A method and system for evaluating the service risk of a drill pipe based on a multi-source data stream

Abstract

The application discloses a kind of based on the drill pipe service risk assessment method and system of multi-source data stream, specifically relates to the technical field of oil drilling equipment failure prediction and health management, for solving the defects that existing method relies on static threshold and average experience cannot effectively evaluate the progressive fatigue damage risk of drill pipe under dynamic alternating load;It is by obtaining drill pipe service multi-source data and reconstructing load time domain data, it is converted into load spectrum to judge high-cycle fatigue risk state, and then analyze the load driving factor of driving material damage trend change, compare the load driving factor of current drill pipe with the historical data of same type drill pipe group to identify abnormal response mode, then it is similar to the damage evolution path library based on historical data to deduce potential damage evolution path, finally based on deduced path determines and outputs dynamic risk assessment value to realize early warning and predictive health management to drill pipe hidden fatigue risk.

Description

Technical Field

[0001] This invention relates to the field of oil drilling equipment failure prediction and health management technology, and more specifically, to a method and system for assessing drill pipe service risk based on multi-source data streams. Background Technology

[0002] In the field of oil and gas drilling, especially in ultra-deep wells and complex structure wells, drill pipe, as a key component of the drill string system, directly affects drilling safety, efficiency, and cost. To achieve drill pipe fault prediction and health management, current technologies generally employ a combination of multi-source data monitoring and human experience-based judgment. Specifically, sensors deployed on the surface and downhole collect multi-source data streams in real time, such as torque, tension, vibration, rotational speed, and drilling fluid parameters. Field systems or engineers typically trigger alarms for out-of-limit single-dimensional data parameters based on preset fixed thresholds. Combining this with the drill pipe's average historical lifespan data and personal experience, a comprehensive risk assessment is conducted, and a decision is made regarding whether to trip and perform maintenance. This is the main technical practice for drill pipe service health management currently.

[0003] However, the risk assessment logic of the aforementioned existing methods is essentially based on static, discrete thresholds and averaged historical experience to evaluate a progressive damage accumulation process driven by continuous, dynamic, and random alternating loads. The core failure modes of drill pipes, such as the initiation and propagation of fatigue cracks, are precisely caused by the continuous accumulation of microscopic damage under conditions of frequent changes in load amplitude and long-term lower than the material's yield strength. Existing technologies simplify this dynamically evolving risk to a judgment on whether instantaneous parameters exceed a certain static safety boundary, resulting in a serious lag or even omission in the identification of early and latent fatigue damage risks, making it impossible to achieve true early warning and predictive health management. Summary of the Invention

[0004] To overcome the aforementioned deficiencies of the prior art, the present invention provides a method and system for assessing the service risk of drill pipe based on multi-source data streams to solve the problems mentioned in the background art.

[0005] To achieve the above objectives, the present invention provides the following technical solution: A method for assessing the service risk of drill pipe based on multi-source data streams includes the following steps: S1. Acquire multi-source sensor data of the drill pipe during its service life, and reconstruct the time-domain load data of the drill pipe based on the multi-source sensor data; S2. Convert the load time domain data into a load spectrum, and determine whether the drill pipe is under a load state that is prone to inducing high cycle fatigue based on the load spectrum; S3. When it is determined that the load is in a load state that is prone to inducing high cycle fatigue, analyze the load driving factors related to the trend change of the micro-damage state of the driving drill pipe material from the load spectrum. S4. Compare and analyze the statistical characteristics of the load driving factor of the current drill pipe with the historical service data of the same type of drill pipe group to identify whether the current drill pipe has an abnormal response pattern that deviates significantly from the group pattern. S5. Match the load driving factor and abnormal response mode of the current drill pipe with the damage evolution path library constructed based on the historical service data of the same type of drill pipe group to deduce the potential damage evolution path of the current drill pipe. S6. Based on the potential damage evolution path, determine and output the dynamic assessment value of the drill pipe service risk.

[0006] Furthermore, multi-source sensor data of the drill pipe during its service life is acquired, and the time-domain load data of the drill pipe is reconstructed based on the multi-source sensor data, including: Acquire sensor data including torque, axial force, triaxial vibration, and rotational speed; Obtain drilling operating parameters, including drilling pressure, drilling fluid density, and displacement. Based on the drill string dynamics model, sensor data and drilling condition parameters are used as inputs to calculate the time-domain data of the drill string load. The time-domain data of the load includes the time history of alternating bending stress and the time history of alternating axial stress.

[0007] Furthermore, the load time-domain data is converted into a load spectrum, and based on the load spectrum, it is determined whether the drill pipe is under a load state that easily induces high-cycle fatigue, including: Rainflow counting was performed on the alternating bending stress time history and alternating axial stress time history to obtain the load spectrum characterized by stress amplitude and mean value; Extract all load cycles whose stress amplitude exceeds a preset stress amplitude threshold from the load spectrum, and calculate the proportion of load cycles whose stress amplitude exceeds the preset stress amplitude threshold in the total number of cycles; If the ratio continues to exceed the preset ratio threshold, the drill pipe is determined to be under a load state that is prone to inducing high-cycle fatigue.

[0008] Furthermore, when the load condition is determined to be prone to inducing high-cycle fatigue, the load driving factors related to the trend change in the micro-damage state of the driving drill pipe material are analyzed from the load spectrum, including: Based on the stress amplitude and mean in the load spectrum, the equivalent bending stress amplitude and equivalent axial stress amplitude are calculated by modifying the Goodman equation; For the equivalent bending stress amplitude and the equivalent axial stress amplitude, calculate their mean and coefficient of variation over the continuous monitoring period, respectively. The mean value of the equivalent bending stress amplitude, the coefficient of variation of the equivalent bending stress amplitude, the mean value of the equivalent axial stress amplitude, and the coefficient of variation of the equivalent axial stress amplitude are used as load driving factors.

