset threshold within a continuous period of time. It then automatically injects recent actual drilling temperature data into the Kalman filter assimilation loop of the formation real temperature calibration module for recalibration, and simultaneously triggers the machine learning real-time temperature prediction module to perform incremental learning and updates.

[0015] Based on the same inventive concept, the present invention also provides a data processing method for a temperature-calibrated portable logging instrument, which processes geothermal exploration data through the steps of the above-mentioned temperature-calibrated portable logging instrument.

[0016] Furthermore, the following steps are included: Step S1: Collect temperature data in real time from multiple temperature sensing sources, and perform time alignment, spatial alignment and quality control preprocessing to output a standardized temperature data stream; Step S2: Receive the standardized temperature data stream, use the wellbore-formation transient heat transfer model and real-time temperature measurement data to assimilate the data, and dynamically calibrate the true formation temperature and its uncertainty information; Step S3: Based on the calibrated actual formation temperature, output the predicted values ​​of bottom hole circulation temperature and wellbore temperature distribution under different drilling conditions through the hybrid prediction engine. Step S4: Based on the predicted values ​​and the uncertainty information, solve for the optimal drilling parameters within the feasible region jointly defined by temperature constraints and well control safety constraints; Step S5: Monitor the temperature response deviation after drilling parameter adjustment, and trigger the formation temperature recalibration in step S2 and the adaptive update of the prediction model in step S3 when the deviation exceeds the limit, forming a closed loop.

[0017] The principle and effect of this solution are as follows: This invention employs a surface receiver to automatically receive and record temperature, time, and well depth data transmitted from temperature sensors at various depths. It automatically collects, stores, and analyzes temperature information according to a pre-defined detection time difference, eliminating the need for on-site personnel throughout the process. This significantly reduces labor costs and fundamentally eliminates the impact of human error on test results. Furthermore, by monitoring real-time data such as wellhead and wellbore temperature, gas content, and lithology, it provides crucial information for geothermal reservoir / aquifer evaluation, well design, system optimization, and safe construction.

[0018] In this invention, once the instrument is deployed and cemented, it enters a long-term autonomous working state, eliminating the need for technicians to travel to remote sites to open wells and manually lower the instrument for temperature measurement each time. This significantly saves on travel, labor, and repeated wellhead opening and closing costs, greatly reducing labor intensity and overall costs. Furthermore, the temperature probe and receiver are connected via a wired connection and inserted into the cementing cement. After the cementing cement solidifies, it directly connects to the formation, and after a certain period of time, the temperature in the area where the temperature probe is located can reach equilibrium with the bottom layer temperature, resulting in more accurate temperature measurements.

[0019] The receiver of this invention breaks away from the open-loop architecture of existing technologies, which separates calibration and decision-making, by constructing a complete closed-loop architecture encompassing real-time temperature sensing, formation temperature calibration, model prediction, drilling parameter optimization, control effect feedback, and model recalibration. It directly and automatically drives the optimization of operational parameters with the calibration results of the formation temperature, and continuously corrects the model through effect feedback. This avoids the uncertainty and lag in human decision-making, significantly improving the intelligent management level of high-temperature geothermal drilling. Attached Figure Description

[0020] To more clearly illustrate the technical solutions in the embodiments of this application, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0021] Figure 1 A schematic diagram of the architecture of a portable logging instrument based on temperature calibration proposed in an embodiment of this application is shown; Figure 2 A system diagram of the receiver in a temperature-calibrated portable logging instrument according to an embodiment of this application is shown; Figure 3 A flowchart of a geothermal exploration data processing method based on temperature calibration proposed in an embodiment of this application is shown. Detailed Implementation

[0022] To further illustrate the technical means and effects of the present invention in achieving its intended purpose, the following detailed description of the specific implementation methods, structures, features, and effects of the present invention, in conjunction with the accompanying drawings and preferred embodiments, is provided below.

