Method for identifying mineshaft micro-leakage through distributed temperature gradient sensing data

Abstract

The invention discloses a method for identifying micro-leakage of a shaft by using distributed temperature gradient sensing data, which comprises the following steps of: acquiring original temperature data of a continuous time sequence of different depths of the shaft through a distributed temperature sensing device, and transmitting the original temperature data to a Modin distributed sensor data processing platform; the platform divides and distributes the data and screens abnormal data; calling a temperature field dynamic evolution prediction algorithm to calculate a predicted temperature value of each depth at each moment; starting a temperature gradient space-time correlation detection algorithm, comparing the predicted temperature with the actual temperature to obtain a deviation sequence, and analyzing and extracting temperature gradient change characteristics; an ultrasonic excitation thermoelastic effect recognition model is applied, ultrasonic excitation is applied, temperature response data are collected, and whether micro leakage exists in the shaft or not and the specific position are determined through comparison and analysis of the incidence relation between the thermoelastic effect and the temperature gradient change. According to the method, an efficient and reliable solution is provided for shaft micro-leakage identification, safe operation of the shaft is guaranteed, and stratum pollution and resource waste are reduced.

Description

Technical Field

[0001] This invention relates to the field of wellbore leakage identification technology, and in particular to a method for identifying wellbore micro-leakage using distributed temperature gradient sensing data. Background Technology

[0002] In oil and gas extraction and related energy engineering activities, the wellbore serves as the core channel for fluid transportation and extraction, and its integrity directly affects extraction efficiency and operational safety. With the extension of wellbore service life and the increasing complexity of downhole conditions, micro-leakage problems are becoming increasingly prominent. These leaks, due to their small pore size and low leakage volume, are difficult to detect initially using conventional detection methods, and their long-term development can easily lead to formation pollution, resource waste, and even safety accidents. Currently, the industry largely relies on distributed temperature sensing technology to obtain wellbore temperature data. However, how to accurately extract micro-leakage characteristics from massive, multi-dimensional temperature data to achieve efficient and reliable leak identification and location has become a critical issue that urgently needs to be addressed to ensure the safe operation of the wellbore. This is especially true in deep wells and complex formations, where higher demands are placed on the processing speed of temperature data and the adaptability of the identification model. A systematic identification method integrating efficient data processing and accurate algorithm models is urgently needed.

[0003] Existing technologies for identifying micro-leaks in wellbore have two significant drawbacks: First, insufficient data processing capabilities. Traditional data processing methods often employ single-node or simple parallel architectures, which are prone to data processing delays and uneven task allocation when faced with continuous spatiotemporal temperature data generated by distributed temperature sensing technology. This makes it difficult to quickly provide effective data support for subsequent leak feature extraction and meet the needs of real-time downhole monitoring. Second, limited identification accuracy. Existing methods often rely on single temperature parameters or simple gradient analysis, failing to fully integrate multi-dimensional information such as the dynamic evolution of the temperature field and the ultrasonic excitation thermoelastic effect. This results in insufficient ability to capture the subtle temperature change features caused by micro-leaks, making them susceptible to interference from normal temperature fluctuations in the wellbore, leading to misjudgments or missed detections of leaks. Furthermore, the lack of a scientific model for multi-parameter collaborative calculation in leak location positioning makes it difficult to accurately pinpoint the location of micro-leaks. Summary of the Invention

[0004] In order to overcome the shortcomings and deficiencies of existing technologies, this invention provides a method for identifying wellbore micro-leakage using distributed temperature gradient sensing data.

[0005] The technical solution adopted in this invention is a method for identifying wellbore micro-leakage using distributed temperature gradient sensing data, comprising the following steps: S1, collecting raw temperature data at different depths in the wellbore over a continuous time series using a distributed temperature sensing device, and transmitting the collected raw temperature data to the Modin distributed sensor data processing platform; S2, based on the Modin distributed sensor data processing platform, dividing and assigning tasks to the transmitted raw temperature data, establishing a multi-node parallel processing architecture, and performing preliminary screening of the temperature data received by different nodes to remove obviously abnormal data; S3, calling a temperature field dynamic evolution prediction algorithm, using the data filtered in S2 as input, constructing a spatiotemporal evolution model of the wellbore temperature field, and calculating the temperature field evolution at different depths at different times. S4. Activate the temperature gradient spatiotemporal correlation detection algorithm, calculate the difference between the predicted temperature value obtained in S3 and the actual temperature data collected at the corresponding time, obtain the temperature deviation sequence, and then perform spatiotemporal correlation analysis on the temperature deviation sequence to extract temperature gradient change characteristics; S5. Apply the ultrasonic excitation thermoelastic effect identification model, apply an ultrasonic excitation signal to a designated area of ​​the wellbore, collect the temperature response data after excitation, and combine it with the temperature gradient change characteristics extracted in S4 to construct the correlation between the thermoelastic effect and the temperature gradient change; S6. Based on the correlation established in S5, set a wellbore micro-leakage identification threshold, compare and analyze the temperature gradient change characteristics at different depths with the identification threshold, and determine whether there is a micro-leakage in the wellbore and the specific location of the micro-leakage.

[0006] Furthermore, the expression used in the temperature field dynamic evolution prediction algorithm is: ,in, Let be the predicted temperature value at depth x at time t. The temperature decay coefficient is Let x be the temperature value at depth x at the initial moment. The time decay factor, Where is the thermal diffusivity, for The Laplace operator for the temperature at depth x at time t. The heat source influence coefficient, Let be the heat source intensity at depth x at time t, which is related to the heat generated by fluid leakage during wellbore micro-leakage.

[0007] Furthermore, the expression used in the temperature gradient spatiotemporal correlation detection algorithm is: ,in, The associated value of the temperature gradient at depth x at time t. Let x be the spatial temperature gradient at depth x at time t. For depth weighting coefficients, Let be the time-temperature gradient at depth x at time t. For time weighting coefficients, The correlation coefficient, At time t, at depth x, and Time depth Covariance of temperature data For depth intervals, The time interval is used to determine the correlation between temperature data at different times and locations, and the covariance reflects the degree of correlation between temperature data at different times and locations. It is used to determine the propagation characteristics of temperature anomalies caused by micro-leakage in the wellbore.

[0008] Furthermore, the expression used in the ultrasonic excitation thermoelastic effect identification model is: ,in, Let be the temperature change at depth x caused by ultrasonic excitation at time t. Thermoelastic conversion coefficient, Let be the ultrasonic excitation pressure at depth x at time t. Let x be the elastic modulus of the wellbore medium at depth x. The dynamic thermoelastic coefficient, Let be the rate of change of ultrasonic excitation pressure at depth x at time t. Let x be the thermal conductivity coefficient of the wellbore structure at depth x, and let x be the temperature change. This is used to correlate the effect of medium changes caused by wellbore microleakage under ultrasonic excitation on the thermoelastic effect.

[0009] Furthermore, the data processing efficiency model expression of the Modin distributed sensor data processing platform is as follows: Where E represents the platform's data processing efficiency, N represents the total number of temperature data entries collected, D represents the number of bytes per data entry, and k represents the number of parallel processing nodes on the platform. Let i be the data reception time of the i-th node. Let i be the data processing time for the i-th node. The node coordination coefficient is used to optimize the platform's processing speed for temperature data associated with wellbore micro-leakage, ensuring timely data processing to support subsequent identification steps.

[0010] Furthermore, the model expression for determining the location of the wellbore micro-leakage is as follows: Where L is the depth of the micro-leakage location, Here, m represents the initial reference depth, and m is the number of depth points used in the calculation. The temperature gradient correlation value at depth xj at time tj is given. The temperature gradient is used as a baseline value when there is no leakage. This model can accurately locate the specific depth where micro-leakage occurs in the wellbore.

[0011] Further, step S3 includes the following sub-steps: S31, retrieve the temperature data filtered in S2 from the Modin distributed sensor data processing platform, reorder the data according to the time series and depth dimension to form a structured temperature dataset, ensuring the continuity of the data in the spatiotemporal dimension; S32, input the structured temperature dataset into the initialization module of the temperature field dynamic evolution prediction algorithm, set the initial parameters of the algorithm, including the temperature diffusion coefficient, the initial temperature reference value and the parameters related to the wellbore environment, and establish the initial conditions for algorithm calculation; S33, based on the initial conditions, run the core calculation module of the temperature field dynamic evolution prediction algorithm, use the finite difference method to solve the spatiotemporal evolution equation of the temperature field, and obtain the intermediate temperature calculation values ​​at different depth positions at each time step; S34, perform spatiotemporal consistency verification on the intermediate temperature calculation values, compare the temperature calculation values ​​at the same depth position at adjacent time steps and adjacent depth positions at the same time step, correct unreasonable intermediate calculation values, and finally output the predicted temperature values ​​at different depth positions at different times.