[0009] Furthermore, the equivalent bending stress amplitude and equivalent axial stress amplitude are calculated by modifying the Goodman equation, including: Based on the stress amplitude and mean, and combined with the tensile strength and fatigue limit of the drill pipe material, the equivalent stress amplitude under bending load and axial load is calculated respectively. The equivalent stress amplitude is used to characterize the stress level equivalent to symmetrical cyclic fatigue damage under the influence of mean stress.

[0010] Furthermore, the statistical characteristics of the current drill pipe's load driving factors are compared and analyzed with historical service data of a group of drill pipes of the same type to identify whether the current drill pipe exhibits an abnormal response pattern that significantly deviates from the group's pattern, including: Based on historical service data of the same type of drill pipe group, a reference distribution of the load driving factor vector is established, which consists of the mean of the equivalent bending stress amplitude, the coefficient of variation of the equivalent bending stress amplitude, the mean of the equivalent axial stress amplitude, and the coefficient of variation of the equivalent axial stress amplitude. Calculate the Mahalanobis distance between the current drill pipe load driving factor vector and the reference distribution; If the Mahalanobis distance exceeds the preset anomaly detection threshold, then an abnormal response pattern that significantly deviates from the group pattern is identified in the current drill pipe.

[0011] Furthermore, the load driving factors and abnormal response patterns of the current drill pipe are matched with a damage evolution path library constructed based on historical service data of a group of drill pipes of the same type to deduce the potential damage evolution path of the current drill pipe, including: The historical service data of the same type of drill pipe group are organized into a time window sequence for each complete record from initial service to damage or decommissioning. Each time window contains a load driving factor vector consisting of the mean of the equivalent bending stress amplitude, the coefficient of variation of the equivalent bending stress amplitude, the mean of the equivalent axial stress amplitude, and the coefficient of variation of the equivalent axial stress amplitude, thereby constructing a damage evolution path library. Calculate the dynamic time warping distance between the load driving factor vector sequence of the latest consecutive time windows of the current drill pipe and the vector sequences of each path in the damage evolution path library at the same time stage; Based on the dynamic time warping distance, several historical damage evolution paths with the highest similarity are selected as potential damage evolution paths for the current drill pipe.

[0012] Furthermore, when constructing the damage evolution path library, the damage state in the complete record of damage or decommissioning is quantitatively calibrated by matching the changing trend of the continuous load driving factor vector of the drill pipe within the corresponding time window with the preset damage evolution mode.

[0013] Furthermore, based on the potential damage evolution path, a dynamic assessment value of the drill pipe service risk is determined and output, including: For each potential damage evolution path, extract the number of remaining time windows experienced from the time window that matches the current drill pipe state until the path terminates. Based on the statistical characteristics of the number of remaining time windows corresponding to all potential damage evolution paths, an assessment value characterizing the remaining safe service life of the drill pipe is determined. The assessment value is output as a dynamic assessment value of the service risk of the drill pipe.

[0014] On the other hand, the present invention provides a drill pipe service risk assessment system based on multi-source data streams, comprising the following modules: The data acquisition module is used to acquire multi-source sensor data of the drill pipe during its service life, and to reconstruct the time-domain load data of the drill pipe based on the multi-source sensor data. The load spectrum analysis module is used to convert load time-domain data into a load spectrum and determine whether the drill pipe is under a load state that is prone to inducing high-cycle fatigue based on the load spectrum. The factor extraction module is used to analyze the load driving factors related to the trend changes in the micro-damage state of the driving drill pipe material from the load spectrum when it is determined that the load state is prone to inducing high-cycle fatigue. The comparison and identification module is used to compare and analyze the statistical characteristics of the load driving factors of the current drill pipe with the historical service data of the same type of drill pipe group to identify whether the current drill pipe has an abnormal response pattern that deviates significantly from the group pattern. The matching and deduction module is used to match the load driving factors and abnormal response patterns of the current drill pipe with the damage evolution path library built based on the historical service data of a group of drill pipes of the same type, and to deduce the potential damage evolution path of the current drill pipe. The evaluation output module is used to determine and output dynamic evaluation values ​​of drill pipe service risk based on potential damage evolution paths.

[0015] Compared with the prior art, the present invention has the following beneficial effects: 1. By constructing a mapping relationship from multi-source data to load-driving factors, discrete physical signals are transformed into quantitative indicators characterizing the material damage evolution trend. This allows for the extraction of key driving elements directly related to the micro-damage accumulation process from complex alternating loads. This enables the assessment of drill pipe health status to move beyond superficial judgments of whether single-point parameters exceed limits, and to a continuous analysis of the intrinsic driving forces of damage development. This analytical approach eliminates reliance on fixed safety thresholds and can keenly capture early fatigue risk signs induced by dynamic characteristics such as load fluctuations below the average load level, thus achieving proactive perception and identification of latent damage development.