[0023] A portable logging tool based on temperature calibration includes a data acquisition module and a receiver wired to the data acquisition module. The data acquisition module includes a temperature probe, a gas detection device, and a lithology identification device. For example, it uses a temperature probe with built-in temperature measurement, positioning, and signal transmission functions, located at the well mud inlet / outlet and well wall. The number of temperature probes can be adjusted according to the actual logging depth and logging length. It can also be configured with distributed fiber optic temperature measurement cables, mud inlet / outlet thermometers, and surface reference thermometers, among other multi-source sensing devices. The gas detection device uses a detachable portable hydrogen sulfide detector, which is deployed at the wellhead, working area, and core storage area. The detection time interval can be designed by the user. By monitoring the hydrogen sulfide concentration at the wellhead, working area, and core storage area in real time, it ensures the environmental safety of on-site construction. The lithology identification device uses an electron microscope with a built-in lens that can magnify 3000 times, thereby accurately identifying the lithology of the core, helping on-site geological engineers to make timely and accurate judgments, and reducing the impact of human error on the core identification results.

[0024] The temperature probe and portable hydrogen sulfide detector are connected to the main unit via a built-in high-strength steel wire cable, and the data is displayed in real time on the receiver desktop. The receiver is equipped with a main unit, keyboard, and display panel. It is connected to the temperature probe and portable hydrogen sulfide detector via a cable, accurately records data such as measurement time, well depth, temperature, and gas content, and automatically stores the measurement data. All data is transmitted to the cloud in real time through the data transmission system for backup.

[0025] In use, the temperature probe is placed at the well mud inlet / outlet and on the well wall. The detachable portable hydrogen sulfide detector is placed at the wellhead, working area, and core storage area. The temperature probes are connected in series via connecting cables, and the detachable portable hydrogen sulfide detectors are connected in parallel via connecting cables. The other end of the connecting cables is connected to the input terminals of the receiver's main unit, keyboard, and display panel. The display panel displays the measurement data in real time, and the receiver's main unit inputs the data to the cloud via network signal.

[0026] Implementation, for example Figure 2 As shown, the receiver incorporates a preprocessing module, a formation real temperature calibration module, a machine learning real-time temperature prediction module, a drilling process parameter optimization decision module, and a closed-loop feedback and model adaptive update module, as detailed below: The preprocessing module is responsible for executing step S1: multi-source temperature data acquisition and preprocessing. Specifically, the preprocessing module receives data signals from downhole measurement-while-drilling (MWD) temperature sensors, gas detection devices, and lithology identification devices, and connects to multi-source sensing devices such as distributed fiber optic temperature measurement cables, mud inlet / outlet thermometers, and surface reference thermometers. Each sensor continuously sends measurement signals to the data processing unit. Due to the different sampling frequencies and depth resolutions of the various sensors, the preprocessing unit first performs time and space alignment operations. Low-frequency mud inlet / outlet temperature and surface temperature data were interpolated onto the same high-frequency time series grid as the drilling measurement-while-drilling temperature and distributed fiber optic thermometry data using cubic spline interpolation. At the same time, based on the depth measurement location of each sensor and the current formation stratification model, the temperature value is mapped onto a standardized spatial node extending along the wellbore. To ensure data quality, a sliding window statistical method was used to detect abnormal jump points. When the temperature at a certain point exceeded the mean within the window by ±3 times the standard deviation, it was judged as abnormal and removed. Subsequently, the spatiotemporal correlation between adjacent depth points and historical time periods was used to complete the data through Kriging interpolation or linear regression.

[0027] The pre-processed standardized temperature vector, along with current construction parameters such as mud discharge, inlet temperature, rotation speed, and drilling pressure, is transmitted in real time to the formation real temperature calibration module.

[0028] The formation true temperature calibration module is responsible for performing step S2: formation true temperature calibration. Specifically, the formation true temperature calibration module restores the measured temperature affected by drilling fluid circulation disturbances to the true original formation temperature and provides uncertainty quantification information. It consists of three sub-units, each performing the following sub-steps: Step S2.1, forward modeling of transient heat transfer between wellbore and formation: The transient heat transfer model unit built into the formation real temperature calibration module establishes the following coupled model: the energy conservation equation of the drilling fluid in the wellbore and the heat conduction equation in the formation are shown in the following equations: in, and These represent drilling fluid temperature and formation temperature, respectively, both at depth. radial coordinates and time The function; , , For drilling fluid density, specific heat capacity, and average annular velocity; The overall heat transfer coefficient; The radius of the wellbore; The term representing frictional heat generation from the drill bit is approximately calculated based on factors such as drilling pressure, rotational speed, and drill bit size. ,in The coefficient of friction of the drill bit. For drilling pressure, For rotational speed, The diameter of the drill bit. The cross-sectional area of ​​the annulus; , , These are the density, specific heat capacity, and thermal conductivity of the geological strata, respectively.

