A multi-parameter integrated casing damage detection method and system

Abstract

The application discloses a kind of multi-parameter integrated casing damage detection method and system, it is related to casing detection technical field.The method is by downhole detection system synchronous acquisition vision, deformation, residual magnetism eddy current and positioning data, after pre-processing and multiplex transmission to ground;Ground system completes data restoration, spatial position alignment and feature level fusion, intelligently identifies defect and generates multidimensional defect distribution atlas;Again according to defect comprehensive density and severity grade self-adaptive control downhole instrument speed, finally output integrated detection report.Downhole detection system uses integrated downhole instrument string, integrates multiple detection modules and FPGA+ARM preprocessing unit, cooperates with ground closed-loop control system.The application realizes multi-parameter synchronous detection, multi-source data accurate fusion, defect intelligent identification and adaptive speed regulation, greatly improves detection precision and efficiency, reduces operation cost, and is suitable for integrated efficient detection of oil and gas well casing damage.

Description

Technical Field

[0001] This invention relates to the field of casing inspection technology, specifically to a multi-parameter integrated casing damage detection method and system. Background Technology

[0002] In fields such as oil and gas field development and groundwater resource utilization, well casing is a core barrier to ensure the safety of downhole operations and isolate formation fluids. Therefore, efficient and high-precision detection of casing damage is a key link in the whole life cycle management of well integrity. Current casing damage detection methods mostly employ single-parameter detection techniques. Visual inspection can only identify surface defects on the inner wall of the casing and cannot detect hidden metal losses inside the casing or in the wall thickness direction. While residual magnetic eddy current and ultrasonic testing can detect metal damage, they cannot visually present surface defects or accurately quantify casing deformation. All single detection methods have blind spots and low accuracy in identifying complex casing defects. Multiple well runs are required to complete the detection of different dimensions of parameters, resulting in long operation cycles, high construction costs, and significant safety risks during downhole operations. Existing multi-parameter combined detection schemes still suffer from insufficient spatial alignment accuracy of multi-source detection data, and data misalignment due to module arrangement and speed fluctuations, leading to misjudgment and location deviations. Furthermore, multi-source data are often simply superimposed at the result level, failing to achieve deep feature-level fusion. This results in insufficient overall confidence in defect identification, and the detection process often adopts a fixed constant speed operation mode, which cannot balance the detection accuracy of defect-dense sections with the operational efficiency of defect-free sections, making it difficult to meet the needs of large-scale and efficient casing inspection operations in the field. Summary of the Invention

[0003] The purpose of this invention is to provide a multi-parameter integrated sleeve damage detection method and system to solve the problems mentioned in the background art.

[0004] This invention is achieved through the following technical solution: A multi-parameter integrated casing damage detection method includes a downhole detection system and a surface control and processing system: S1. Synchronous data acquisition: The detection system acquires visual inspection data of the inner wall of the casing, deformation inspection data of the casing, residual magnetic eddy current inspection data of the casing, and positioning inspection data of the instrument as multiple raw data. S2. Data Preprocessing and Transmission: Preprocess and reuse multiple raw data streams, and upload the processed data to the ground control and processing system in real time via logging armored cable; S3. Ground data restoration and fusion analysis: The processed data is demultiplexed and signal restored to obtain multi-channel detection data. The multi-channel detection data is then sequentially subjected to spatial location alignment, feature-level fusion, and intelligent defect identification to generate a multi-dimensional defect distribution map. S4. Adaptive Motion Control: Based on a multi-dimensional defect distribution map, the comprehensive defect density of the current detection area is calculated through a preset sliding window. Motion control commands are generated in combination with the severity of the defects. The ground control system adjusts the running speed of the downhole detection system in real time through a cable winch according to the motion control commands. S5. Inspection Results Output: After the inspection of the entire pipe section is completed, the ground control and processing system generates an integrated inspection report that includes defect visualization information, quantitative parameters, location coordinates, and severity level.

[0005] Furthermore, The spatial alignment described in S3 specifically refers to: Based on the visual inspection module, deformation inspection module, and residual magnetic eddy current detection module of the detection system, the fixed axial spacing within the multi-parameter detection section is measured. The axial movement speed of the downhole detection system is acquired in real time by the ground control system. Calculate the time difference between each detection module passing the same defect location on the sleeve. : The ground control and processing system uses the calculated time difference By combining the sampling rates of each detection module, the time axis offset alignment is performed on the multi-channel detection data, which includes detection data from different detection modules, so that the multi-channel detection data corresponds to the same physical position of the sleeve.

[0006] Furthermore, In S3, feature-level fusion is achieved using a strategy that combines feature vector concatenation with weighted interactive fusion, specifically: Extract texture features and grayscale distribution features from visual inspection data, deformation features and deformation gradient features from deformation inspection data, and signal amplitude and phase features from remanent eddy current detection data; Multi-dimensional feature vectors at the same spatial location are concatenated end to end to form a high-dimensional joint feature vector; then, based on the detection sensitivity of each detection module corresponding to different defect types, learnable fusion weights are assigned to complete weighted interactive fusion; finally, the fused feature vector, the comprehensive confidence of the defect, and the overall defect judgment result are output.