[0012] Further, step S4 includes the following sub-steps: S41, obtaining predicted temperature values ​​at different depths at different times from the temperature field dynamic evolution prediction algorithm, and simultaneously retrieving actual temperature data collected at the corresponding times from the distributed temperature sensing device, establishing a one-to-one correspondence between the predicted temperature and the actual temperature; S42, performing difference calculations on each set of corresponding data to obtain temperature deviation values ​​at different depths at different times, arranging all temperature deviation values ​​in order of time and depth to form a temperature deviation sequence; S43, calling the correlation analysis module of the temperature gradient spatiotemporal correlation detection algorithm to perform gradient calculation in the spatial dimension and trend analysis in the time dimension on the temperature deviation sequence, extracting the rate of change of temperature deviation at different depths and the degree of correlation between temperature deviations at adjacent depths; S44, constructing a temperature gradient change feature vector based on the correlation analysis results, the feature vector including spatial gradient values, time rate of change, and spatiotemporal correlation coefficient parameters, which are used as input for the subsequent thermoelastic effect identification model.

[0013] Further, step S5 includes the following sub-steps: S51, applying ultrasonic excitation signals of different frequencies and intensities to multiple preset detection areas in the wellbore using an ultrasonic excitation device, and recording the excitation signal parameters of each detection area, including excitation frequency, excitation intensity, and excitation duration; S52, using a distributed temperature sensing device to synchronously collect temperature response data of each detection area at different times during ultrasonic excitation, ensuring the time synchronization of temperature response data and excitation signals; S53, inputting the collected temperature response data into the ultrasonic excitation thermoelastic effect identification model, and combining it with the temperature gradient change feature vector extracted in S4 to establish a mapping relationship between temperature response and temperature gradient change under the thermoelastic effect; S54, performing a significance test on the mapping relationship, and selecting mapping relationship pairs with strong correlation as the core correlation basis for subsequent wellbore micro-leakage identification.

[0014] A method for identifying wellbore micro-leakage using distributed temperature gradient sensing data is implemented through different units, including: a distributed temperature data acquisition unit, which consists of multiple distributed temperature sensors spaced along the depth direction of the wellbore to collect raw temperature data at different depths and transmit the collected data to a data transmission unit; a data transmission unit, which employs an optical fiber transmission module and a wireless backup transmission module. The optical fiber transmission module is used for high-speed and stable transmission of large amounts of raw temperature data, while the wireless backup transmission module ensures data transmission continuity in case of optical fiber transmission failure, transmitting the data to a Modin distributed data processing unit; and a Modin distributed data processing unit, which includes multiple parallel processing nodes and a node collaborative control module. The parallel processing nodes are used to filter the received raw temperature data, and the node collaborative control module is used to manage data between different nodes. The data is allocated and scheduled, and the processed data is transmitted to the algorithm processing unit. The algorithm processing unit integrates a temperature field dynamic evolution prediction algorithm module, a temperature gradient spatiotemporal correlation detection algorithm module, and an ultrasonic excitation thermoelastic effect identification model module. Different modules perform calculations and analyses on the processed data, outputting temperature gradient change characteristics and correlation data to the micro-leakage identification unit. The micro-leakage identification unit includes a threshold setting module and a comparison analysis module. The threshold setting module sets the micro-leakage identification threshold based on wellbore parameters, while the comparison analysis module compares the temperature gradient change characteristics with the identification threshold, outputting the micro-leakage judgment result and location information to the result display unit. The result display unit uses a visual interface and a data storage module. The visual interface displays the wellbore micro-leakage status and location information in real time, while the data storage module stores the raw data, calculation data, and identification results for querying and analysis.

[0015] Beneficial Effects: This invention proposes a method for identifying wellbore micro-leakage using distributed temperature gradient sensing data. By introducing the Modin distributed sensor data processing platform and constructing a multi-node parallel processing architecture, it achieves efficient partitioning and task allocation of massive spatiotemporal temperature data, solving the problems of data processing latency and uneven task allocation under traditional single-node or simple parallel architectures. This quickly provides effective data support for subsequent feature extraction, meeting the real-time monitoring needs of downhole systems. Simultaneously, it integrates a temperature field dynamic evolution prediction algorithm, a temperature gradient spatiotemporal correlation detection algorithm, and an ultrasonic excitation thermoelastic effect identification model, eliminating reliance on a single temperature parameter or simple gradient... Instead of relying solely on temperature analysis, this method combines multi-dimensional information such as the dynamic evolution of the temperature field, spatiotemporal correlation characteristics, and thermoelastic effects to enhance the ability to capture subtle temperature changes in micro-leaks. This reduces interference from normal temperature fluctuations in the wellbore, avoiding misjudgments or omissions of leaks. Furthermore, a micro-leak location determination model based on multi-parameter collaborative calculations enables precise location pinpointing. In addition, this method forms a systematic identification process, with each step—from data acquisition and processing to algorithm calculation and result judgment—being interconnected. It maintains good adaptability even in deep wells and complex formations, providing an efficient and reliable solution for wellbore micro-leak identification, ensuring safe wellbore operation, and reducing formation pollution and resource waste. Attached Figure Description

[0016] Figure 1 This is a flowchart of the method steps of the present invention;

[0017] Figure 2 This is a diagram showing the unit composition for implementing the method of the present invention. Detailed Implementation

[0018] It should be noted that, unless otherwise specified, the embodiments and features described in this application can be combined with each other. The application will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0019] like Figure 1 As shown, a method for identifying wellbore micro-leakage using distributed temperature gradient sensing data includes the following steps:

[0020] S1. Collect raw temperature data at different depths in the wellbore under continuous time series using distributed temperature sensing devices, and transmit the collected raw temperature data to the Modin distributed sensor data processing platform.

[0021] Specifically, step S1 completes the acquisition and transmission of wellbore temperature data. In practice, distributed temperature sensing devices are first deployed along the depth direction within the wellbore. Fiber optic sensors based on the Raman scattering principle are selected, and the spacing between them is determined by the wellbore depth. When the wellbore depth is between 1000 and 3000 meters, the sensor spacing is set to 0.5 meters; when the depth exceeds 3000 meters, the spacing is adjusted to 0.3 meters to ensure spatial resolution of the data acquisition. The data acquisition time series is set to acquire data every 10 seconds, with a continuous acquisition duration of no less than 72 hours to ensure coverage of the dynamic temperature change cycle within the wellbore. The accuracy of the acquired raw temperature data is controlled within ±0.1℃ to avoid affecting subsequent analysis due to insufficient data accuracy. During the acquisition process, the optical signal is converted into an electrical signal by the signal conversion module of the sensing device, and then transmitted to the Modin distributed sensor data processing platform via a data interface. The transmission rate is set to 100 Mbps to ensure efficient transmission of massive amounts of data without loss or delay. This step provides the foundational data for all subsequent analyses. Only by ensuring the spatial resolution, temporal continuity, and accuracy of the original data can we accurately capture the subtle temperature changes that may be caused by micro-leakage in the wellbore. If the sampling interval is too large or the duration is insufficient, leakage characteristic data may be missed, affecting the final identification results.

[0022] S2. Based on the Modin distributed sensor data processing platform, the transmitted raw temperature data is divided and tasks are allocated. A multi-node parallel processing architecture is established, and the temperature data received by different nodes is initially screened to remove obviously abnormal data.

[0023] Specifically, step S2 utilizes the Modin distributed sensor data processing platform to preprocess the raw temperature data. During implementation, the platform's multi-node parallel processing architecture is first activated. The number of nodes is determined based on the amount of data collected. When the amount of collected data is 100GB-500GB, 10-20 parallel processing nodes are activated. Each node is allocated 16GB of memory resources and 8 CPU cores to ensure that the nodes have sufficient computing power to process the data. The platform divides the raw data into multiple data blocks according to the time dimension through the data partitioning module. Each data block includes 1 hour of continuous temperature data. Then, the task allocation module evenly distributes the data blocks to each parallel node. After receiving the data, each node initiates a preliminary screening process. The screening criterion is to remove data exceeding the normal temperature range. The normal temperature range for wellbore is determined based on geological conditions; in onshore oil and gas wells, the normal temperature range is typically 25-150℃. Data exceeding this range by ±20℃ is considered significantly abnormal. Simultaneously, null and duplicate values ​​generated during data transmission are also removed. After each node completes its processing, the filtered data is transmitted to the platform's data integration module. The integration module sorts the data according to the depth-time dimension, forming a structured dataset. This step reduces the interference of invalid data on subsequent algorithm calculations, improving algorithm processing efficiency. Failure to remove abnormal data may lead to deviations in subsequent temperature field predictions, affecting the accuracy of leak identification.