[0029] This coupled model is solved spatially and temporally using the finite volume method or the finite difference method, thus constructing the original temperature profile of the formation to be determined. To simulate temperature The forward mapping relationship.

[0030] Step S2.2, Layered Temperature Correction and Inversion: The layered temperature correction and inversion unit of the formation true temperature calibration module uses historical temperature measurement data or temperature recovery data during pump shutdown to obtain the initial profile of formation temperature. The inversion objective function is shown in the following formula: In the formula, These are the spatiotemporal sampling points for the preprocessed measured temperature. The total number of sampling points used for inversion. For smooth difference operators, a first- or second-order discrete difference matrix is ​​typically used to constrain the formation temperature profile morphology and suppress the non-uniqueness of the inversion. This is the regularization coefficient.

[0031] When distinct lithological strata exist, independent geothermal gradients and thermal conductivity parameters for each layer can be introduced to achieve differentiated inversion based on stratification. The Levenberg-Marquardt algorithm is used to solve this least-squares problem, and the initial result is obtained after iterative convergence. and thermal conductivity of each layer The estimate.

[0032] Step S2.3, Real-time assimilation calibration based on ensemble Kalman filtering: The ensemble Kalman filtering real-time assimilation calibration unit of the formation true temperature calibration module dynamically tracks the formation thermal state during drilling. First, the state vector is defined. It includes the original formation temperature values ​​and thermal conductivity parameters of different layers at discrete nodes along the depth. It constructs a system containing... An initial set of members. Typically, the initial values ​​are taken as 50 to 100. The initial values ​​of the members are generated by superimposing a small statistical perturbation on the initial inversion result obtained in step S2.2 to characterize the initial uncertainty.

[0033] As the drilling process progresses, new construction parameters are obtained. At that time, perform the prediction step: for each set member Using forward model Its state is changed from the analysis value of the previous time step. Integrate forward to the current time step and add the model error term. , thus obtaining the prediction set As shown in the following formula: Subsequently, when new real-time temperature observation vectors containing drilling-while-drilling bottom hole temperature, distributed fiber optic multi-depth point temperature, etc., are received... Then, the update steps are performed: First, the background error covariance matrix is ​​calculated based on the prediction set. Calculate the Kalman gain using the following formula. : Finally, update each set member using the following formula to obtain the analytic set. : In the formula, For linear observation operators, In its nonlinear form, the state vector is mapped to the observation space; The observation error covariance matrix; To conform to a specific distribution of observation perturbations.

[0034] The mean of the analysis set is the formation temperature distribution and thermophysical parameter vector after calibration at that moment. At the same time, based on the statistical dispersion of the set, the 90% confidence interval of the formation temperature at each depth point is calculated, providing uncertainty quantification for subsequent optimization. This assimilation process is automatically triggered the moment new temperature measurement data is obtained, so that the calibration results continuously approach the true value. In addition, this unit also identifies working conditions such as tripping in and out of the drill string and sudden changes in mud parameters through discrete event simulation, and automatically triggers the assimilation process. It also uses the asymptotic fitting of the temperature recovery curve during pump shutdown for supplementary correction.

[0035] By introducing a real-time data assimilation technique based on ensemble Kalman filtering (EnKF), real-time observation data such as measurement while drilling (MWD) and distributed fiber optic thermometry (DTS) are dynamically fused with the wellbore-formation transient heat transfer model. This enables online inversion and continuous approximation of the true original formation temperature and thermophysical parameters, while quantifying the uncertainty of calibration results. This solves the problem of measured temperatures deviating from the true formation temperature due to drilling disturbances, providing a more accurate and reliable benchmark for subsequent optimization decisions.