[0007] Furthermore, The specific steps for generating a multi-dimensional defect distribution map as described in S3 are: Based on the positioning and detection data collected by the inertial navigation unit in the positioning module of the detection system, a spatial coordinate system for the detection path is established. The multi-path detection data after spatial alignment and the fused feature values ​​obtained by feature-level fusion are mapped to the corresponding coordinate points in the spatial coordinate system. The unsampled positions between sampling intervals are interpolated and made continuous. Finally, the geometric contour of the detection path is used as the base layer, and feature layers of various dimensions and the fused defect comprehensive layer are superimposed to form a multi-dimensional defect distribution map.

[0008] Furthermore, In S4, the comprehensive defect density of the current detection area is calculated using a preset sliding window as follows: Calculate the first in the sliding window Equivalent area of ​​class defects Its calculation formula is In the formula, For the first in the sliding window The number of class defects, For the first in the sliding window The actual area of ​​each defect, and the equivalent area of ​​linear defects such as cracks. For the first The severity coefficient of a defect is positively correlated with the severity level of the defect; Calculate the first Single-class defect density of class defects In the calculation formula, The area covered by the sliding window. Let be the axial length of the sliding window. The circumferential circumference of the sleeve to be tested, and the density of single-type defects. Mapping to the corresponding severity level density interval to determine the first The severity level of the defect is determined, and the maximum severity level within the sliding window is obtained accordingly. Calculate the overall defect density within the sliding window In the formula, This represents the total number of defect types. For the first The weight coefficients of the class defects, and satisfying .

[0009] Furthermore, The weighting coefficients are also dynamically corrected based on the sensor health coefficients, and their calculation formula is as follows: In the formula, For the first Weighting coefficients for class defects For the preset first Static base weights of class defects For the first The sensor health coefficient of the corresponding defect detection module, with a value ranging from 0 to 1, is provided in real time by the signal self-diagnosis data of the corresponding detection module; after correction, for all Normalization is performed.

[0010] Furthermore, The specific operating speed of the real-time control downhole monitoring system described in S4 is as follows: When the defect comprehensive density Below the preset low threshold At that time, the ground control system controls the downhole detection system to operate at a normal speed via a cable winch; When the defect comprehensive density At the preset low threshold With preset high threshold Between, and within the sliding window, the maximum severity level of the defects is light or moderate, and the surface control system controls the downhole detection system to operate at a normal speed via a cable winch; When the defect comprehensive density At the preset low threshold With preset high threshold Between, and the maximum severity level within the sliding window is severe, or the overall defect density Higher than or equal to the preset high threshold At that time, the ground control system immediately switches the downhole detection system to low-speed operation via a cable winch.

[0011] Furthermore, The integrated inspection report described in S5 includes defect type determination results, comprehensive defect severity level, defect axial and circumferential positioning coordinates, deformation parameters, metal loss assessment data, defect video screenshots, and multi-dimensional defect distribution maps.

[0012] Furthermore, A multi-parameter integrated casing damage detection system is used to implement the multi-parameter integrated casing damage detection method described above, including a downhole detection system and a surface control and processing system. The downhole detection system is an integrated instrument string structure, consisting of a cable adapter, a data preprocessing module, a multi-parameter detection sub, and a centering support from top to bottom. The multi-parameter detection sub integrates a visual inspection module, a deformation detection module, a residual magnetic eddy current detection module, a positioning module, and a storage module. The positioning module incorporates a photoelectric coded odometer and an inertial navigation unit. The data preprocessing module uses an FPGA+ARM architecture, incorporating both an FPGA and an ARM unit. The visual inspection module, deformation detection module, residual magnetic eddy current detection module, and positioning module all possess signal self-diagnostic capabilities. The outputs of the visual inspection module, deformation detection module, residual magnetic eddy current detection module, and positioning module are all electrically connected to the inputs of the storage module and data preprocessing module, respectively; the data preprocessing module is bidirectionally connected to the ground control and processing system via a logging armored cable. The ground control and processing system includes a cable winch, a ground power supply system, a ground control system, a signal processing and data acquisition system, and an industrial control computer. The input end of the signal processing and data acquisition system is connected to the logging armored cable, and the output end is connected to the industrial control computer. The output end of the industrial control computer is connected to the ground control system. The ground control system is electrically connected to the cable winch and is used to regulate the operating speed of the downhole detection system.

[0013] The beneficial effects of this invention are as follows: 1. This invention solves the industry pain point of asynchronous sampling time and spatial misalignment of the same defect caused by the axial arrangement of multiple detection modules by calculating the time difference based on the fixed axial spacing of the detection modules and the instrument running speed, and completing the spatial position calibration by combining the sampling rate to complete the time axis offset alignment. It provides a reliable spatial benchmark for subsequent feature fusion and accurate defect positioning, and effectively avoids the problem of defect misjudgment and positioning deviation caused by data misalignment.