[0024] S3. Call the temperature field dynamic evolution prediction algorithm, use the data filtered in S2 as input, construct the spatiotemporal evolution model of the wellbore temperature field, and calculate the predicted temperature values ​​at different depths at different times.

[0025] Specifically, step S3 constructs a spatiotemporal evolution model of the wellbore temperature field using a dynamic temperature field evolution prediction algorithm and calculates the predicted temperature value. During implementation, the structured dataset formed in step S2 is retrieved from the Modin platform and imported into the algorithm's input module. The input module performs data format conversion to ensure the data meets the algorithm's processing requirements. The algorithm's parameter configuration module sets key parameters based on the wellbore parameters, where the thermal diffusivity is determined according to the wellbore casing material. If the casing is P110 steel, the thermal diffusivity is set to 1.1 × 10⁻⁻⁻⁶. 5The initial temperature baseline is the average temperature of the previous 24 hours selected in step S2, measured in square meters per second. The time step is set to 10 seconds, consistent with the data acquisition time step, and the spatial step is consistent with the sensor deployment spacing to ensure spatiotemporal matching. After the algorithm starts, the core calculation module performs evolution calculations on the temperature field based on the heat conduction equation. An iterative calculation method is used during the calculation process. After each iteration, the calculation results are checked for errors, with an error threshold set at 0.5℃. When the iteration error is less than the threshold, the iteration stops and the predicted temperature values ​​at each depth position at that moment are output. After the calculation is completed, the predicted temperature values ​​are stored in the algorithm's result output module according to the depth-time dimension, and simultaneously transmitted to the Modin platform's temporary database for easy retrieval in subsequent steps. This step establishes a predictive model for normal temperature changes in the wellbore, providing a benchmark for subsequent comparative analysis. By comparing the predicted temperature with the actual temperature, it is possible to preliminarily determine whether there are abnormal temperature changes, laying the foundation for leak feature extraction.

[0026] S4. Activate the temperature gradient spatiotemporal correlation detection algorithm, calculate the difference between the predicted temperature value obtained in S3 and the actual temperature data collected at the corresponding time, obtain the temperature deviation sequence, and then perform spatiotemporal correlation analysis on the temperature deviation sequence to extract the temperature gradient change characteristics.

[0027] Specifically, step S4 uses a temperature gradient spatiotemporal correlation detection algorithm to analyze temperature deviation and extract temperature gradient change features. During implementation, the predicted temperature value from step S3 is retrieved from the Modin platform temporary database, and the actual collected temperature data at the corresponding time is retrieved from the historical database of the distributed temperature sensing device. The two sets of data are imported into the algorithm's data matching module. The matching module matches the data according to the "depth-time" one-to-one correspondence principle to ensure that each set of data includes the predicted temperature value and the actual temperature value at the same depth and time. After pairing, the algorithm's difference calculation module performs difference calculations on each set of data to obtain the temperature deviation value at each depth and time. Then, the sequence construction module arranges the temperature deviation values ​​in ascending order of depth and time, forming a temperature deviation sequence. Next, the spatiotemporal correlation analysis module is activated. This module first calculates the spatial temperature gradient by dividing the difference in temperature deviation values ​​between adjacent depths by the sensor spacing. Then, it calculates the temporal temperature gradient by dividing the difference in temperature deviation values ​​between adjacent times by the time step. Next, the correlation analysis algorithm calculates the correlation degree between temperature deviation values ​​at different depths and times. The correlation degree is determined by the correlation coefficient; an absolute correlation coefficient greater than 0.8 indicates a strong correlation, and less than 0.3 indicates a weak correlation. Finally, the feature extraction module extracts temperature gradient change features such as spatial temperature gradient, temporal temperature gradient, and strongly correlated regions, and stores the feature data in the Modin platform's feature database. This step extracts features with leakage identification value from the temperature deviation. Through spatiotemporal correlation analysis, it can distinguish between normal temperature fluctuations and abnormal temperature changes caused by leaks, avoiding misjudging normal fluctuations as leaks.

[0028] S5. Apply the ultrasonic excitation thermoelastic effect identification model, apply ultrasonic excitation signal to a designated area of ​​the wellbore, collect temperature response data after excitation, and combine the temperature gradient change characteristics extracted in S4 to construct the correlation between thermoelastic effect and temperature gradient change.

[0029] Specifically, step S5 utilizes an ultrasonic-excited thermoelastic effect identification model to construct the correlation between the thermoelastic effect and temperature gradient changes. During implementation, ultrasonic excitation devices are first deployed along the depth direction on the outer wall of the wellbore. The deployment positions of the excitation devices correspond one-to-one with the positions of the distributed temperature sensors. The frequency adjustment range of the excitation devices is set to 20-100kHz, and the power adjustment range is set to 50-200W. Before starting the excitation devices, the excitation parameters are set through the parameter setting module. The excitation frequency is determined according to the wellbore diameter. When the wellbore diameter is 150-200 mm, the frequency is set to 30kHz; when the diameter is 200-300 mm, the frequency is set to 25kHz. The excitation duration is set to 5 minutes to ensure a stable thermoelastic effect is generated. After the excitation device is activated, the distributed temperature sensing device is simultaneously activated, collecting temperature response data of the excitation area at the original acquisition frequency (10 seconds / time) for a duration of 10 minutes, including temperature data for 2 minutes before excitation, 5 minutes during excitation, and 3 minutes after excitation, to ensure complete capture of the temperature change process caused by the thermoelastic effect. The collected temperature response data and the temperature gradient change features extracted in step S4 are imported into the ultrasonic excitation thermoelastic effect recognition model. The model calculates the matching degree between the temperature response data and the temperature gradient change features through the correlation analysis module. The matching degree calculation uses a cosine similarity algorithm, and the matching degree is corrected by combining the thermophysical properties of the wellbore medium (such as specific heat capacity and thermal conductivity). The correction coefficient is determined according to the medium type; if the medium is crude oil, the correction coefficient is set to 0.92, and if it is brine, the correction coefficient is set to 0.95. Finally, a correlation table between the thermoelastic effect and the temperature gradient change is generated and stored in the correlation database of the Modin platform. This step enhances the temperature characteristics of the leak area through active ultrasonic excitation, establishes the correlation between the thermoelastic effect and the leak characteristics, improves the sensitivity of micro-leakage identification, and avoids missed detection due to weak leak characteristics.

[0030] S6. Based on the correlation established in S5, set the wellbore micro-leakage identification threshold, compare and analyze the temperature gradient change characteristics at different depths with the identification threshold, and determine whether there is a micro-leakage in the wellbore and the specific location of the micro-leakage.

[0031] Specifically, step S6 determines and locates wellbore micro-leakage based on the established correlation. During implementation, the correlation table formed in step S5 is retrieved from the correlation database of the Modin platform and imported into the leak identification module. The leak identification module first sets a wellbore micro-leakage identification threshold. The threshold setting adopts a statistical method, based on the temperature gradient change characteristics and thermoelastic effect correlation data of historical leak-free wellbores, and calculates the upper limit of the 95% confidence interval as the identification threshold. If there is no historical data for the wellbore, leak-free data of the same type of wellbore under similar geological conditions are used as a reference to ensure that the threshold is targeted. Then, the temperature gradient change characteristics at each depth location are compared and analyzed with the identification threshold. The comparison method is to compare the characteristic value with the threshold at each depth and at each time. When the temperature gradient change characteristic value at a certain depth location exceeds the identification threshold for three consecutive times, the depth location is marked as a suspected leak point. For suspected leak points, the temperature response data and correlation table at that depth are further retrieved to verify whether the temperature response of the suspected leak point conforms to the thermoelastic effect characteristics caused by micro-leakage. The verification standard is that the matching degree between the temperature response data and the leakage characteristics in the correlation table exceeds 0.85. If the verification passes, a micro-leakage is determined to exist at that depth, and the start time and duration of the micro-leakage are recorded. If the verification fails, the suspected leak point is marked as an abnormal fluctuation point, and the possibility of leakage is ruled out. Finally, the micro-leakage location (accurate to ±0.1 meters), the leak determination result, and related data (predicted temperature, actual temperature, temperature deviation, temperature response curve) are compiled into a report by the positioning output module and transmitted to the display terminal of the wellbore monitoring center. Simultaneously, the data is stored in the historical database for subsequent traceability and analysis. This step completes the final transformation from data to results, ensuring the accuracy of leak determination through multi-dimensional verification, accurately locating the leak location, providing a clear basis for wellbore maintenance, and reducing increased maintenance costs and operational delays caused by inaccurate leak location.