[0036] The machine learning real-time temperature prediction module is responsible for executing step S3. Specifically, the machine learning real-time temperature prediction module uses the calibrated geothermal field output in step S2 as a benchmark to construct a hybrid prediction engine that includes a physical information neural network sub-model, a deep long short-term memory network sub-model, a lightweight gradient boosting tree sub-model, a working condition discrimination and switching unit, and an online drilling heat map query unit.

[0037] Physical information neural networks with deep and time Input to the network, output predicted temperature The network's loss function is a weighted average of the data matching error and the physical residual term, as shown in the following equation: Among them, the physical residual term As shown in the following formula: In the formula, and These are the data loss weighting coefficient and the physical loss weighting coefficient, respectively. For the number of physical points, The calibrated formation temperature given in Module B is used as the boundary condition.

[0038] The calibrated formation temperature is provided by the formation true temperature calibration module. As a boundary condition, the network output is forced to satisfy the wellbore heat transfer control equation, so that the model maintains physical consistency under unseen operating conditions.

[0039] The deep long short-term memory network directly learns the nonlinear mapping from time-series features such as displacement, rotational speed, drilling pressure, inlet temperature, calibration geothermal gradient, and historical bottom-hole circulation temperature to future bottom-hole circulation temperature, with the loss function being standard mean square error. The operating condition discrimination and switching unit monitors the rate of change of drilling parameters in real time: when the rate of change exceeds a preset threshold, it automatically switches to the deep long short-term memory network to capture transient responses; when the operating conditions tend to stabilize, it switches to the computationally less expensive lightweight gradient boosting tree model to balance prediction accuracy and computational efficiency. Furthermore, the online drilling heatmap query unit incorporates a multi-dimensional parameter-temperature mapping table generated offline by the transient hydraulic-thermal model, allowing real-time viewing of the cooling potential of different parameter combinations through interpolation, assisting in optimization decisions.

[0040] A hybrid prediction engine combining Physical Information Neural Network (PINN) and Deep Long Short-Term Memory Network (DLSTM) is employed. PINN ensures the mechanistic rationality of the prediction results by embedding the heat transfer equation as a physical constraint into the network training; DLSTM excels at capturing transient responses under abrupt changes in construction parameters. By leveraging the synergistic switching between the two, the shortcomings of slow computation of pure physical models and insufficient generalization of pure data models under unseen conditions are overcome, providing fast and reliable prediction support for real-time optimization. The drilling process parameter optimization decision module is responsible for executing step S4: drilling process parameter optimization decision. Specifically, the drilling process parameter optimization decision module uses the prediction from step S3 and the calibration information from S2 as a basis to solve for the optimal drilling parameters under the dual constraints of temperature and well control safety. The drilling process parameter optimization decision module includes a multi-objective optimization solution unit and a high-temperature anomaly identification and prevention unit.

[0041] The multi-objective optimization solution unit uses mud discharge rate, inlet temperature, rotational speed, and drilling pressure as decision variables to construct a multi-objective optimization model as shown in the following equation: In the formula, The bottom hole circulation temperature predicted by the machine learning real-time temperature prediction module; For safety targets, this value is taken as the upper limit of the temperature resistance of downhole tools. minus To reserve a safety margin; Power of the mud pump; This refers to the mechanical drilling rate; with subscripts. Represents the baseline value. Weighting coefficient. , , The system adapts to the severity of the overheating emergency.

[0042] Constraints include: temperature constraints ,in The calibration uncertainty confidence interval provided in step S2 is used to determine the appropriate safety margin; pressure constraint. That is, the bottom hole circulation pressure must be between the formation pore pressure and the fracture pressure, and the formula is used in the calculation. The problem is designed to correct for the effect of temperature on drilling fluid density; it also includes mechanical limitations on mud pump displacement, drilling pressure, and rotational speed. This nonlinear constrained optimization problem is solved efficiently using a sequential quadratic programming algorithm, which outputs the optimal parameter combination.

[0043] The high-temperature anomaly identification and prevention unit uses the high-precision temperature gradient profile calibrated in step two to identify abnormal strata that exceed the normal rate of temperature increase. It also combines the current mechanical drilling rate to estimate the drilling time of the abnormal strata and trigger preventive cooling strategies in advance, such as appropriately increasing the discharge rate or reducing the inlet temperature, thus transforming passive emergency response into proactive management.