[0014] 2. This invention employs a multi-source data feature-level fusion strategy that combines feature vector splicing with weighted interactive fusion. Learnable fusion weights are assigned to the detection sensitivity of corresponding detection modules for different defect types. This achieves a complete representation of all dimensions of casing defects through multi-dimensional feature splicing, while weighted interaction amplifies the feature contribution of high-sensitivity detection modules and suppresses noise interference from low-sensitivity modules. Compared to single detection methods, this significantly improves the accuracy of identifying complex casing defects such as corrosion pits, microcracks, and casing deformation, greatly reduces the defect false negative rate, and optimizes the data processing volume, fully meeting the performance requirements of real-time downhole detection processing.

[0015] 3. This invention uses a defect comprehensive density quantification system to standardize and quantify different types and severity of casing defects, allowing for unified comparison and comprehensive evaluation of dissimilar defects such as linear cracks and planar corrosion pits.

[0016] 4. This invention utilizes an adaptive motion closed-loop control logic based on the comprehensive defect density and defect severity level. It calculates the damage state of the current detection area in real time through a sliding window, automatically matching the operating speed of the downhole detection system. In sections of the casing without defects or with low defect density, it operates at a conventional speed to ensure detection efficiency. In sections with severe defects or high defect density, it automatically switches to low-speed operation to increase sampling density. This avoids the problem of insufficient detection accuracy caused by insufficient sampling in defect-dense sections in traditional constant-speed detection, and also solves the problem of low operational efficiency caused by low-speed detection throughout the process. Furthermore, the fully automated closed-loop speed control requires no manual intervention, significantly reducing the workload of on-site operators and improving the automation and intelligence level of the detection operation. Attached Figure Description

[0017] Figure 1 This is a schematic diagram of the system workflow of the present invention; Figure 2 This is a schematic diagram of the logical flow of the method of the present invention; Figure 3 This is a schematic diagram of the system of the present invention. Detailed Implementation

[0018] The present invention will be further described in detail below with reference to the embodiments and accompanying drawings, but the embodiments of the present invention are not limited thereto.

[0019] See the example. Figures 1 to 3 : A multi-parameter integrated casing damage detection method includes a downhole detection system and a surface control and processing system: S1. Synchronous data acquisition: The detection system acquires visual inspection data of the inner wall of the casing, deformation inspection data of the casing, residual magnetic eddy current inspection data of the casing, and positioning inspection data of the instrument as multiple raw data. S2. Data Preprocessing and Transmission: Preprocess and reuse multiple raw data streams, and upload the processed data to the ground control and processing system in real time via logging armored cable; S3. Ground data restoration and fusion analysis: The processed data is demultiplexed and signal restored to obtain multi-channel detection data. The multi-channel detection data is then sequentially subjected to spatial location alignment, feature-level fusion, and intelligent defect identification to generate a multi-dimensional defect distribution map. S4. Adaptive Motion Control: Based on a multi-dimensional defect distribution map, the comprehensive defect density of the current detection area is calculated through a preset sliding window. Motion control commands are generated in combination with the severity of the defects. The ground control system adjusts the running speed of the downhole detection system in real time through a cable winch according to the motion control commands. S5. Inspection Results Output: After the inspection of the entire pipe section is completed, the ground control and processing system generates an integrated inspection report that includes defect visualization information, quantitative parameters, location coordinates, and severity level.

[0020] Furthermore, The spatial alignment described in S3 specifically refers to: Based on the visual inspection module, deformation inspection module, and residual magnetic eddy current detection module of the detection system, the fixed axial spacing within the multi-parameter detection section is measured. The axial movement speed of the downhole detection system is acquired in real time by the ground control system. Calculate the time difference between each detection module passing the same defect location on the sleeve. : The ground control and processing system uses the calculated time difference By combining the sampling rates of each detection module, the time axis offset alignment is performed on the multi-channel detection data, which includes detection data from different detection modules, so that the multi-channel detection data corresponds to the same physical position of the sleeve.

[0021] By combining sampling rate with time axis offset alignment, the industry pain point of asynchronous sampling time and spatial misalignment of the same defect caused by the axial arrangement of multiple detection modules is solved. There is no need to design additional hardware synchronization trigger circuits for multiple modules in the well, which greatly reduces the hardware complexity and failure risk of the instrument in the high temperature and high pressure environment in the well. At the same time, it can adapt to speed fluctuations during winch speed adjustment, ensuring that the spatial position of multi-source data corresponds one-to-one. This provides an accurate spatial benchmark for subsequent feature fusion and defect localization, fundamentally avoiding the problems of defect misjudgment and positioning deviation caused by data misalignment. In practical implementation, the visual inspection module, deformation inspection module, and residual magnetic eddy current inspection module within the multi-parameter inspection section are arranged sequentially from top to bottom along the instrument axis. The fixed axial distance d1 between the visual inspection module and the deformation inspection module is 0.3m, and the fixed axial distance d2 between the deformation inspection module and the residual magnetic eddy current inspection module is 0.2m. When the axial movement velocity v of the downhole inspection system, acquired in real time by the ground control system, is 0.6m / s, the time difference Δt1 between the visual inspection module and the deformation inspection module passing the same defect location is calculated to be 0.3 / 0.6 = 0.5s, and the time difference Δt1 between the deformation inspection module and the residual magnetic eddy current inspection module is calculated to be 0.5s. t2 = 0.2 / 0.6 ≈ 0.333s; If the sampling rate of each detection module is 100Hz, the ground control and processing system will shift the detection data of the deformation detection module forward by 50 sampling points along the time axis and the detection data of the residual magnetic eddy current detection module forward by 83 sampling points along the time axis to complete the time axis offset alignment of the three detection data. This ensures that each sampling point of the three detection data corresponds to the same axial physical position of the casing. This implementation method can control the spatial alignment error of multi-source data to within 1mm, while greatly simplifying the circuit design of downhole instruments and improving the operational stability of the instruments in complex downhole environments.