[0016] 2. By introducing historical service data from a group of drill pipes of the same type, a dynamically changing statistical benchmark and evolutionary reference is provided for the current assessment of the drill pipe. By comparing real-time individual data with historical patterns of the group, abnormal response patterns caused by individual differences or early defects can be effectively identified. Based on this, similarity matching and inference are performed using a historical damage evolution path library, which can associate the current state with complete service histories with similar starting points in history. This provides multiple statistically significant potential evolutionary trajectories for the future damage development of the current drill pipe. The final output dynamic assessment value is a quantitative prediction result about the remaining safe service life based on historical similarity reasoning and statistical induction. This provides direct and objective data support for predictive maintenance decisions on site, realizing a shift from passive alarm to proactive prediction in health management. Attached Figure Description

[0017] Figure 1 This is a flowchart of a drill pipe service risk assessment method based on multi-source data streams according to the present invention; Figure 2 This is a schematic diagram of the structure of a drill pipe service risk assessment system based on multi-source data streams according to the present invention. Detailed Implementation

[0018] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0019] Example 1: Figure 1 This invention presents a method for assessing the service risk of drill pipes based on multi-source data streams, which includes the following steps: S1. Acquire multi-source sensor data of the drill pipe during its service life, and reconstruct the time-domain load data of the drill pipe based on the multi-source sensor data; S2. Convert the load time domain data into a load spectrum, and determine whether the drill pipe is under a load state that is prone to inducing high cycle fatigue based on the load spectrum; S3. When it is determined that the load is in a load state that is prone to inducing high cycle fatigue, analyze the load driving factors related to the trend change of the micro-damage state of the driving drill pipe material from the load spectrum. S4. Compare and analyze the statistical characteristics of the load driving factor of the current drill pipe with the historical service data of the same type of drill pipe group to identify whether the current drill pipe has an abnormal response pattern that deviates significantly from the group pattern. S5. Match the load driving factor and abnormal response mode of the current drill pipe with the damage evolution path library constructed based on the historical service data of the same type of drill pipe group to deduce the potential damage evolution path of the current drill pipe. S6. Based on the potential damage evolution path, determine and output the dynamic assessment value of the drill pipe service risk.

[0020] To implement step S1 of this method, multi-source sensor data generated by the drill pipe during drilling operations is acquired. Raw data is collected using sensors installed on the surface portion of the drill string system and on the downhole measurement-while-drilling (MWD) tool. The surface sensors include a torque meter and a weight indicator. The torque meter measures the torque signal transmitted to the top of the drill string, and the weight indicator measures the change in suspended weight of the drill string. Based on the difference between the weight indicator reading and the weight of the drill string in air, combined with empirical values ​​of the wellbore trajectory friction coefficient, the axial force borne by the downhole drill pipe is calculated. The downhole MWD tool is installed in the drill string assembly near the drill bit. Its built-in accelerometer measures the vibration acceleration of the drill pipe in three mutually perpendicular directions at a high sampling rate, acquiring triaxial vibration data. The rotational speed sensor in the downhole MWD tool simultaneously measures the angular velocity of the drill pipe rotating about its axis, acquiring rotational speed data. The torque signal, calculated axial force, triaxial vibration data, and rotational speed data are transmitted to the surface computer system via mud pulse telemetry or electromagnetic waves.

[0021] Obtain drilling condition parameters directly related to the drilling process. These parameters are derived from instrument readings on the driller's console and a real-time drilling database. The drilling condition parameters include pressure on the drill bit (PBD), drilling fluid density, and drilling fluid displacement. PBD is read directly from the weight indicator or calculated from the suspended weight and the target pressure applied to the drill bit; the unit of PBD is kilonewtons (kN). Drilling fluid density is measured in real-time by a densitometer installed on the mud circulation line; the unit of drilling fluid density is grams per cubic centimeter (g / cm³). Drilling fluid displacement is calculated by multiplying the number of pump strokes recorded by the mud pump's stroke counter by the pump's volumetric efficiency, or by directly measuring it using an electromagnetic flowmeter; the unit of drilling fluid displacement is liters per second (L / s). These drilling condition parameters are synchronized with multi-source sensor data using a unified time synchronization system to ensure that each set of data has a corresponding timestamp.

[0022] Based on the drill string dynamics model, multi-source sensor data and drilling condition parameters are used as inputs to calculate the time-domain load data of the drill pipe. The drill string dynamics model is established using the finite element method or the lumped mass method. This model discretizes the drill string into multiple elements, each with mass, stiffness, and damping properties. The torque signal measured on the surface is used as the torque boundary condition input at the top of the model; the axial force on the downhole drill pipe, calculated from the model, is used as the axial force boundary condition input at the top of the model. The triaxial vibration acceleration data measured downhole is high-pass filtered to remove the gravitational acceleration component, and then numerically integrated in the time domain to convert it into a vibration velocity sequence. This vibration velocity sequence serves as a motion calibration reference or additional excitation for the model at corresponding downhole nodes. The drilling parameters, including rotational speed, drilling pressure, drilling fluid density, and drilling fluid displacement, are used as environmental parameters input to the model. Rotational speed is used to determine the angular velocity boundary of the model rotation. Drilling pressure is converted into a concentrated force acting at the bottom of the model, i.e., the drill bit. Drilling fluid density and drilling fluid displacement are used to calculate the buoyancy of the drilling fluid on the drill string, the annular pressure drop, and the viscous damping force generated by the relative motion with the drill string.