[0032] Preferably, the expression used in the temperature field dynamic evolution prediction algorithm is: ,in, Let be the predicted temperature value at depth x at time t. The temperature decay coefficient is Let x be the temperature value at depth x at the initial moment. The time decay factor, Where is the thermal diffusivity, for The Laplace operator for the temperature at depth x at time t. The heat source influence coefficient, Let be the heat source intensity at depth x at time t, which is related to the heat generated by fluid leakage during wellbore micro-leakage.

[0033] Specifically, when implementing the dynamic evolution prediction algorithm for the temperature field, it is first clarified that the algorithm calculates the predicted temperature at each time and depth through multi-parameter collaborative calculation. The temperature decay coefficient is determined based on the lithology of the surrounding formation. If the formation is sandstone, the coefficient is set to 0.95; if it is mudstone, it is set to 0.92, ensuring that it conforms to the temperature decay law under different geological conditions. The initial temperature values ​​at each depth are the average temperature of the first 24 hours after screening in step S2. The time decay factor is set according to the heat dissipation rate of the wellbore, set to 0.001 near the wellhead, gradually decreasing with increasing depth, reaching 0.0005 at a depth of 3000 meters, to reflect the influence of depth on the temperature decay over time. The thermal diffusivity coefficient is determined by combining the comprehensive thermal conduction characteristics of the well casing and the formation medium. When the casing is P110 steel and the formation is sandstone, this coefficient is set to 1.1 × 10⁻⁻⁻⁶. 5 Square meters per second; if the stratum is mudstone, adjust to 0.9 × 10⁻ 5 The heat source intensity parameter is related to the heat generated by fluid leakage during micro-leakage in the wellbore. In the absence of leakage, this parameter is set to 0. When leakage occurs, it is calculated based on the type of leaking fluid (crude oil or brine) and the estimated leakage volume. For crude oil leakage, the heat source intensity increases by 50 per cubic meter of leakage, and for brine leakage, it increases by 30. During algorithm implementation, initial values ​​for each parameter are input first, and then the spatial distribution of the temperature field at different times is calculated through integration. After each time step (10 seconds), the calculation results are compared with historical temperature field data from leak-free wellbores. If the deviation exceeds 0.5℃, the thermal diffusivity coefficient and heat source intensity parameters are adjusted until the deviation meets the requirements. By accurately setting multi-dimensional parameters, the accuracy of temperature field prediction is improved, providing a reliable benchmark for subsequent temperature deviation analysis and avoiding excessive deviations between predicted and actual temperatures due to unreasonable parameter settings, which could affect leakage characteristic identification.

[0034] Preferably, the expression used in the temperature gradient spatiotemporal correlation detection algorithm is: ,in, The associated value of the temperature gradient at depth x at time t. Let x be the spatial temperature gradient at depth x at time t. For depth weighting coefficients, Let be the time-temperature gradient at depth x at time t. For time weighting coefficients, The correlation coefficient, At time t, at depth x, and Time depth Covariance of temperature data For depth intervals, The time interval is used to determine the correlation between temperature data at different times and locations, and the covariance reflects the degree of correlation between temperature data at different times and locations. It is used to determine the propagation characteristics of temperature anomalies caused by micro-leakage in the wellbore.

[0035] Specifically, when implementing the temperature gradient spatiotemporal correlation detection algorithm, the values ​​of each parameter of the algorithm are first determined. The depth weight coefficient is set according to the temperature sensitivity of different depths in the wellbore. It is set to 1.2 in the middle of the wellbore (1000-2000 meters), and 0.8 at the wellhead (0-500 meters) and the bottom of the well (above 3000 meters), because the temperature in the middle region is more significantly affected by leakage. The time weight coefficient is dynamically adjusted according to the time series. It is set to 0.9 in the initial stage of collection (first 24 hours), 1.0 in the middle stage (24-48 hours), and 1.1 in the later stage (48-72 hours) to highlight the impact of the temperature change trend in the later stage on the correlation analysis. The correlation coefficient is determined according to the strength of the spatiotemporal correlation of the wellbore temperature data. It is set to 0.7 under normal operating conditions and adjusted to 0.9 when a suspected temperature anomaly is detected to enhance the correlation analysis weight of abnormal data. The depth interval is consistent with the sensor deployment spacing (0.3-0.5 meters), and the time interval is the same as the data acquisition time step (10 seconds) to ensure the matching of spatiotemporal dimension parameters. During algorithm implementation, the spatial and temporal temperature gradients at each depth-time point are first calculated. Then, the correlation between temperature data at different spatiotemporal locations is analyzed through covariance calculation. When the absolute value of the covariance is greater than 0.8, a strong spatiotemporal correlation is determined in the corresponding area, indicating possible temperature anomaly propagation caused by leakage. If the absolute value of the covariance is less than 0.3, it is determined to be a weak correlation, considered as normal temperature fluctuation. After each full-wellbore correlation analysis (once per hour), the temperature gradient data of the strongly correlated area is stored. At the same time, the results of two adjacent analyses are compared. If the location of the strongly correlated area is fixed and its range expands, it is marked as a suspected leakage area. By dynamically adjusting the spatiotemporal weights and correlation coefficients, the algorithm accurately distinguishes between normal temperature fluctuations and abnormal temperature changes caused by leakage, improving the targeting of temperature gradient correlation analysis and avoiding misjudging weakly correlated normal fluctuations as leakage characteristics or missing strongly correlated leakage anomalies.

[0036] Preferably, the expression used in the ultrasonic excitation thermoelastic effect identification model is: ,in, Let be the temperature change at depth x caused by ultrasonic excitation at time t. Thermoelastic conversion coefficient, Let be the ultrasonic excitation pressure at depth x at time t. Let x be the elastic modulus of the wellbore medium at depth x. The dynamic thermoelastic coefficient, Let be the rate of change of ultrasonic excitation pressure at depth x at time t. Let x be the thermal conductivity coefficient of the wellbore structure at depth x, and let x be the temperature change. This is used to correlate the effect of medium changes caused by wellbore microleakage under ultrasonic excitation on the thermoelastic effect.

[0037] Specifically, when implementing the ultrasonic excitation thermoelastic effect identification model, the physical meaning and value standards of each parameter in the model are first clarified. The thermoelastic conversion coefficient is determined by combining the ultrasonic excitation intensity with the thermoelastic response characteristics of the wellbore medium. When the excitation frequency is 30kHz, the power is 100W, and the medium is crude oil, this coefficient is set to 0.85; if the medium is brine, it is adjusted to 0.75. The elastic coefficient of the wellbore medium is determined according to the medium type and temperature. The elastic coefficient of crude oil at 50℃ is set to 200MPa, and decreases by 10MPa for every 10℃ increase in temperature. The pressure of brine is set to 300 MPa at 50℃, and decreases by 15 MPa for every 10℃ increase in temperature. The dynamic thermoelastic coefficient reflects the effect of the ultrasonic excitation pressure change rate on temperature change. It is set to 0.6 when the excitation power change rate is 5 W / s and 0.8 when the change rate is 10 W / s. The thermal conductivity coefficient of the wellbore structure is determined in combination with the casing material and cement sheath characteristics. When P110 steel casing is combined with conventional cement sheath, the coefficient is set to 0.8 W / (m·K). If the cement sheath is of high strength type, it is adjusted to 1.0 W / (m·K). During model implementation, an excitation signal with a set frequency (25-30kHz) and power (50-200W) is first applied through an ultrasonic excitation device, and temperature response data is collected simultaneously. Then, the ultrasonic excitation pressure and pressure change rate are calculated and substituted into the model to calculate the temperature change. The calculated temperature change is compared with the actual collected temperature response data. If the deviation exceeds 0.3℃, the thermoelastic conversion coefficient and dynamic thermoelastic coefficient are adjusted until the deviation is less than 0.3℃. By refining the medium and structural parameters, a precise mapping relationship between ultrasonic excitation and temperature response is established, enhancing the ability to capture the weak thermoelastic effect caused by micro-leakage. This avoids the temperature change calculation deviation caused by the mismatch between model parameters and actual working conditions, which would affect the correlation analysis between leakage and thermoelastic effect.

[0038] Preferably, the data processing efficiency model expression of the Modin distributed sensor data processing platform is: Where E represents the platform's data processing efficiency, N represents the total number of temperature data entries collected, D represents the number of bytes per data entry, and k represents the number of parallel processing nodes on the platform. Let i be the data reception time of the i-th node. Let i be the data processing time for the i-th node. The node coordination coefficient is used to optimize the platform's processing speed for temperature data associated with wellbore micro-leakage, ensuring timely data processing to support subsequent identification steps.