[0022] Furthermore, In S3, feature-level fusion is achieved using a strategy that combines feature vector concatenation with weighted interactive fusion, specifically: Extract texture features and grayscale distribution features from visual inspection data, deformation features and deformation gradient features from deformation inspection data, and signal amplitude and phase features from remanent eddy current detection data; Multi-dimensional feature vectors at the same spatial location are concatenated end to end to form a high-dimensional joint feature vector; then, based on the detection sensitivity of each detection module corresponding to different defect types, learnable fusion weights are assigned to complete weighted interactive fusion; finally, the fused feature vector, the comprehensive confidence of the defect, and the overall defect judgment result are output.

[0023] To address corrosion pit defects on the inner wall of casing, a convolutional neural network is first used to extract texture features and grayscale distribution features from visual inspection data, deformation features and deformation gradient features from deformation inspection data, and signal amplitude and phase features from remanent magnetic eddy current inspection data. The feature vectors of these six dimensions at the same spatial location are then concatenated to form a 256-dimensional high-dimensional joint feature vector. Based on a pre-trained defect recognition model, the fusion weights for the visual inspection module features (0.5), remanent magnetic eddy current inspection module features (0.35), and deformation inspection module features (0.15) are assigned according to the corrosion pit defect type, performing weighted interactive fusion. The final output includes the fused feature vector, the comprehensive confidence score of the corrosion pit defect, and the defect determination result. This implementation method amplifies the feature contribution of detection modules with higher sensitivity to target defect types and suppresses noise interference from low-sensitivity modules. Compared to single detection methods, it improves the accuracy of identifying composite defects such as corrosion pits, microcracks, and casing deformation, reduces the false negative rate, and decreases the data processing volume, fully meeting the real-time processing requirements of downhole inspection.

[0024] Furthermore, The specific steps for generating a multi-dimensional defect distribution map as described in S3 are: Based on the positioning and detection data collected by the inertial navigation unit in the positioning module of the detection system, a spatial coordinate system for the detection path is established. The multi-path detection data after spatial alignment and the fused feature values ​​obtained by feature-level fusion are mapped to the corresponding coordinate points in the spatial coordinate system. The unsampled positions between sampling intervals are interpolated and made continuous. Finally, the geometric contour of the detection path is used as the base layer, and feature layers of various dimensions and the fused defect comprehensive layer are superimposed to form a multi-dimensional defect distribution map.

[0025] By establishing a spatial coordinate system using inertial navigation data, data mapping, interpolation continuity, and multi-layer overlay, absolute spatial positioning of the entire casing inspection data was achieved. This solved the problem of relative mileage positioning deviation caused by cable stretching and winch slippage in traditional well logging. The interpolation continuity processing filled the data gaps between discrete sampling points, realizing continuous defect characterization of the casing in the entire circumference and axis without detection blind spots. The multi-layer overlay method can not only intuitively display single-dimensional detection features, but also display the integrated defect information after fusion, realizing multi-dimensional visualization of defects and providing intuitive and comprehensive data support for subsequent casing repair and safety assessment. In practical implementation, the three-axis acceleration and three-axis angular velocity positioning and detection data collected by the inertial navigation unit in the positioning module are used as the basis. Combined with the mileage data of the photoelectric odometer, a three-dimensional spatial coordinate system for the casing detection path is established. The axis of the coordinate system is the depth direction of the casing, the circumferential direction is the circumferential direction of the casing, and the radial direction is the wall thickness direction of the casing. The visual inspection grayscale data, deformation detection deformation data, residual magnetic eddy current metal loss data, and the defect comprehensive confidence fusion feature value obtained by feature-level fusion are mapped one by one to the corresponding depth and circumferential coordinate points in the spatial coordinate system. For the unsampled positions between sampling intervals, a cubic spline interpolation algorithm is used for continuous processing to achieve 0 on the casing axis. Continuous data coverage with 1mm intervals and 1° circumferential intervals; finally, using the cylindrical geometric contour of the casing inspection path as the base layer, a visual texture feature layer, a deformation distribution layer, a metal loss distribution layer, and a fused comprehensive defect distribution layer are sequentially superimposed to form a multi-dimensional defect distribution map that can be freely switched and zoomed in for viewing; this implementation method controls the casing defect positioning error to within 0.5%, achieves defect characterization without blind spots along the entire casing section, and the generated multi-dimensional map can intuitively and comprehensively display all-dimensional information such as the location, size, type, and severity of defects, greatly improving the readability and practicality of the inspection results, and providing accurate and intuitive visualization basis for casing integrity assessment and repair plan formulation.