[0023] After setting all boundary conditions, loads, and excitations, the drill string dynamics model is solved. The solution process is completed using numerical integration methods, such as the Newmark-Beta method or the Runge-Kutta method, to calculate the stress tensor of the target drill string segment, for example, at a depth of 3,000 meters downhole, at each discrete time point within the model simulation time domain. From the calculated stress tensor, two key components are decomposed: one is the maximum normal stress in the plane perpendicular to the drill string axis, generated by the overall bending or local whirl deformation of the drill string; its time-varying sequence is extracted to obtain the alternating bending stress time history. The other component is the normal stress along the drill string axis, generated by the overall tension, compression, and longitudinal vibration of the drill string; its time-varying sequence is extracted to obtain the alternating axial stress time history. The alternating bending stress time history and the alternating axial stress time history together constitute the load time-domain data, which contains complete time-series information of the dynamic fatigue loads borne by the drill string. The unit of the alternating bending stress time history is megapascals (MPA), and the unit of the alternating axial stress time history is also MPA. The payload time-domain data is stored in computer memory or non-volatile storage media for use in subsequent steps.

[0024] To implement step S2 of this method, the load time-domain data is converted into a load spectrum. The load time-domain data includes alternating bending stress time histories and alternating axial stress time histories. Rainflow counting algorithms are executed on the alternating bending stress time histories and alternating axial stress time histories respectively. The execution process of the rainflow counting algorithm involves rearranging the time series data of the alternating bending stress time histories, using the peaks and valleys in the sequence as turning points to generate a new peak-valley sequence. A starting point is set, and each stress point in the peak-valley sequence is examined sequentially in chronological order, simulating rainflow flowing along a stress-time trajectory. The rainflow stops when it encounters rainflow from the opposite direction at a higher stress level, thus identifying a complete stress cycle. For an identified complete stress cycle, the stress amplitude and mean of the cycle are calculated. The stress amplitude is equal to the difference between the maximum and minimum stress values ​​in the cycle divided by 2. The mean is equal to the sum of the maximum and minimum stress values ​​in the cycle divided by 2. After performing rainflow counting on the alternating bending stress time histories, a series of cyclic data pairs consisting of bending stress amplitude and bending stress mean are obtained. After rainflow counting on the alternating axial stress time history, a series of cyclic data pairs consisting of axial stress amplitude and mean axial stress are obtained. By collecting and statistically analyzing all identified cyclic data pairs, a load spectrum characterized by stress amplitude and mean is obtained. The load spectrum represents the statistical distribution of cyclic loads at various stress levels experienced by the drill pipe within a given time period.

[0025] The load spectrum is used to determine whether the drill pipe is under a load state that easily induces high-cycle fatigue. All load cycles with stress amplitudes exceeding a preset stress amplitude threshold are extracted from the load spectrum. The preset stress amplitude threshold is set based on the fatigue limit of the drill pipe material. The fatigue limit of the drill pipe material is obtained through a standard rotating bending fatigue test; for example, for API S-135 grade drill pipe material, its fatigue limit value may be 200 MPa. The preset stress amplitude threshold can be set to a value equal to or slightly higher than the fatigue limit value, for example, a preset stress amplitude threshold set to 210 MPa. Another method for setting the preset stress amplitude threshold is based on statistical analysis of historical failure data of a group of drill pipes of the same type. Specifically, load spectrum data of historical high-cycle fatigue failure cases are collected, and the typical stress amplitude range leading to failure in these cases is analyzed. The lower limit of the typical stress amplitude range is used as the preset stress amplitude threshold. The extraction process involves traversing all load cycles in the load spectrum, comparing the stress amplitude of each load cycle with the preset stress amplitude threshold, and filtering out all load cycles with stress amplitudes greater than the preset stress amplitude threshold. The number of the filtered load cycles is counted and recorded as the high-stress amplitude cycle number. The total number of all load cycles in the statistical load spectrum is denoted as the total number of cycles. The ratio of the number of cycles with high stress amplitude to the total number of cycles is calculated. This ratio represents the proportion of load cycles with stress amplitude exceeding a preset stress amplitude threshold in the total number of cycles.

[0026] The determination of whether the proportion consistently exceeds a preset proportion threshold is crucial. The preset proportion threshold is an empirical or statistical value used to define the significance of high-cycle fatigue risk. The preset proportion threshold can be set based on engineering experience. For example, in drilling engineering practice, when the proportion of stress cycles exceeding the material fatigue limit consistently exceeds 5%, it is considered that the drill pipe has entered a rapid accumulation stage of high-cycle fatigue, and therefore the preset proportion threshold can be set to 5%. Alternatively, the preset proportion threshold can be obtained by training a machine learning model on historical data. Specifically, a large number of historical drilling data segments are collected. Each data segment contains a sequence of proportion values ​​calculated within a time window before drill pipe failure, along with a label indicating whether high-cycle fatigue failure ultimately occurred. Algorithms such as logistic regression or support vector machines are used for training, and the decision boundary value output by the model for classification can be used as the preset proportion threshold. The concept of consistently exceeding the threshold needs to be concretely implemented by defining a time window and a continuous condition. An evaluation time window is set, for example, with a length of 30 minutes. Within the evaluation time window, at fixed time intervals, such as 1 minute, the proportion of load cycles with stress amplitude exceeding the preset stress amplitude threshold in the total number of cycles is calculated, generating a time series of proportion values. If, in the time series of this ratio value, the ratio value at several consecutive calculation intervals is greater than the preset ratio threshold (e.g., the ratio value at 5 consecutive calculation intervals is greater than the preset ratio threshold), then the ratio is determined to continuously exceed the preset ratio threshold. When the condition of continuously exceeding the preset ratio threshold is met, the drill pipe is determined to be under a load state that easily induces high-cycle fatigue. If the time series of the ratio value does not meet the condition of continuous exceedance, or the amount of data within the evaluation time window is insufficient for calculation, then the drill pipe is determined not to be under a load state that easily induces high-cycle fatigue. The output result of determining whether the drill pipe is under a load state that easily induces high-cycle fatigue is a binary logic state identifier, which is stored in association with the corresponding load spectrum data.