[0044] Based on the high-precision formation temperature gradient profile obtained after calibration, it is possible to identify and determine the high-temperature abnormality zone in front of the drill bit, predict its drilling time, and thus trigger preventive cooling strategies in advance. This transforms temperature management from an emergency response to a critical temperature state into a proactive, predictive control that anticipates risks, significantly improving the safety margin of drilling operations.

[0045] The closed-loop feedback and model adaptive update module is used to execute step S5: closed-loop feedback and model adaptive update. The response-prediction deviation monitoring unit of the closed-loop feedback and model adaptive update module continuously compares the deviation between the actual response of the bottom hole circulating temperature and the predicted value in step three. The model mismatch judgment and triggering unit determines that the model mismatch is caused by a deviation exceeding 2°C for a continuous period (e.g., 15 minutes) after excluding abnormal conditions such as well leakage and well kick. At this time, the system automatically labels the recent actual drilling temperature data and injects it as a new observation into the ensemble Kalman filter assimilation loop in step two, and updates the formation temperature distribution and thermophysical parameters to achieve emergency recalibration of the formation temperature model. At the same time, it uses the latest accumulated data to perform small-batch incremental learning on the depth long short-term memory network and fine-tunes the network weights to adapt to the new formation conditions. Even if the deviation does not exceed the limit, the ensemble Kalman filter assimilation loop continues to run at a fixed frequency (e.g., every 5 meters or every 30 minutes), so that the system can adaptively maintain calibration and prediction accuracy without human intervention, and completely close the closed-loop link of calibration, prediction, optimization, execution, feedback, and recalibration.

[0046] Based on the same inventive concept, another embodiment also provides a data processing method for a portable logging instrument based on temperature calibration, such as... Figure 3 As shown, steps S1 to S5 in the above embodiments are included, namely: Step S1: Multi-source temperature data acquisition and preprocessing It collects multi-source data in real time, including downhole drilling temperature measurement, distributed fiber optic temperature measurement, mud inlet and outlet temperature, and surface reference temperature; it achieves time alignment through cubic spline interpolation and spatial alignment by depth mapping; and outputs a standardized temperature data stream after outlier detection and spatiotemporal correlation completion.

[0047] Step S2: Formation True Temperature Calibration Step S2.1: Establish a forward model of transient heat transfer between the wellbore and the formation to describe the dynamic response of drilling fluid temperature and formation temperature; Step S2.2: Utilize historical or pump-stop temperature recovery data and construct a cost function with a smooth regularization term to obtain the original formation temperature initial profile and thermophysical parameters through inversion; Step S2.3: During the drilling process, a state vector is constructed using the formation temperature node values ​​and the thermal conductivity of the section to generate an initial set; whenever new temperature measurement data is obtained, a prediction step is executed, the Kalman gain is calculated, and the set members are updated. The set mean is analyzed as the calibrated true formation temperature, and the uncertainty confidence interval is output using the set statistical dispersion.

[0048] Step S3: Machine Learning for Real-Time Temperature Prediction Based on the calibrated geothermal field output in step two, a hybrid prediction engine is constructed: the residual of the wellbore heat transfer control equation is embedded as a physical loss term into a physical information neural network to train and output a physically consistent wellbore temperature distribution; a deep long short-term memory network is constructed to learn the mapping between construction parameters and calibrated geothermal temperature to future bottom hole circulation temperature; the transient prediction model and the steady-state prediction model are automatically switched according to the rate of change of drilling parameters; and the interpolation results of the drilling heat map are used to assist in judging the cooling potential.

[0049] Step S4: Drilling process parameter optimization decision Using mud discharge, inlet temperature, rotational speed, and drilling pressure as decision variables, a multi-objective optimization model is constructed: the optimization objectives are the deviation between the predicted bottom hole circulating temperature and the safety target, pump power, and mechanical drilling speed, with the constraints being the temperature safety limit plus uncertainty safety margin, the bottom hole circulating pressure being between pore pressure and fracture pressure, and equipment capacity; the optimal drilling parameters are solved using a sequential quadratic programming algorithm and executed; at the same time, based on the calibrated temperature gradient profile, the high temperature anomaly zone ahead is identified, and preventive cooling is initiated in advance.