[0026] Furthermore, In S4, the comprehensive defect density of the current detection area is calculated using a preset sliding window as follows: Calculate the first in the sliding window Equivalent area of ​​class defects Its calculation formula is In the formula, For the first in the sliding window The number of class defects, For the first in the sliding window The actual area of ​​each defect, and the equivalent area of ​​linear defects such as cracks. For the first The severity coefficient of a defect is positively correlated with the severity level of the defect; Calculate the first Single-class defect density of class defects In the formula, The area covered by the sliding window. Let be the axial length of the sliding window. The circumferential circumference of the sleeve to be tested, and the density of single-type defects. Mapping to the corresponding severity level density interval to determine the first The severity level of the defect is determined, and the maximum severity level within the sliding window is obtained accordingly. Calculate the overall defect density within the sliding window In the formula, This represents the total number of defect types. For the first The weight coefficients of the class defects, and satisfying .

[0027] By using equivalent area and severity coefficients, defects of different types and severity are standardized and quantified, solving the problem of inconsistent comparison and comprehensive evaluation of different defect types such as linear cracks and planar corrosion pits. The mapping between single-type defect density and severity level can accurately identify severe single-point defects within the sliding window, avoiding the problem of high-risk defects being masked by the averaging of comprehensive density. The weighted summation of comprehensive density calculation can allocate weights according to the degree of impact of different defect types on casing safety, realizing a scientific and accurate comprehensive quantitative evaluation of casing damage status. This provides a reliable quantitative basis for subsequent adaptive speed regulation and avoids false triggering and missed triggering of speed regulation logic. In practical implementation, the axial length L of the sliding window is preset to be 1m, the circumferential circumference W of the sleeve to be inspected is 0.5m, the coverage area L×W of the sliding window is 0.5㎡, and the total number of defect types m is 3, namely corrosion defects, crack defects, and deformation defects. For the corrosion defects within the current sliding window, the number of defects n1 is counted as 5. Two mild defects have actual areas of 50mm² and 60mm², with a severity coefficient α of 1 for each; two moderate defects have actual areas of 100mm² and 120mm², with a severity coefficient α of 3 for each; and one severe defect has an actual area of ​​200mm², with a severity coefficient α of 5. The equivalent area Seq,1 of the corrosion defects is calculated as (50×1)+(60×1)+(100×3)+(120×3)+(200×5)=2070mm²=0.00207㎡. Furthermore, the single-type defect density of the corrosion defects is calculated. =0.00207 / 0.5=0.414%, which is mapped to the preset severity level density range. 0.3%~0.6% corresponds to a moderate severity level, so the severity level of corrosion defects is determined to be moderate. The single-type defect density of crack defects is calculated in the same way. The density of single-type defects is 0.2%, corresponding to a mild level, representing the single-type defect density of deformation defects. The severity level is set at 0.8%, corresponding to a severe level. Therefore, the maximum severity level within the sliding window is determined to be severe. The weighting coefficients w1, w2, and w3 for corrosion, crack, and deformation defects are preset to 0.4, 0.35, and 0.25 respectively, and the sum of these weights is 1. The overall defect density within the sliding window is then calculated. =(0.4×0.414%)+(0.35×0.2%)+(0.25×0.8%)=0.4356%; This implementation method enables standardized quantitative assessment of defects of different types and severity, accurately capturing high-risk and severe defects within the sliding window, providing a reliable quantitative basis for subsequent adaptive motion control, and effectively avoiding misjudgments in speed regulation logic.

[0028] Furthermore, The weighting coefficients are also dynamically corrected based on the sensor health coefficients, and their calculation formula is as follows: In the formula, For the first Weighting coefficients for class defects For the preset first Static base weights of class defects For the first The sensor health coefficient of the corresponding defect detection module, with a value ranging from 0 to 1, is provided in real time by the signal self-diagnosis data of the corresponding detection module; after correction, for all Normalization is performed.

[0029] During downhole inspection, when a sensor in a certain detection module experiences performance degradation or signal anomalies, the fixed weights can lead to distorted comprehensive density calculations and incorrect defect assessments. By real-time correction of weights based on signal self-diagnosis data, the weight ratio of modules with abnormal signals can be automatically reduced, while the weight ratio of modules with normal signals can be increased. This ensures the reliability of the comprehensive density calculation results, significantly improving the system's fault tolerance and adaptability to complex downhole environments. The corrected normalization process ensures a unified standard for comprehensive density calculations across the entire well section, making the damage states of different well sections comparable laterally and preventing changes in the calculation benchmark. In practical implementation, the static base weight w_static1 for the visual inspection module corresponding to corrosion defects is preset to 0.4, the static base weight w_static2 for the remanent magnetic eddy current detection module corresponding to crack defects is preset to 0.35, and the static base weight w_static3 for the deformation detection module corresponding to deformation defects is preset to 0.25. During downhole inspection, each detection module performs real-time signal self-diagnosis. When the visual inspection module's lens image becomes blurred due to downhole mud contamination, the signal self-diagnosis data gives its sensor health coefficient β1 as 0.6. The remanent magnetic eddy current detection module and the deformation detection module are working normally, and the health coefficient β2 is... When β3 is all 1, the corrected initial weights are first calculated: w1=0.4×0.6=0.24, w2=0.35×1=0.35, w3=0.25×1=0.25, and the total initial weights are 0.84. Then, the corrected weights are normalized to obtain the final weight coefficients: w1=0.24 / 0.84≈0.286, w2=0.35 / 0.84≈0.417, w3=0.25 / 0.84≈0.297, and the total weights are 1. In the subsequent calculation of the defect comprehensive density, the above normalized dynamic weights are used for calculation.