[0027] To implement step S3 of this method, when it is determined that the load is in a load state that is prone to inducing high-cycle fatigue, the load driving factors related to the trend change of the micro-damage state of the driving drill pipe material are analyzed from the load spectrum. The load spectrum is derived from the output of step S2 and is characterized by stress amplitude and mean. The load spectrum includes the bending stress amplitude and mean bending stress for bending load and the axial stress amplitude and mean axial stress for axial load.

[0028] Based on the stress amplitude and mean value in the load spectrum, the equivalent bending stress amplitude and equivalent axial stress amplitude are calculated using the modified Goodman equation. The modified Goodman equation is an engineering model that considers the influence of mean stress on the fatigue life of materials. The specific process for calculating the equivalent bending stress amplitude is as follows: For each cyclic data pair in the load spectrum consisting of the bending stress amplitude and the mean bending stress, the bending stress amplitude and mean bending stress of that cyclic data pair are read. The tensile strength of the drill pipe material is obtained. The tensile strength of the drill pipe material can be obtained by consulting the material standard handbook; for example, for a specific grade of drill pipe material, its tensile strength is 1000 MPa, or it can be directly measured by sampling and performing a uniaxial tensile test. The bending fatigue limit of the drill pipe material is obtained. The bending fatigue limit of the drill pipe material can be obtained through a standard high-cycle fatigue test; for example, the fatigue limit value is 200 MPa. When calculating the equivalent bending stress amplitude, the bending stress amplitude is divided by a correction factor, which is equal to 1 minus the ratio of the mean bending stress to the tensile strength of the drill pipe material. The calculation process can be described as follows: the equivalent bending stress amplitude equals the bending stress amplitude divided by 1 minus the mean bending stress divided by the tensile strength of the drill pipe material. This calculation process is performed once for each pair of bending load cycles in the load spectrum, generating the equivalent bending stress amplitude value corresponding to each original bending cycle. The calculation of the equivalent axial stress amplitude uses the same principle but with axial load parameters. For each pair of cycle data consisting of the axial stress amplitude and the mean axial stress in the load spectrum, the axial stress amplitude and the mean axial stress are read. The tensile strength of the drill pipe material used here is the same as the value used when calculating the equivalent bending stress amplitude. The axial fatigue limit of the drill pipe material is obtained, which needs to be obtained through axial tensile-compressive fatigue tests; for example, the axial fatigue limit value is 180 MPa. When calculating the equivalent axial stress amplitude, the axial stress amplitude is divided by a correction factor, which is equal to 1 minus the ratio of the mean axial stress to the tensile strength of the drill pipe material. The calculation process can be described as follows: the equivalent axial stress amplitude equals the axial stress amplitude divided by 1 minus the mean axial stress divided by the tensile strength of the drill pipe material. This calculation is performed on each pair of axial load cycles in the load spectrum to generate the equivalent axial stress amplitude value corresponding to each original axial cycle. The equivalent bending stress amplitude and equivalent axial stress amplitude calculated by modifying the Goodman equation are stress parameters used for fatigue damage analysis.

[0029] For the equivalent bending stress amplitude and equivalent axial stress amplitude, calculate their mean and coefficient of variation over the continuous monitoring period. The continuous monitoring period is a pre-set sliding time window, for example, a continuous monitoring period of 30 minutes. Calculate the mean of the equivalent bending stress amplitude over the continuous monitoring period. Specifically, collect all equivalent bending stress amplitude values ​​calculated within this continuous monitoring period to form an equivalent bending stress amplitude dataset. Calculate the arithmetic mean of all values ​​in this dataset; this arithmetic mean is the mean of the equivalent bending stress amplitude over the continuous monitoring period. Calculate the coefficient of variation of the equivalent bending stress amplitude over the continuous monitoring period. The coefficient of variation is the ratio of the standard deviation to the mean. Calculate the standard deviation of the equivalent bending stress amplitude dataset. The standard deviation is calculated by taking the square root of the average of the squares of the differences between each value in the dataset and the mean of the equivalent bending stress amplitude. Then, divide the calculated standard deviation by the mean of the equivalent bending stress amplitude; the quotient is the coefficient of variation of the equivalent bending stress amplitude over the continuous monitoring period. Calculate the mean of the equivalent axial stress amplitude over the continuous monitoring period. The method involves collecting all equivalent axial stress amplitude values ​​calculated within the continuous monitoring period, forming an equivalent axial stress amplitude dataset. The arithmetic mean of all values ​​in this dataset is calculated to obtain the mean of the equivalent axial stress amplitude during the continuous monitoring period. The coefficient of variation of the equivalent axial stress amplitude during the continuous monitoring period is then calculated. Finally, the standard deviation of the equivalent axial stress amplitude dataset is calculated. The standard deviation is obtained by taking the square root of the average of the squares of the differences between each value and the mean of the equivalent axial stress amplitude, and then dividing this standard deviation by the mean of the equivalent axial stress amplitude.

[0030] The mean, coefficient of variation, mean, and coefficient of variation of the equivalent bending stress amplitude, equivalent axial stress amplitude, and equivalent axial stress amplitude are used as load driving factors. Each load driving factor comprises four specific values: the mean, coefficient of variation, and coefficient of variation of the equivalent bending stress amplitude. These four load driving factors collectively constitute a quantitative description of the key characteristics of the fatigue load currently applied to the drill pipe. These four load driving factors are output and used in subsequent analysis steps.