[0050] Step S5: Closed-loop feedback and adaptive model update The deviation between the actual bottom hole circulation temperature response and the predicted value in step three is continuously compared. When the deviation exceeds the limit continuously, the model is judged to be mismatched, and the recent actual drilling temperature data is automatically injected into the ensemble Kalman filter assimilation loop in step two as new observations to update the formation temperature model. At the same time, the latest data is used to perform small batch incremental learning on the depth long short-term memory network to achieve adaptive continuous optimization of the model.

[0051] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any way. Although the present invention has been disclosed above with reference to preferred embodiments, it is not intended to limit the present invention. Any person skilled in the art can make some modifications or alterations to the above-disclosed technical content to create equivalent embodiments without departing from the scope of the present invention. Any indirect modifications, equivalent changes, and alterations made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the scope of the present invention.

Claims

1. A portable logging instrument based on temperature calibration, characterized in that, The system includes a data acquisition module and a receiver wired to the data acquisition module. The data acquisition module includes a temperature probe, a gas detection device, and a lithology identification device. The receiver has the following built-in components: The preprocessing module preprocesses the real-time temperature data and outputs a standardized temperature data stream. The formation true temperature calibration module receives a standardized temperature data stream, uses the wellbore-formation transient heat transfer model and real-time temperature measurement data to assimilate the data, and dynamically calibrates the formation true temperature and its uncertainty information. The machine learning real-time temperature prediction module uses the calibrated actual formation temperature as a benchmark and outputs the predicted values ​​of bottom hole circulation temperature and wellbore temperature distribution under different drilling conditions through a hybrid prediction engine. The drilling process parameter optimization decision module solves for the optimal drilling parameters within the feasible region jointly defined by temperature constraints and well control safety constraints, based on predicted values ​​and uncertainty information. The closed-loop feedback and model adaptive update module monitors the temperature response deviation after drilling parameter adjustment, and triggers the formation real temperature calibration module and the machine learning real-time temperature prediction module to perform adaptive updates when the deviation exceeds the limit.

2. The portable logging instrument based on temperature calibration according to claim 1, characterized in that, The formation true temperature calibration module includes: The wellbore-formation transient heat transfer model unit establishes a coupled forward model consisting of the drilling fluid energy conservation equation and the formation heat conduction equation, describing the dynamic response of the temperature field of the wellbore and the formation under given engineering parameters and formation thermal properties. The layered temperature correction inversion unit constructs a cost function with a smooth regularization term based on the residual between the measured temperature and the model simulated temperature, and obtains the original formation temperature profile and thermal conductivity parameters of each layer through inversion. The Kalman filter real-time assimilation calibration unit defines a state vector containing formation temperature node values ​​and thermal conductivity parameters. Through iterative prediction and update steps, the real-time temperature observations are dynamically fused with the coupled forward model to output the calibrated true formation temperature distribution and uncertainty confidence interval.

3. A portable logging instrument based on temperature calibration according to claim 2, characterized in that, In the aforementioned Kalman filter real-time assimilation calibration unit: An initial set containing multiple members is generated based on the output of the hierarchical temperature correction inversion unit. The initial value of each member is formed by superimposing statistical perturbations on the inversion result. When new construction parameters are obtained during drilling, the coupled forward model is used to integrate each set member from the analysis state of the previous moment to the current moment, and the model error term is added to obtain the prediction set. Upon receiving a new real-time temperature observation vector, the Kalman gain is calculated based on the statistical background error covariance matrix of the prediction set, and the Kalman gain and observation perturbation are used to update each set member to obtain the analysis set. The mean of the analysis set is used as the calibrated formation temperature distribution and thermophysical parameters, and the statistical dispersion of the analysis set is used to output the calibrated temperature confidence interval for each depth point.