[0030] Furthermore, The specific operating speed of the real-time control downhole monitoring system described in S4 is as follows: When the defect comprehensive density Below the preset low threshold At that time, the ground control system controls the downhole detection system to operate at a normal speed via a cable winch; When the defect comprehensive density At the preset low threshold With preset high threshold Between, and within the sliding window, the maximum severity level of the defects is light or moderate, and the surface control system controls the downhole detection system to operate at a normal speed via a cable winch; When the defect comprehensive density At the preset low threshold With preset high threshold Between, and the maximum severity level within the sliding window is severe, or the overall defect density Higher than or equal to the preset high threshold At that time, the ground control system immediately switches the downhole detection system to low-speed operation via a cable winch.

[0031] In practical implementation, a low threshold for the overall defect density is preset. The threshold is 0.3%. The defect density is 0.8%, the normal operating speed is 0.6 m / s, and the low-speed operating speed is 0.1 m / s. The severity of the defect is divided into three levels: mild, moderate, and severe. When inspecting the normal well section of the casing, the comprehensive defect density within the current sliding window is calculated. The value is 0.15%, which is lower than the preset low threshold. The surface control system controls the downhole detection system to operate at a conventional speed of 0.6 m / s via a cable winch to ensure detection efficiency; when a certain well section is detected, calculations are performed. It is 0.5%, which is in and Between these points, and given that the maximum severity level of defects within the sliding window is moderate, the ground control system continues to control the instrument at a conventional speed of 0.6 m / s to avoid frequent speed adjustments affecting detection efficiency; when detecting the next well section, the calculation is... It remains at 0.5%, at a level where and However, within the sliding window, there was a severe deformation defect, with a maximum severity level of severe. The ground control system immediately switched the instrument's operating speed to a low speed of 0.1 m / s via a cable winch to increase the sampling density in that well section, ensuring the detection accuracy and data integrity of the severe defect. When the detection reached a well section with dense defects, the calculations yielded... The figure is 1.2%, which is higher than the preset high threshold. Regardless of the severity of the defects, the ground control system immediately switches the instrument to a low speed of 0.1 m / s to perform detailed inspections of densely defective areas. The fully automated closed-loop control significantly reduces the workload of on-site operators and improves the automation level of the inspection operation.

[0032] Furthermore, The integrated inspection report described in S5 includes defect type determination results, comprehensive defect severity level, defect axial and circumferential positioning coordinates, deformation parameters, metal loss assessment data, defect video screenshots, and multi-dimensional defect distribution maps.

[0033] An integrated process of "inspection data - defect identification - quantitative assessment - result output" is established. In specific implementation, after the inspection of the entire 2000m casing section is completed, the ground control and processing system automatically generates an integrated inspection report. The report first includes a summary of defect statistics for the entire casing section, including the total number and proportion of various defects, and the distribution of defects of different severity levels. Secondly, for each identified defect, the report clearly marks the defect type judgment result, such as internal wall corrosion pit, axial crack, casing diameter reduction deformation, etc., the comprehensive severity level of the defect, the axial depth positioning coordinates and circumferential angle positioning coordinates of the defect, deformation parameters (such as diameter reduction, ellipticity), and metal loss assessment data (such as maximum metal loss rate, average metal loss depth). At the same time, the report includes visual inspection video screenshots of the corresponding defect locations, local magnified views of multi-dimensional defect distribution maps, and a comprehensive multi-dimensional defect distribution map of the entire well section. This provides complete data support for the safety level assessment of the casing and the accurate formulation of repair plans. The standardized report format also facilitates the user unit to establish a casing integrity management file and realize standardized operation and maintenance management throughout the entire life cycle of the casing.

[0034] Furthermore, A multi-parameter integrated casing damage detection system is used to implement the multi-parameter integrated casing damage detection method described above, including a downhole detection system and a surface control and processing system. The downhole detection system is an integrated instrument string structure, consisting of a cable adapter, a data preprocessing module, a multi-parameter detection sub, and a centering support from top to bottom. The multi-parameter detection sub integrates a visual inspection module, a deformation detection module, a residual magnetic eddy current detection module, a positioning module, and a storage module. The positioning module incorporates a photoelectric coded odometer and an inertial navigation unit. The data preprocessing module uses an FPGA+ARM architecture, incorporating both an FPGA and an ARM unit. The visual inspection module, deformation detection module, residual magnetic eddy current detection module, and positioning module all possess signal self-diagnostic capabilities. The outputs of the visual inspection module, deformation detection module, residual magnetic eddy current detection module, and positioning module are all electrically connected to the inputs of the storage module and data preprocessing module, respectively; the data preprocessing module is bidirectionally connected to the ground control and processing system via a logging armored cable. The ground control and processing system includes a cable winch, a ground power supply system, a ground control system, a signal processing and data acquisition system, and an industrial control computer. The input end of the signal processing and data acquisition system is connected to the logging armored cable, and the output end is connected to the industrial control computer. The output end of the industrial control computer is connected to the ground control system. The ground control system is electrically connected to the cable winch and is used to regulate the operating speed of the downhole detection system.