[0039] Specifically, the parameter settings and optimization of the data processing efficiency model of the Modin distributed sensor data processing platform are implemented by first determining the calculation method of each parameter. The total number of data entries is calculated based on the acquisition duration, the number of sensors, and the acquisition frequency. When the acquisition duration is 72 hours, the number of sensors is 6000 (3000-meter wellbore, 0.5-meter spacing), and the acquisition frequency is 10 seconds / time, the total number of entries is 72 × 360 × 6000 = 155,520,000. The number of bytes per data entry is determined based on the accuracy of the temperature data; 4 bytes are used for storage when the accuracy is ±0.1℃, so the number of bytes per data entry is 4. The number of parallel processing nodes is dynamically adjusted according to the total amount of data. 10 nodes are activated when the data volume is 100GB, and the number of nodes increases by 50GB for every additional 50GB. The number of nodes can be increased by 2, with a maximum of 20. The node data reception time is obtained through real-time monitoring on the platform. Under normal circumstances, the time for each node to receive 1 hour of data (approximately 216,000 records) should be controlled within 30 seconds. If it exceeds this time, the data transmission link should be checked and the transmission rate adjusted (100-120Mbps). The node data processing time is also monitored in real time. The time for each node to process 1 hour of data should be controlled within 1 minute. If it exceeds this time, the node memory resources should be increased (16-24GB). The node coordination coefficient is set according to the smoothness of data transmission and task scheduling between nodes. When the data transmission delay between nodes is less than 1 second, it is set to 0.95; when the delay is 1-2 seconds, it is set to 0.9; when the delay exceeds 2 seconds, it is set to 0.8, and the scheduling strategy is optimized. During model implementation, the platform's data processing efficiency is calculated every hour. When the efficiency is below 80 (unit: MB / s), efficiency is improved by increasing the number of nodes, adjusting the transmission rate, or optimizing memory allocation to ensure that the efficiency is maintained between 80-100. By quantifying the parameters of each stage of platform data processing, an efficiency evaluation and optimization mechanism is established to ensure the timeliness of processing massive amounts of temperature data and avoid data backlog due to insufficient processing efficiency, which would affect the real-time performance of subsequent algorithm calculations.

[0040] Preferably, the model expression for determining the location of micro-leakage in the wellbore is: Where L is the depth of the micro-leakage location, Here, m represents the initial reference depth, and m is the number of depth points used in the calculation. The temperature gradient correlation value at depth xj at time tj is given. The temperature gradient is used as a baseline value when there is no leakage. This model can accurately locate the specific depth where micro-leakage occurs in the wellbore.

[0041] Specifically, the parameter selection and implementation steps for the wellbore micro-leakage location determination model are as follows: First, clarify the method for determining each parameter of the model. The initial reference depth is selected from the middle depth of the historically high-leakage area of ​​the wellbore. If no historical data is available, any depth in the middle of the wellbore (1500-2000 meters), such as 1800 meters, is selected. The number of depth points involved in the calculation is determined based on the range of the suspected leak area. If the suspected area is 50 meters (100 sensor spacing), then depth points within that area and 20 meters above and below it are selected, for a total of 140 depth points. Each depth point corresponds to multiple correlation values ​​at three consecutive times, and the average value is taken as the correlation value for that depth point. The temperature gradient correlation benchmark value when there is no leak is calculated based on historical data from wellbores at the same depth and during the same period without leaks. If no historical data is available, the temperature data from the initial stage (first 24 hours) of current wellbore data collection without anomalies is selected. The benchmark value for each depth point is the average value of the correlation values ​​for that depth point over the previous 24 hours, with the deviation controlled within ±0.1. During model implementation, the correlation values ​​and baseline values ​​of each depth point are first collected, and the difference between the correlation value and the baseline value of each depth point is calculated. Then, the depth of the leak location is calculated through weighted summation. During the calculation, if the difference of a certain depth point exceeds the threshold (0.3), the weight of that depth point is increased (adjusted from 1.0 to 1.2). If the difference is less than 0.1, the weight is decreased (adjusted from 1.0 to 0.8) to highlight the impact of abnormal depth points on the location results. After the preliminary leak location is calculated, it is compared with the suspected area marked by the ultrasonic excitation thermoelastic effect identification model. If the deviation between the two exceeds 0.5 meters, the number of depth points participating in the calculation is reselected (increased by 20), the initial reference depth is adjusted (±50 meters), and the calculation is repeated until the deviation meets the requirements. Through weighted calculation and multiple rounds of verification, the accuracy of micro-leak location is improved, avoiding excessive positioning deviation due to individual abnormal data or improper parameter selection. This provides accurate location basis for well maintenance and reduces the blindness of maintenance operations.

[0042] Preferably, step S3 includes the following sub-steps: S31, retrieve the temperature data filtered in S2 from the Modin distributed sensor data processing platform, reorder the data according to the time series and depth dimension to form a structured temperature dataset, ensuring the continuity of the data in the spatiotemporal dimension; S32, input the structured temperature dataset into the initialization module of the temperature field dynamic evolution prediction algorithm, set the initial parameters of the algorithm, including the temperature diffusion coefficient, the initial temperature reference value and the parameters related to the wellbore environment, and establish the initial conditions for algorithm calculation; S33, based on the initial conditions, run the core calculation module of the temperature field dynamic evolution prediction algorithm, use the finite difference method to solve the spatiotemporal evolution equation of the temperature field, and obtain the intermediate temperature calculation values ​​at different depth positions at each time step; S34, perform spatiotemporal consistency verification on the intermediate temperature calculation values, compare the temperature calculation values ​​at the same depth position at adjacent time steps and adjacent depth positions at the same time step, correct unreasonable intermediate calculation values, and finally output the predicted temperature values ​​at different depth positions at different times.

[0043] Specifically, step S3 includes four sub-steps. S31 retrieves the temperature data filtered in S2 from the Modin platform and performs a dual sorting based on both time series (one data point every 10 seconds) and depth dimension (one collection point every 0.3-0.5 meters). The bubble sort algorithm is used during the sorting process to ensure the spatiotemporal continuity of the data. If data is missing (the missing rate must be controlled within 0.5%), it is supplemented through linear interpolation of adjacent data, ultimately forming a structured dataset in CSV format for easy reading by subsequent algorithms. In S32, the structured dataset is imported into the algorithm initialization module. When setting initial parameters, the thermal diffusivity coefficient is determined based on the casing material and formation medium. For example, for P110 steel casing paired with sandstone formation, it is set to 1.1 × 10⁻⁻⁻⁶. 5The initial temperature baseline is the average temperature of the 24 hours prior to the data selection in S2, with a deviation of less than 0.2℃. The maximum number of iterations is set to 100 to avoid infinite iterations. When the S33 core calculation module runs, the finite difference method is used to solve the spatiotemporal evolution equation of the temperature field. The spatial step size is consistent with the sensor deployment spacing (0.3-0.5 meters), and the time step size is the same as the data acquisition interval (10 seconds). After each iteration, the error between the current iteration result and the previous result is calculated using the mean square error method. When S34 performs spatiotemporal consistency verification, it compares the calculated temperature values ​​at the same depth location with adjacent time steps. If the difference exceeds 0.5℃, it is deemed unreasonable and needs to be recalculated. If the difference exceeds 0.3℃ with the calculated temperature values ​​at adjacent depth locations with the same time step, the thermal diffusivity coefficient is adjusted until all calculated values ​​meet the verification standard. The final output predicted temperature value needs to be stored in the Modin platform's temporary database in JSON format for easy S4 access. By refining the step-by-step parameters and operations, the accuracy of temperature field prediction is ensured, laying the foundation for subsequent deviation analysis and avoiding inaccurate prediction results due to disordered data sorting, improper parameter settings, or calculation errors.

[0044] Preferably, step S4 includes the following sub-steps: S41, obtaining predicted temperature values ​​at different depths at different times from the temperature field dynamic evolution prediction algorithm, and simultaneously retrieving actual temperature data collected at the corresponding times from the distributed temperature sensing device, establishing a one-to-one correspondence between the predicted temperature and the actual temperature; S42, performing difference calculations on each set of corresponding data to obtain temperature deviation values ​​at different depths at different times, arranging all temperature deviation values ​​in order of time and depth to form a temperature deviation sequence; S43, calling the correlation analysis module of the temperature gradient spatiotemporal correlation detection algorithm to perform gradient calculation in the spatial dimension and trend analysis in the time dimension on the temperature deviation sequence, extracting the rate of change of temperature deviation at different depths and the degree of correlation between temperature deviations at adjacent depths; S44, constructing a temperature gradient change feature vector based on the correlation analysis results, the feature vector including spatial gradient values, time rate of change, and spatiotemporal correlation coefficient parameters, for use as input to the subsequent thermoelastic effect identification model.