[0031] To implement step S4 of this method, the statistical characteristics of the load driving factor of the current drill pipe are compared and analyzed with the historical service data of a group of drill pipes of the same type to identify whether the current drill pipe has an abnormal response pattern that significantly deviates from the group's pattern. The load driving factor of the current drill pipe is derived from the output of step S3 and includes four specific values: the mean of the equivalent bending stress amplitude, the coefficient of variation of the equivalent bending stress amplitude, the mean of the equivalent axial stress amplitude, and the coefficient of variation of the equivalent axial stress amplitude. The group of drill pipes of the same type refers to a batch of drill pipes with the same specifications, steel grade, and manufacturing process as the current drill pipe. The historical service data of the group of drill pipes of the same type refers to the load driving factor data recorded by each drill pipe in the group during its respective service period during past drilling operations, obtained after processing by steps S1 to S3, as well as related operation records and final status records.

[0032] Based on historical service data of a group of drill pipes of the same type, a reference distribution of the load driving factor vector is established, consisting of the mean of the equivalent bending stress amplitude, the coefficient of variation of the equivalent bending stress amplitude, the mean of the equivalent axial stress amplitude, and the coefficient of variation of the equivalent axial stress amplitude. Specifically, all historical data entries for the group of drill pipes of the same type are extracted from the historical service database. Each historical data entry corresponds to the monitoring results of a drill pipe within a specific continuous monitoring period. This entry contains four values: the mean of the equivalent bending stress amplitude calculated within that period, the coefficient of variation of the equivalent bending stress amplitude calculated within that period, the mean of the equivalent axial stress amplitude calculated within that period, and the coefficient of variation of the equivalent axial stress amplitude calculated within that period. These four values ​​are arranged in a fixed order to form a four-dimensional load driving factor vector. The arrangement order can be: the first element is the mean of the equivalent bending stress amplitude, the second element is the coefficient of variation of the equivalent bending stress amplitude, the third element is the mean of the equivalent axial stress amplitude, and the fourth element is the coefficient of variation of the equivalent axial stress amplitude. By traversing all available historical data entries, each entry is transformed into a load-driving factor vector, resulting in a set of historical load-driving factor vectors. This set of historical vectors represents the load response patterns exhibited by a group of drill pipes of the same type under various historical operating conditions. The reference distribution is a statistical description of this set of historical vectors. Establishing the reference distribution requires calculating the statistical characteristics of this set of historical vectors, primarily the mean vector and covariance matrix of all historical vectors. The mean vector is a four-dimensional vector, where each element is the arithmetic mean of the values ​​of all vectors in the historical vector set in the corresponding dimension. The covariance matrix is ​​a 4x4 matrix that describes the relationships between the values ​​in the four dimensions of the historical vector set and their respective dispersion. The covariance matrix is ​​calculated by taking the average of the products of all historical vector values ​​in each pair of dimensions minus the mean of that pair. The mean vector and the covariance matrix together define the reference distribution. When establishing a reference distribution, historical data can be filtered, for example, only data from normal service periods when the drill pipe did not fail in the subsequent period can be selected to ensure that the reference distribution reflects a healthy or typical response pattern.

[0033] Calculate the Mahalanobis distance between the current drill pipe load driving factor vector and the reference distribution. The current drill pipe load driving factor vector is constructed based on the four load driving factor values ​​output in step S3, arranged in the same element order as when the reference distribution was established. That is, the current drill pipe load driving factor vector is also a four-dimensional vector containing the mean of the equivalent bending stress amplitude, the coefficient of variation of the equivalent bending stress amplitude, the mean of the equivalent axial stress amplitude, and the coefficient of variation of the equivalent axial stress amplitude. Mahalanobis distance is a distance metric that considers the correlation between data dimensions and standardizes the variance. Calculating the Mahalanobis distance requires the mean vector and covariance matrix of the reference distribution established in the previous steps. Specifically, the calculation steps are: subtract the mean vector of the reference distribution from the current drill pipe load driving factor vector to obtain a difference vector; calculate the inverse matrix of the covariance matrix of the reference distribution; calculate the product of this difference vector and the inverse matrix of the covariance matrix; calculate the dot product of this product vector and the original difference vector, and take the square root of the dot product result; the resulting value is the Mahalanobis distance. Mahalanobis distance is a dimensionless scalar value. When calculating the inverse of the covariance matrix, it must be ensured that the covariance matrix is ​​full rank and invertible.

[0034] If the Mahalanobis distance exceeds a preset anomaly threshold, the system identifies an abnormal response pattern in the current drill pipe that significantly deviates from the group's pattern. The preset anomaly threshold is a critical value used to determine the degree of anomaly. This threshold is set based on the statistical characteristics of historical data. One method is to use the Mahalanobis distance distribution of historical vector sets. The Mahalanobis distance from each historical load driving factor vector to its own reference distribution is calculated, resulting in a set of historical Mahalanobis distance values. The preset anomaly threshold can be set as a high percentile of this set of historical Mahalanobis distance values, such as the 95th or 99th percentile. Another method combines hypothesis testing principles, setting the threshold as a critical Mahalanobis distance value corresponding to a certain significance level. In real-time monitoring, the Mahalanobis distance of the current drill pipe is continuously calculated at the end of each continuous monitoring period. When the latest calculated Mahalanobis distance value is greater than the preset anomaly threshold, a judgment condition is triggered, and the system identifies an abnormal response pattern in the current drill pipe that significantly deviates from the group's pattern. This identification result is a logical flag. If the Mahalanobis distance does not exceed the preset anomaly detection threshold, then the current drill pipe is identified as not exhibiting an abnormal response pattern that significantly deviates from the group pattern. The identified abnormal response pattern serves as a key risk indication signal and is output to subsequent steps for further risk extrapolation and analysis.