[0035] In practical implementation, the downhole inspection system is designed as an integrated instrument string structure with an outer diameter of 102mm, suitable for inspection operations on 5.5-inch casing. The instrument string, from top to bottom, includes a cable adapter, a data preprocessing module, a multi-parameter inspection section, and a centering support. The cable adapter is compatible with a 7-core logging armored cable, enabling power supply and bidirectional communication. The multi-parameter inspection section, from top to bottom, integrates a vision inspection module, a deformation detection module, a residual magnetic eddy current detection module, a positioning module, and a storage module. The vision inspection module uses a high-temperature resistant wide-angle lens and a ring-shaped supplementary lighting source to achieve 360° panoramic imaging of the casing inner wall. The deformation detection module employs a 16-channel circumferentially arranged ultrasonic transducer to detect the full circumferential deformation of the casing. The residual magnetic eddy current detection module uses a circumferential array of eddy current sensors to detect metal loss in the casing wall thickness. The positioning module integrates a photoelectric coded odometer and a MEMS inertial navigation unit to achieve odometer and spatial attitude positioning of the instrument. The storage module uses a high-temperature resistant industrial-grade storage chip with a capacity of 128GB, enabling local backup of raw detection data for the entire well section. The data preprocessing module adopts an FPGA+ARM architecture, with the FPGA unit performing parallel filtering, format normalization, and time-division multiplexing of multi-channel detection data. The ARM unit performs data protocol encapsulation and transmission control through preprocessing. The outputs of each detection module are electrically connected to the inputs of the storage module and the data preprocessing module, enabling local data storage and real-time uploading. The ground control and processing system integrates a cable winch, a ground power supply system, a ground control system, a signal processing and data acquisition system, and an industrial control computer. The logging armored cable is wound around the cable winch, and the ground power supply system powers the downhole instruments. The signal processing and data acquisition system demultiplexes and restores the data uploaded from the downhole, transmitting it to the industrial control computer for data analysis and processing. The motion control commands generated by the industrial control computer are transmitted to the ground control system, which controls the speed of the cable winch, enabling real-time control of the downhole instrument's operating speed. Compared to the traditional method of multiple instruments being deployed into the well for testing, this integrated instrument string structure shortens the operation time and significantly reduces the cost and safety risks of well workover operations. The FPGA+ARM architecture preprocessing module meets the real-time uploading requirements of massive amounts of detection data. Downhole local storage enables dual backup of the original data. The bidirectional communication and closed-loop control between the ground and downhole systems achieve fully automated operation of the detection process, significantly improving the efficiency and reliability of the detection operation.

[0036] It is understood that the above embodiments are merely exemplary implementations used to illustrate the principles of the present invention, and the present invention is not limited thereto. For those skilled in the art, various modifications and improvements can be made without departing from the spirit and essence of the present invention, and these modifications and improvements are also considered to be within the scope of protection of the present invention.

Claims

1. A multi-parameter integrated casing damage detection method, characterized in that, This includes downhole detection systems and surface control and processing systems, as detailed below: S1. Synchronous data acquisition: The detection system acquires visual inspection data of the inner wall of the casing, deformation inspection data of the casing, residual magnetic eddy current inspection data of the casing, and positioning inspection data of the instrument as multiple raw data. S2. Data Preprocessing and Transmission: Preprocess and reuse multiple raw data streams, and upload the processed data to the ground control and processing system in real time via logging armored cable; S3. Ground data restoration and fusion analysis: The processed data is demultiplexed and signal restored to obtain multi-channel detection data. The multi-channel detection data is then sequentially subjected to spatial location alignment, feature-level fusion, and intelligent defect identification to generate a multi-dimensional defect distribution map. S4. Adaptive Motion Control: Based on a multi-dimensional defect distribution map, the comprehensive defect density of the current detection area is calculated through a preset sliding window. Motion control commands are generated in combination with the severity of the defects. The ground control system adjusts the running speed of the downhole detection system in real time through a cable winch according to the motion control commands. Specifically, the calculation of the overall defect density of the current detection area through a preset sliding window is as follows: Calculate the first in the sliding window Equivalent area of ​​class defects Its calculation formula is In the formula, For the first in the sliding window The number of class defects, For the first in the sliding window The actual area of ​​each defect, and the equivalent area of ​​linear defects such as cracks. For the first The severity coefficient of a defect is positively correlated with the severity level of the defect; Calculate the first Single-class defect density of class defects In the formula, The area covered by the sliding window. Let be the axial length of the sliding window. The circumferential circumference of the sleeve to be tested, and the density of single-type defects. Mapping to the corresponding severity level density interval to determine the first The severity level of the defect is determined, thereby obtaining the maximum severity level within the sliding window; Calculate the overall defect density within the sliding window In the formula, This represents the total number of defect types. For the first The weight coefficients of the class defects, and satisfying ; S5. Inspection Results Output: After the inspection of the entire pipe section is completed, the ground control and processing system generates an integrated inspection report that includes defect visualization information, quantitative parameters, location coordinates, and severity level.