[0045] Specifically, step S4 includes four sub-steps. S41 retrieves the predicted temperature values ​​at each depth and time from the output module of the temperature field dynamic evolution prediction algorithm, and simultaneously retrieves the actual temperature data at the corresponding time from the historical database of the distributed temperature sensing device. Data pairing uses a "depth-time" dual-keyword matching method, with a matching accuracy of 100%. If a matching failure occurs (failure rate must be less than 0.1%), the data is retrieved again, and the data timestamp and depth identifier are checked to ensure that each set of data includes the predicted and actual temperature values ​​at the same depth and time. S42 calculates the temperature deviation value by subtracting the predicted temperature value from the actual temperature value, retaining two decimal places. The deviation values ​​are then arranged in order of depth from shallow to deep and time from early to late, forming a temperature deviation sequence. The sequence length is consistent with the total data acquisition time (e.g., if 72 hours of acquisition is used, the sequence includes 25920 data points). The sequence needs to be stored in the algorithm's dedicated cache, and the cache capacity must be sufficient to store at least three complete sequences. When the S43 correlation analysis module is started, the spatial dimension gradient calculation is performed by dividing the difference between adjacent depth deviations by the sensor spacing (0.3-0.5 meters), with the result rounded to three decimal places. The temporal dimension trend analysis is performed by dividing the difference between adjacent time deviations by the time step (10 seconds). Simultaneously, the correlation coefficient between adjacent depths and adjacent time deviations is calculated using the Pearson correlation coefficient method, with the result controlled between -1 and 1. This is used to extract the rate of change and correlation characteristics. When constructing the feature vector in S44, the vector includes three core parameters: spatial gradient value, temporal rate of change, and spatiotemporal correlation coefficient. Each parameter is arranged in depth-time order, forming a multi-dimensional feature vector. The vector dimension is consistent with the number of data collection points (e.g., 6000 collection points in a 3000-meter wellbore, resulting in a vector dimension of 6000). The feature vector needs to be converted into a matrix format recognizable by the algorithm and transmitted to the Modin platform feature database to provide input data for S5. Through standardized step-by-step operations and parameter standards, the accurate extraction of temperature gradient change features is ensured, avoiding subsequent correlation analysis failures due to data pairing errors or feature calculation deviations.

[0046] Preferably, step S5 includes the following sub-steps: S51, applying ultrasonic excitation signals of different frequencies and intensities to multiple preset detection areas in the wellbore using an ultrasonic excitation device, and recording the excitation signal parameters of each detection area, including excitation frequency, excitation intensity, and excitation duration; S52, using a distributed temperature sensing device to synchronously collect temperature response data of each detection area at different times during ultrasonic excitation, ensuring the time synchronization of temperature response data and excitation signals; S53, inputting the collected temperature response data into the ultrasonic excitation thermoelastic effect identification model, and combining it with the temperature gradient change feature vector extracted in S4 to establish a mapping relationship between temperature response and temperature gradient change under the thermoelastic effect; S54, performing a significance test on the mapping relationship, and selecting mapping relationship pairs with strong correlation as the core correlation basis for subsequent wellbore micro-leakage identification.

[0047] Specifically, step S5 includes four sub-steps. In S51, when deploying the ultrasonic excitation device, the device is positioned one-to-one with the distributed temperature sensors along the outer wall of the wellbore, with the spacing consistent with the sensor deployment spacing (0.3-0.5 meters). The excitation frequency adjustment range is set to 20-100kHz, and the power adjustment range is set to 50-200W. Specific parameters are set according to the wellbore diameter, such as 30kHz frequency and 100W power for wellbore with a diameter of 150-200 mm, and 25kHz frequency and 120W power for wellbore with a diameter of 200-300 mm. The excitation duration is uniformly set to 5 minutes. Signal calibration is required before the device is started, and the calibration error must be less than 5%. When collecting temperature response data in S52, the distributed temperature sensing device maintains its original collection frequency (10 seconds / time), with a collection duration covering 2 minutes before excitation, 5 minutes during excitation, and 3 minutes after excitation, for a total of 10 minutes. The collected data must be synchronized with the excitation signal time, and the time synchronization error must be controlled within 1 second. If a synchronization deviation occurs, the data must be collected again and the clock calibration module checked. The collected temperature response data must be transmitted to the Modin platform's temporary storage area in real time, and the storage area's read / write speed must be greater than 100Mbps to avoid data transmission delay. When importing the ultrasonic excitation thermoelastic effect identification model in S53, the temperature response data is first divided according to the excitation region (each excitation device corresponds to one region), and then matched with the temperature gradient change feature vector extracted in S4 according to the region. The matching adopts the region number association method. After the matching is completed, a linear regression algorithm is used to establish the mapping relationship between temperature response and temperature gradient change. The regression coefficients must be calculated using the least squares method, and the goodness of fit R² must be greater than 0.85 to ensure the reliability of the mapping relationship. When performing significance testing on S54, the t-test method is used to verify the significance of the mapping relationship. The significance level is set to 0.05. When the P-value is less than 0.05, the mapping relationship is considered significant and selected as a valid correlation pair. When the P-value is greater than or equal to 0.05, the excitation parameters need to be readjusted (e.g., power increase or decrease of 5-10W) and data needs to be collected again until a significant mapping pair is obtained. Finally, the valid correlation pairs are organized into a correlation table by region and stored in the Modin platform correlation database, providing the core basis for S6 leak identification. By refining the step-by-step parameters and testing standards, the correlation between thermoelastic effect and temperature gradient change is ensured to be reliable, avoiding missed leak identification due to improper excitation parameters, data asynchrony, or insignificant correlation.

[0048] The temperature field dynamic evolution prediction algorithm of this invention is used to construct a spatiotemporal variation model of the wellbore temperature field and calculate the predicted temperature values ​​at each time and depth. The algorithm first retrieves filtered structured temperature data from the Modin platform, sorting it by time (one data point every 10 seconds) and depth (one acquisition point every 0.3-0.5 meters). Missing data (missing rate ≤ 0.5%) is supplemented by linear interpolation of adjacent data. Next, initial parameters are set, with the thermal diffusivity determined based on the casing material and formation medium (e.g., 1.1 × 10⁻ for P110 steel casing paired with sandstone formation). 5 The initial temperature baseline is the average temperature of the previous 24 hours (deviation ≤ 0.2℃), and the maximum number of iterations is set to 100. Then, the finite difference method is used to solve the spatiotemporal evolution equation of the temperature field. The spatial and temporal steps are matched with the sensor spacing and acquisition interval, respectively. The error is calculated using the mean square error method for each iteration. Finally, a spatiotemporal consistency check is performed. If the temperature difference at the same depth in adjacent time steps exceeds 0.5℃ or the temperature difference between adjacent depths in the same time step exceeds 0.3℃, the parameters are recalculated or adjusted. The final predicted temperature value is output and stored in JSON format. This algorithm provides a normal temperature change baseline for subsequent temperature deviation analysis, accurately predicts the temperature patterns at different spatiotemporal locations in the wellbore, avoids misjudgments of leakage characteristics due to the lack of a temperature baseline, lays a data foundation for the entire micro-leakage identification process, and ensures that subsequent steps can accurately capture the weak temperature anomalies caused by micro-leakage.

[0049] The temperature gradient spatiotemporal correlation detection algorithm of this invention is used to analyze temperature deviation and extract leakage-related temperature gradient change features. Predicted temperature values ​​and actual temperature data are retrieved from the algorithm output module and the historical database of the sensing device, respectively. "Depth-time" dual-keyword matching (accuracy must be 100%, failure rate ≤0.1%) is used to ensure that each set of data corresponds to the same spatiotemporal location. The temperature deviation value (actual temperature minus predicted temperature, rounded to two decimal places) is calculated and arranged in depth and time order to form a deviation sequence (e.g., 72 hours of data collection contains 25920 data points), which is stored in a dedicated cache (capacity required to store three complete sequences). Correlation analysis is initiated. In the spatial dimension, the gradient is calculated by dividing the difference in deviation between adjacent depths by the sensor spacing (rounded to three decimal places). In the temporal dimension, the rate of change is calculated by dividing the difference in deviation between adjacent time points by the time step. Simultaneously, the Pearson correlation coefficient method (result range -1 to 1) is used to analyze the degree of spatiotemporal correlation. A multidimensional feature vector containing spatial gradient values, time rate of change, and spatiotemporal correlation coefficients is constructed, converted into matrix format, and transmitted to the Modin platform feature database. This algorithm filters out features with leakage identification value from temperature deviations, distinguishes between normal temperature fluctuations and leakage anomalies, solves the problem of easy misjudgment in traditional single-parameter analysis, and enhances the ability to capture weak leakage features through spatiotemporal correlation, providing accurate feature input for subsequent thermo-elastic effect correlation.