[0035] To implement step S5 of this method, the load driving factor and abnormal response pattern of the current drill pipe are matched with a damage evolution path library constructed based on historical service data of a group of drill pipes of the same type, and the potential damage evolution path of the current drill pipe is deduced. The load driving factor of the current drill pipe is derived from the output of step S3 and includes the mean of the equivalent bending stress amplitude, the coefficient of variation of the equivalent bending stress amplitude, the mean of the equivalent axial stress amplitude, and the coefficient of variation of the equivalent axial stress amplitude. The abnormal response pattern is derived from the output of step S4 and is a binary logical flag indicating whether the load driving factor vector of the current drill pipe deviates significantly from the reference distribution of the group of drill pipes of the same type. The historical service data of the group of drill pipes of the same type refers to the complete service records of a batch of drill pipes of the same type as the current drill pipe collected and stored in past drilling operations. These records have been processed by steps S1 to S3 to obtain load driving factor data in units of continuous monitoring periods.

[0036] A damage evolution path database is constructed. This database is built based on historical service data of a group of drill pipes of the same type. Drill pipe data with complete service records are selected from the historical service database. A complete service record refers to the data of all continuous monitoring periods recorded throughout the entire process of a drill pipe from its initial commissioning until it fails due to damage or reaches its normal service life. For each drill pipe with a complete service record, its entire service life is divided into a series of continuous and non-overlapping time windows. The length of each time window is consistent with the continuous monitoring period used in step S3, for example, a time window of 30 minutes. For each time window, four load driving factor values ​​calculated within that time window are extracted from the historical data of the drill pipe: the mean of the equivalent bending stress amplitude, the coefficient of variation of the equivalent bending stress amplitude, the mean of the equivalent axial stress amplitude, and the coefficient of variation of the equivalent axial stress amplitude. These four values ​​are arranged in the exact same order as the load driving factor vector constructed in step S4, forming a four-dimensional load driving factor vector representing the load state of that time window. By connecting the load-driving factor vectors corresponding to all time windows throughout the entire service life of a drill pipe in chronological order, an ordered sequence of load-driving factor vectors is formed. This sequence constitutes a damage evolution path for the drill pipe. The damage evolution path library is a collection of damage evolution paths for all drill pipes of the same type with complete service records. When constructing the damage evolution path library, the endpoint state of each damage evolution path is calibrated, indicating whether the drill pipe corresponding to the path ultimately suffered damage or was normally decommissioned. The damage state in the complete record of damage or decommissioning is quantified by matching the changing trend of the continuous load-driving factor vectors of the drill pipe within the corresponding time window with a preset damage evolution pattern. The preset damage evolution pattern is a typical change pattern of the load-driving factor vectors in the period before failure, summarized from historical failure cases. The method for establishing the preset damage evolution pattern can be to collect load-driving factor vector sequences from multiple consecutive time windows before failure in a large number of historical failure cases, perform cluster analysis on these sequences, and use the cluster centers as the preset damage evolution pattern. Another approach involves domain experts defining typical failure precursor modes based on experience. For example, one mode could be defined as a continuous increase in the mean of the equivalent bending stress amplitude over five consecutive time windows, with a cumulative increase exceeding 20%. By matching the load driving factor vector sequence at the end of several consecutive time windows of each damage evolution path with various preset damage evolution modes, the damage state of the damage evolution path is determined based on the mode with the highest matching degree, such as flexural fatigue failure, axial fatigue failure, or no obvious damage.

[0037] The dynamic time warping distance is calculated between the load driving factor vector sequence of the most recent consecutive time windows of the current drill pipe and the vector sequences of each path in the damage evolution path library at the same time stage. The most recent consecutive time windows of the current drill pipe refer to several consecutive time windows of the same length as when the damage evolution path library was constructed, starting from the current moment and moving backward; for example, tracing back the last 10 consecutive monitoring periods. The mean, coefficient of variation of equivalent bending stress amplitude, mean, and coefficient of variation of equivalent axial stress amplitude calculated within each of these 10 time windows are extracted from the real-time monitoring data of the current drill pipe. These are then arranged in the same order to form a current load driving factor vector sequence consisting of 10 four-dimensional load driving factor vectors. For each historical damage evolution path in the damage evolution path library, an initial segment load driving factor vector sequence with the same length as the current load driving factor vector sequence, starting from the service start point, is selected; that is, the load driving factor vector sequence of the previous 10 time windows. Dynamic time warping distance is an algorithm used to measure the similarity between two time series. The specific process for calculating the dynamic time warping distance involves constructing a distance matrix. The number of rows in the distance matrix is ​​equal to the length of the current load driving factor vector sequence (10), and the number of columns is equal to the length of the load driving factor vector sequence of the initial segment of the historical damage evolution path (10). Each element in the distance matrix represents the Euclidean distance between a load driving factor vector in the current load driving factor vector sequence and a load driving factor vector in the initial segment of the historical damage evolution path. The Euclidean distance is calculated by summing the squares of the differences between the two four-dimensional load driving factor vectors in each dimension and then taking the square root. A path is found in the distance matrix from the top left corner to the bottom right corner that minimizes the sum of the matrix element values ​​traversed by the path. This path must satisfy monotonicity and continuity constraints, meaning that each step of the path can only move one unit to the right, one unit down, or one unit to the lower right. This path is the optimal warping alignment path. The sum of the matrix element values ​​traversed by the path is the dynamic time warping distance. For each historical damage evolution path in the damage evolution path library, the dynamic time warping distance between its initial segment load driving factor vector sequence and the current load driving factor vector sequence is calculated.