2. The multi-parameter integrated sleeve damage detection method according to claim 1, characterized in that, The spatial alignment described in S3 specifically refers to: Based on the visual inspection module, deformation inspection module, and residual magnetic eddy current detection module of the detection system, the fixed axial spacing within the multi-parameter detection section is measured. The axial movement speed of the downhole detection system is acquired in real time by the ground control system. Calculate the time difference between each detection module passing the same defect location on the sleeve. : The ground control and processing system uses the calculated time difference By combining the sampling rates of each detection module, the time axis offset alignment is performed on the multi-channel detection data, which includes detection data from different detection modules, so that the multi-channel detection data corresponds to the same physical position of the sleeve.

3. The multi-parameter integrated sleeve damage detection method according to claim 1, characterized in that, In S3, feature-level fusion is achieved using a strategy that combines feature vector concatenation with weighted interactive fusion, specifically: Extract texture features and grayscale distribution features from visual inspection data, deformation features and deformation gradient features from deformation inspection data, and signal amplitude and phase features from remanent eddy current detection data; Multi-dimensional feature vectors at the same spatial location are concatenated end to end to form a high-dimensional joint feature vector; then, based on the detection sensitivity of each detection module corresponding to different defect types, learnable fusion weights are assigned to complete weighted interactive fusion; finally, the fused feature vector, the comprehensive confidence of the defect, and the overall defect judgment result are output.

4. The multi-parameter integrated sleeve damage detection method according to claim 1, characterized in that, The specific steps for generating a multi-dimensional defect distribution map as described in S3 are: Based on the positioning and detection data collected by the inertial navigation unit in the positioning module of the detection system, a spatial coordinate system for the detection path is established; the multi-path detection data after spatial alignment and the fused feature values ​​obtained by feature-level fusion are mapped to the corresponding coordinate points in the spatial coordinate system. Unsampled locations between sampling intervals are interpolated to become continuous; finally, a multi-dimensional defect distribution map is formed by overlaying feature layers of various dimensions and a fused defect comprehensive layer on the geometric contour of the detection path as the base layer.

5. The multi-parameter integrated casing damage detection method according to claim 1, characterized in that, The weighting coefficients are also dynamically corrected based on the sensor health coefficients, and their calculation formula is as follows: In the formula, For the first Weighting coefficients for class defects For the preset first Static base weights of class defects For the first The sensor health coefficient of the corresponding defect detection module, with a value ranging from 0 to 1, is provided in real time by the signal self-diagnosis data of the corresponding detection module; after correction, for all Normalization is performed.

6. The multi-parameter integrated casing damage detection method according to claim 1, characterized in that, The specific operating speed of the real-time control downhole monitoring system described in S4 is as follows: When the defect comprehensive density Below the preset low threshold At that time, the ground control system controls the downhole detection system to operate at a normal speed via a cable winch; When the defect comprehensive density At the preset low threshold With preset high threshold Between, and within the sliding window, the maximum severity level of the defects is light or moderate, and the surface control system controls the downhole detection system to operate at a normal speed via a cable winch; When the defect comprehensive density At the preset low threshold With preset high threshold Between, and the maximum severity level within the sliding window is severe, or the overall defect density Higher than or equal to the preset high threshold At that time, the ground control system immediately switches the downhole detection system to low-speed operation via a cable winch.

7. The multi-parameter integrated casing damage detection method according to claim 1, characterized in that, The integrated inspection report described in S5 includes defect type determination results, comprehensive defect severity level, defect axial and circumferential positioning coordinates, deformation parameters, metal loss assessment data, defect video screenshots, and multi-dimensional defect distribution maps.

8. A multi-parameter integrated casing damage detection system, utilizing any one of the multi-parameter integrated casing damage detection methods described in claims 1-7, characterized in that, Includes downhole detection systems and surface control and processing systems; The downhole detection system is an integrated instrument string structure, consisting of a cable adapter, a data preprocessing module, a multi-parameter detection sub, and a centering support from top to bottom. The multi-parameter detection sub integrates a visual inspection module, a deformation detection module, a residual magnetic eddy current detection module, a positioning module, and a storage module. The positioning module incorporates a photoelectric coded odometer and an inertial navigation unit. The data preprocessing module uses an FPGA+ARM architecture, incorporating both an FPGA and an ARM unit. The visual inspection module, deformation detection module, residual magnetic eddy current detection module, and positioning module all possess signal self-diagnostic capabilities. The outputs of the visual inspection module, deformation detection module, residual magnetic eddy current detection module, and positioning module are all electrically connected to the inputs of the storage module and data preprocessing module, respectively; the data preprocessing module is bidirectionally connected to the ground control and processing system via a logging armored cable. The ground control and processing system includes a cable winch, a ground power supply system, a ground control system, a signal processing and data acquisition system, and an industrial control computer. The input end of the signal processing and data acquisition system is connected to the logging armored cable, and the output end is connected to the industrial control computer. The output end of the industrial control computer is connected to the ground control system. The ground control system is electrically connected to the cable winch and is used to regulate the operating speed of the downhole detection system.

Images