[0050] The ultrasonic-excited thermoelastic effect identification model of this invention is used to establish the correlation between temperature response and temperature gradient change under ultrasonic excitation and enhance the identification of leakage characteristics. The model first deploys an ultrasonic excitation device, corresponding one-to-one with temperature sensors along the outer wall of the wellbore (0.3-0.5 meters apart). Parameters are set according to the wellbore diameter (e.g., 30kHz frequency, 100W power for a 150-200 mm diameter). A calibration signal is applied before excitation (error ≤ 5%), and excitation is continued for 5 minutes. Then, temperature response data is collected synchronously, with the sensors maintaining a collection frequency of 10 seconds per acquisition, covering the periods before excitation (2 minutes), during excitation (5 minutes), and after excitation (3 minutes), ensuring that the data is synchronized with the excitation signal (error ≤ 1 second). The data is transmitted to the Modin platform's temporary storage area (read / write rate > 100Mbps). Then, temperature response data is divided according to the excitation region and associated with the temperature gradient change feature vector through region numbering. A linear regression algorithm (least squares method to calculate coefficients, goodness of fit R² > 0.85) is used to establish a mapping relationship. Finally, a t-test (significance level 0.05) is used to verify the significance of the association. When the p-value is ≥ 0.05, the excitation parameters are adjusted (power increased / decreased by 5-10W), and data is collected again. The effective association pairs are then compiled into a table and stored in the association database. This model amplifies the temperature characteristics of the leakage area through active ultrasonic excitation, establishing a correlation between the thermoelastic effect and leakage characteristics. This solves the problem of weak and difficult-to-identify micro-leakage characteristics, improves identification sensitivity, avoids missed detections, and provides core association evidence for final leakage determination.

[0051] The Modin distributed sensor data processing platform of this invention is a distributed data processing system for efficiently processing massive amounts of wellbore temperature data and supporting subsequent algorithm calculations. The architecture employs a multi-node parallel approach: First, it receives raw data from distributed temperature sensors (transmission rate 100Mbps). Based on the data volume (100-500GB), it allocates 10-20 parallel processing nodes (each node has 16GB of memory and an 8-core CPU). Then, it partitions and assigns tasks, dividing the data into 1-hour blocks based on time and distributing them evenly across nodes. Nodes remove abnormal data, null values, and duplicate values ​​that exceed the normal temperature range (25-150℃±20℃ for onshore oil and gas wells). Next, a node collaborative control module implements data allocation and task scheduling, monitoring the data reception (1-hour data reception time ≤ 30 seconds) and processing time (1-hour data processing time ≤ 1 minute) of each node in real time. When the transmission delay between nodes exceeds 2 seconds, the scheduling strategy is optimized. Simultaneously, the performance is evaluated using a data processing efficiency model (efficiency = total data bytes / total node time × collaborative coefficient). When the efficiency is below 80MB / s, efficiency is improved by adding nodes, adjusting the transmission rate, or optimizing memory. This platform enables rapid screening, structured processing, and efficient scheduling of massive temperature data, providing timely and effective data support for algorithms. It solves the problems of processing delay and uneven task allocation in traditional single-node architectures, ensures real-time data processing, and avoids data backlog affecting the efficiency of the entire identification process. It is the basic support platform for achieving efficient identification of micro-leakage in wells.

[0052] like Figure 2As shown, a method for identifying wellbore micro-leakage using distributed temperature gradient sensing data is implemented through different units, including: a distributed temperature data acquisition unit, which consists of multiple distributed temperature sensors spaced along the depth direction of the wellbore to collect raw temperature data at different depths and transmit the collected data to a data transmission unit; a data transmission unit, which employs an optical fiber transmission module and a wireless backup transmission module. The optical fiber transmission module is used for high-speed and stable transmission of large amounts of raw temperature data, while the wireless backup transmission module ensures data transmission continuity in case of optical fiber transmission failure, transmitting the data to a Modin distributed data processing unit; and a Modin distributed data processing unit, which includes multiple parallel processing nodes and a node collaborative control module. The parallel processing nodes are used to filter the received raw temperature data, and the node collaborative control module is used to manage communication between different nodes. Data allocation and task scheduling are performed, and the processed data is transmitted to the algorithm processing unit. The algorithm processing unit integrates a temperature field dynamic evolution prediction algorithm module, a temperature gradient spatiotemporal correlation detection algorithm module, and an ultrasonic excitation thermoelastic effect identification model module. Different modules perform calculations and analyses on the processed data, outputting temperature gradient change characteristics and correlation data to the micro-leakage identification unit. The micro-leakage identification unit includes a threshold setting module and a comparison analysis module. The threshold setting module sets the micro-leakage identification threshold based on wellbore parameters, while the comparison analysis module compares the temperature gradient change characteristics with the identification threshold, outputting the micro-leakage judgment result and location information to the result display unit. The result display unit uses a visual interface and a data storage module. The visual interface displays the wellbore micro-leakage status and location information in real time, while the data storage module stores the raw data, calculation data, and identification results for querying and analysis.

[0053] A method for identifying wellbore micro-leakage using distributed temperature gradient sensing data is proposed. On the one hand, by introducing the Modin distributed sensor data processing platform and building a multi-node parallel processing architecture, the massive spatiotemporal temperature data collected by distributed temperature sensing devices can be efficiently divided and task-assigned. This avoids the problems of data processing delay and uneven task allocation under traditional single-node or simple parallel architectures, and quickly completes data filtering and structured processing, providing timely and effective data support for subsequent algorithm calculations, thus meeting the core requirements of real-time downhole monitoring. On the other hand, by integrating temperature field dynamic evolution prediction algorithms, temperature gradient spatiotemporal correlation detection algorithms, and ultrasonic excitation thermoelastic effect identification models, the method no longer relies on a single temperature parameter or simple gradient analysis. Instead, it combines the dynamic evolution law of the temperature field, spatiotemporal correlation characteristics, and thermoelastic effect information under ultrasonic excitation, significantly enhancing the ability to capture the subtle temperature change characteristics caused by micro-leakage, reducing the interference caused by normal temperature fluctuations in the wellbore, and effectively avoiding misjudgment or omission of leaks in traditional methods. At the same time, through multi-parameter collaborative calculation, the precise location of micro-leakage is achieved, solving the problem of insufficient accuracy of traditional positioning models.

[0054] This method forms a complete closed-loop process from data acquisition and processing to algorithm calculation and result judgment. Each step is interconnected, and from the acquisition of raw temperature data to the determination of micro-leakage location, each step has clear technical support, avoiding the problems of process fragmentation and poor connection between links in traditional methods. At the same time, this method fully considers complex working conditions such as deep wells and complex formations in data processing and algorithm design. The Modin distributed platform can adapt to the needs of large-scale data processing, and the multi-algorithm collaboration can cope with temperature field changes under different formation conditions. Compared with the poor adaptability of traditional methods in complex environments, it can stably and reliably realize the identification of wellbore micro-leakage, providing strong protection for the safe operation of the wellbore and reducing formation pollution and resource waste caused by leakage.

[0055] In the description of this invention, it should be noted that, unless otherwise explicitly specified and limited, the terms "set," "install," "connect," "link," and "fix" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can refer to the internal communication between two components. Those skilled in the art will understand the specific meaning of the above terms in this invention based on the specific circumstances.

[0056] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various equivalent changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A method for identifying wellbore micro-leakage using distributed temperature gradient sensing data, characterized in that, Includes the following steps: S1. Collect raw temperature data at different depths in the wellbore over a continuous time series using distributed temperature sensing devices, and transmit the collected raw temperature data to the Modin distributed sensor data processing platform; S2. Based on the Modin distributed sensor data processing platform, perform data partitioning and task allocation on the transmitted raw temperature data, establish a multi-node parallel processing architecture, and perform preliminary screening of the temperature data received by different nodes to remove obviously abnormal data; S3. Call the temperature field dynamic evolution prediction algorithm, use the data filtered in S2 as input, construct a spatiotemporal evolution model of the wellbore temperature field, and calculate the predicted temperature values ​​at different depths at different times; S4. Activate the temperature gradient spatiotemporal correlation detection algorithm, calculate the difference between the predicted temperature value obtained in S3 and the actual temperature data collected at the corresponding time, obtain the temperature deviation sequence, and then perform spatiotemporal correlation analysis on the temperature deviation sequence to extract temperature gradient change characteristics; S5. Apply the ultrasonic excitation thermoelastic effect identification model, apply an ultrasonic excitation signal to a designated area of ​​the wellbore, collect the temperature response data after excitation, and combine it with the temperature gradient change characteristics extracted in S4 to construct the correlation between the thermoelastic effect and the temperature gradient change; S6. Based on the correlation established in S5, set the wellbore micro-leakage identification threshold, compare and analyze the temperature gradient change characteristics at different depths with the identification threshold, and determine whether there is a micro-leakage in the wellbore and the specific location of the micro-leakage.