4. A portable logging instrument based on temperature calibration according to claim 1, characterized in that, The machine learning real-time temperature prediction module includes: The physical information neural network sub-model takes depth and time as network inputs, embeds the wellbore heat transfer control equation as a physical residual term into the loss function, and outputs the predicted wellbore temperature distribution that satisfies the physical constraints. The temporal neural network sub-model learns the temporal mapping relationship between construction parameters, calibration geothermal gradient and historical bottom hole circulation temperature, and outputs the predicted value of future bottom hole circulation temperature. The working condition discrimination and switching unit automatically switches between the physical information neural network sub-model, the temporal neural network sub-model, or the lightweight gradient boosting tree model for prediction based on the rate of change of drilling parameters. The online drilling heat map query unit, based on an offline pre-calculated parameter-temperature mapping table, can query the cooling potential of different parameter combinations in real time.

5. A portable logging instrument based on temperature calibration according to claim 4, characterized in that, The loss function of the physical information neural network sub-model is composed of a weighted sum of a data matching error term and a physical residual term. The physical residual term is the mean square value of the residual of the wellbore drilling fluid energy conservation equation, and the calibrated formation temperature output by the formation true temperature calibration module is used as the boundary condition.

6. A portable logging instrument based on temperature calibration according to claim 1, characterized in that, The drilling process parameter optimization decision module includes: The multi-objective optimization solution unit constructs a multi-objective optimization model with cooling deviation, pump power saving and mechanical drilling speed improvement as optimization objectives and temperature safety constraints and wellbore pressure balance constraints as constraints, and solves for the optimal drilling parameters; The high-temperature anomaly identification and prevention unit identifies the high-temperature anomaly zone ahead based on the calibrated formation temperature gradient profile, estimates the drilling time to the anomaly section, and outputs preventive cooling strategies in advance.

7. A portable logging instrument based on temperature calibration according to claim 1, characterized in that, The optimization model constructed by the multi-objective optimization solution unit uses mud discharge, inlet temperature, rotational speed and drilling pressure as decision variables; The temperature safety constraints include a predicted bottomhole circulation temperature plus a safety margin determined by the calibration uncertainty confidence interval that is less than or equal to the upper limit of the downhole tool temperature resistance. The wellbore pressure balance constraint includes the bottom hole circulating pressure being between the formation pore pressure and the fracture pressure, and the bottom hole circulating pressure calculation incorporates the correction for drilling fluid density due to temperature.

8. A portable logging instrument based on temperature calibration according to claim 1, characterized in that, The closed-loop feedback and model adaptive update module includes: The response-prediction deviation monitoring unit continuously compares the deviation between the actual response value of the bottom hole circulation temperature after drilling parameter adjustment and the predicted value output by the machine learning real-time temperature prediction module. The model mismatch determination and triggering unit determines that the model is mismatched when the deviation continuously exceeds a preset threshold within a continuous period of time. It then automatically injects recent actual drilling temperature data into the Kalman filter assimilation loop of the formation real temperature calibration module for recalibration, and simultaneously triggers the machine learning real-time temperature prediction module to perform incremental learning and updates.

9. A data processing method for a portable logging instrument based on temperature calibration, characterized in that, Each step of geothermal exploration data processing is achieved using a portable logging instrument based on temperature calibration as described in any one of claims 1-8.

10. A portable logging instrument data processing method based on temperature calibration according to claim 9, characterized in that, Includes the following steps: Step S1: Collect temperature data in real time from multiple temperature sensing sources, and perform time alignment, spatial alignment and quality control preprocessing to output a standardized temperature data stream; Step S2: Receive the standardized temperature data stream, use the wellbore-formation transient heat transfer model and real-time temperature measurement data to assimilate the data, and dynamically calibrate the true formation temperature and its uncertainty information; Step S3: Based on the calibrated actual formation temperature, output the predicted values ​​of bottom hole circulation temperature and wellbore temperature distribution under different drilling conditions through the hybrid prediction engine. Step S4: Based on the predicted values ​​and the uncertainty information, solve for the optimal drilling parameters within the feasible region jointly defined by temperature constraints and well control safety constraints; Step S5: Monitor the temperature response deviation after drilling parameter adjustment, and trigger the formation temperature recalibration in step S2 and the adaptive update of the prediction model in step S3 when the deviation exceeds the limit, forming a closed loop.

Images