2. The method for identifying wellbore micro-leakage using distributed temperature gradient sensing data according to claim 1, characterized in that, The expression used in the temperature field dynamic evolution prediction algorithm is: ,in, Let be the predicted temperature value at depth x at time t. The temperature decay coefficient is Let x be the temperature value at depth x at the initial moment. The time decay factor, Where is the thermal diffusivity, for The Laplace operator for the temperature at depth x at time t. The heat source influence coefficient, Let be the heat source intensity at depth x at time t, which is related to the heat generated by fluid leakage during wellbore micro-leakage.

3. The method for identifying wellbore micro-leakage using distributed temperature gradient sensing data according to claim 1, characterized in that, The expression used in the temperature gradient spatiotemporal correlation detection algorithm is: ,in, The associated value of the temperature gradient at depth x at time t. Let x be the spatial temperature gradient at depth x at time t. For depth weighting coefficients, Let be the time-temperature gradient at depth x at time t. For time weighting coefficients, The correlation coefficient, At time t, at depth x, and Time depth Covariance of temperature data For depth intervals, The time interval is used to determine the correlation between temperature data at different times and locations, and the covariance reflects the degree of correlation between temperature data at different times and locations. It is used to determine the propagation characteristics of temperature anomalies caused by micro-leakage in the wellbore.

4. The method for identifying wellbore micro-leakage using distributed temperature gradient sensing data according to claim 1, characterized in that, The expression used in the ultrasonic-excited thermoelastic effect identification model is: ,in, Let be the temperature change at depth x caused by ultrasonic excitation at time t. Thermoelastic conversion coefficient, Let be the ultrasonic excitation pressure at depth x at time t. Let x be the elastic modulus of the wellbore medium at depth x. The dynamic thermoelastic coefficient, Let be the rate of change of ultrasonic excitation pressure at depth x at time t. Let x be the thermal conductivity coefficient of the wellbore structure at depth x, and let x be the temperature change. This is used to correlate the effect of medium changes caused by wellbore microleakage under ultrasonic excitation on the thermoelastic effect.

5. The method for identifying wellbore micro-leakage using distributed temperature gradient sensing data according to claim 1, characterized in that, The data processing efficiency model expression of the Modin distributed sensor data processing platform is as follows: Where E represents the platform's data processing efficiency, N represents the total number of temperature data entries collected, D represents the number of bytes per data entry, and k represents the number of parallel processing nodes on the platform. Let i be the data reception time of the i-th node. The data processing time for the i-th node is [time]. The node coordination coefficient is used to optimize the platform's processing speed for temperature data associated with wellbore micro-leakage, ensuring timely data processing to support subsequent identification steps.

6. The method for identifying wellbore micro-leakage using distributed temperature gradient sensing data according to claim 1, characterized in that, The model expression for determining the location of micro-leakage in the wellbore is as follows: Where L is the depth of the micro-leakage location, Here, m represents the initial reference depth, and m is the number of depth points used in the calculation. The temperature gradient correlation value at depth xj at time tj is given. The temperature gradient is used as a baseline value when there is no leakage. This model can accurately locate the specific depth where micro-leakage occurs in the wellbore.

7. The method for identifying wellbore micro-leakage using distributed temperature gradient sensing data according to claim 1, characterized in that, Step S3 includes the following sub-steps: S31, retrieve the temperature data filtered in S2 from the Modin distributed sensor data processing platform, reorder the data according to the time series and depth dimension to form a structured temperature dataset, ensuring the continuity of the data in the spatiotemporal dimension; S32, input the structured temperature dataset into the initialization module of the temperature field dynamic evolution prediction algorithm, set the initial parameters of the algorithm, including the temperature diffusion coefficient, the initial temperature reference value and the parameters related to the wellbore environment, and establish the initial conditions for algorithm calculation; S33, based on the initial conditions, run the core calculation module of the temperature field dynamic evolution prediction algorithm, use the finite difference method to solve the spatiotemporal evolution equation of the temperature field, and obtain the intermediate temperature calculation values ​​at different depth positions at each time step; S34, perform spatiotemporal consistency verification on the intermediate temperature calculation values, compare the temperature calculation values ​​at the same depth position at adjacent time steps and adjacent depth positions at the same time step, correct unreasonable intermediate calculation values, and finally output the predicted temperature values ​​at different depth positions at different times.

8. The method for identifying wellbore micro-leakage using distributed temperature gradient sensing data according to claim 1, characterized in that, Step S4 includes the following sub-steps: S41. Obtain predicted temperature values ​​at different depths at different times from the temperature field dynamic evolution prediction algorithm, and simultaneously retrieve the actual temperature data collected at the corresponding times from the distributed temperature sensing device to establish a one-to-one correspondence between the predicted temperature and the actual temperature; S42. Perform difference calculation on each set of corresponding data to obtain temperature deviation values ​​at different depths at different times, and arrange all temperature deviation values ​​in order of time and depth to form a temperature deviation sequence; S43. Call the correlation analysis module of the temperature gradient spatiotemporal correlation detection algorithm to perform gradient calculation in the spatial dimension and trend analysis in the time dimension on the temperature deviation sequence, and extract the correlation between the rate of change of temperature deviation at different depths and the degree of correlation between temperature deviation at adjacent depths; S44. Based on the correlation analysis results, construct a temperature gradient change feature vector, which includes spatial gradient values, time rate of change, and spatiotemporal correlation coefficient parameters, for use as input to the subsequent thermoelastic effect identification model.

9. The method for identifying wellbore micro-leakage using distributed temperature gradient sensing data according to claim 1, characterized in that, Step S5 includes the following sub-steps: S51, applying ultrasonic excitation signals of different frequencies and intensities to multiple detection areas preset in the wellbore through an ultrasonic excitation device, and recording the excitation signal parameters of each detection area, including excitation frequency, excitation intensity, and excitation duration. S52. Utilize distributed temperature sensing devices to synchronously collect temperature response data from each detection area at different times during ultrasonic excitation, ensuring the time synchronization of temperature response data with the excitation signal; S53. Input the collected temperature response data into the ultrasonic excitation thermoelastic effect identification model, and combine it with the temperature gradient change feature vector extracted in S4 to establish a mapping relationship between temperature response and temperature gradient change under the thermoelastic effect; S54. Perform a significance test on the mapping relationship, and select mapping relationship pairs with strong correlation as the core correlation basis for subsequent wellbore micro-leakage identification.

10. A method for identifying wellbore micro-leakage using distributed temperature gradient sensing data according to any one of claims 1-9, characterized in that, This method is implemented through different units, including: a distributed temperature data acquisition unit, which consists of multiple distributed temperature sensors spaced along the wellbore depth to collect raw temperature data at different depths within the wellbore and transmit the collected data to a data transmission unit; a data transmission unit, which employs an optical fiber transmission module and a wireless backup transmission module. The optical fiber transmission module is used for high-speed and stable transmission of large amounts of raw temperature data, while the wireless backup transmission module ensures data transmission continuity in case of optical fiber transmission failure, transmitting the data to a Modin distributed data processing unit; and a Modin distributed data processing unit, which includes multiple parallel processing nodes and a node collaborative control module. The parallel processing nodes are used to filter and process the received raw temperature data, while the node collaborative control module is used for data allocation and task scheduling among different nodes. The data is transmitted to the algorithm processing unit; the algorithm processing unit integrates a temperature field dynamic evolution prediction algorithm module, a temperature gradient spatiotemporal correlation detection algorithm module, and an ultrasonic excitation thermoelastic effect identification model module. Different modules perform calculations and analyses on the processed data, outputting temperature gradient change characteristics and correlation data to the micro-leakage identification unit; the micro-leakage identification unit includes a threshold setting module and a comparison analysis module. The threshold setting module is used to set the micro-leakage identification threshold according to the wellbore parameters, and the comparison analysis module is used to compare the temperature gradient change characteristics with the identification threshold, outputting the micro-leakage judgment result and location information to the result display unit; the result display unit uses a visual interface and a data storage module. The visual interface is used to display the wellbore micro-leakage status and location information in real time, and the data storage module is used to store the raw data, calculation data, and identification results for querying and analysis.

Images