[0038] Based on the dynamic time warping distance, several historical damage evolution paths with the highest similarity are selected as potential damage evolution paths for the current drill pipe. After calculating the dynamic time warping distance between the current load driving factor vector sequence and the initial segments of all paths in the damage evolution path library, these dynamic time warping distance values ​​are sorted, and several historical damage evolution paths with the smallest dynamic time warping distance values ​​are selected. The selection threshold can be set empirically, for example, fixedly selecting the 5 historical damage evolution paths with the smallest dynamic time warping distance. The selection threshold can also be dynamically determined based on the distribution of dynamic time warping distance values, for example, selecting all historical damage evolution paths with dynamic time warping distance values ​​less than a certain absolute threshold. This absolute threshold can be determined by analyzing the distribution of matching distances of similar sequences in historical data, for example, taking the 75th percentile of historical distance values. If step S4 identifies an abnormal response pattern in the current drill pipe, then during the selection process, those historical damage evolution paths that are ultimately identified as damaged are given higher priority, or only those historical damage evolution paths that ultimately lead to damage are selected. The subsequent portions of these selected historical damage evolution paths—that is, the complete evolution trajectory from the time point aligned with the current load-driving factor vector sequence to the end of the damage evolution path—are taken as the potential damage evolution paths for the current drill pipe. The derived potential damage evolution paths include the subsequent load-driving factor vector sequence for each potential damage evolution path and its calibrated final damage state. The potential damage evolution paths are output to step S6 for risk assessment.

[0039] To implement step S6 of this method, a dynamic assessment value of the drill pipe service risk is determined and output based on the potential damage evolution path. The potential damage evolution path is derived from the output of step S5 and is a set containing one or more historical evolution trajectories selected from the historical damage evolution path library that are most similar to the early state of the current drill pipe. Each potential damage evolution path contains a complete load driving factor vector sequence from the alignment point with the current drill pipe state to the end point of the historical damage evolution path, as well as the damage state finally marked by the historical damage evolution path.

[0040] For each potential damage evolution path, the number of remaining time windows from the time window matching the current drill pipe state until the end of recording is extracted. The time window matching the current drill pipe state refers to the time window corresponding to the alignment point found on the historical damage evolution path through the dynamic time warping distance calculation in step S5, where the load driving factor vector sequence is most similar to the load driving factor vector sequence of the latest consecutive time windows of the current drill pipe. This alignment point divides the historical damage evolution path into two parts. The specific method for extracting the number of remaining time windows is to determine the number of complete time windows contained in the potential damage evolution path from the first time window after the alignment point to the last time window of the potential damage evolution path. The number of time windows is an integer. For example, if a potential damage evolution path has 50 time windows of data after the alignment point until the end of recording, then the number of remaining time windows extracted from the potential damage evolution path is 50. If the record of the historical damage evolution path terminates due to drill pipe damage and failure, then the remaining time window number represents the duration taken in historical cases to progress from a similar state to failure. The extracted remaining time window number is a time-related quantity, and its actual duration equals the remaining time window number multiplied by the physical duration of each time window. For example, if each time window is 30 minutes and the remaining time window number is 50, then the corresponding remaining physical duration is 1500 minutes.

[0041] Based on the statistical characteristics of the number of remaining time windows corresponding to all potential damage evolution paths, an assessment value characterizing the remaining safe service life of the drill pipe is determined. The number of remaining time windows corresponding to all potential damage evolution paths constitutes a dataset. The method for determining the assessment value is to perform statistical analysis on this dataset and select or calculate a representative value. One method for determining the assessment value is to take the minimum value of all remaining time windows. This method is based on the worst-case assumption. Another method is to calculate the arithmetic mean of all remaining time windows. This method is based on the assumption of average expectation. A weighted average can also be used as the assessment value, and the weights can be set according to the dynamic time warp distance calculated in step S5. The method for setting the weights is that the smaller the dynamic time warp distance, the more similar the historical damage evolution path is to the current state, and the greater its weight should be. A specific weight setting method is to set the weight as the reciprocal of the dynamic time warp distance. For example, if the dynamic time warp distance of a potential damage evolution path is 2, then its weight is 1 / 2 = 0.5. Another weight setting method is to use an exponential decay function, where the weight is equal to e raised to the power of the negative dynamic time warp distance. When calculating the weighted average, the number of remaining time windows for each potential damage evolution path is multiplied by its corresponding weight. All products are summed, and then divided by the sum of all weights to obtain the weighted average number of remaining time windows as the assessment value. A third method for determining the assessment value is to take a lower percentile of the number of remaining time windows, such as the 10th percentile. To convert the assessment value into a more intuitive representation of the remaining safe service life, the assessment value in terms of the number of time windows can be multiplied by the physical duration of each time window to obtain an assessment value in terms of time, such as an assessment value in terms of hours. The final assessment value characterizing the remaining safe service life of the drill pipe is a numerical value.

[0042] The assessed value is output as a dynamic assessment of the drill pipe's service risk. Output actions include storing the calculated assessment value in a designated record field of the database and sending the assessment value to a human-machine interface (HMI) for display. The HMI can be a graphical dashboard on a computer screen. The assessment value can be displayed in pure numerical form, such as the remaining hours. Alternatively, the assessment value can be converted into a risk level for display, which requires mapping the assessment value to discrete risk levels using preset risk level thresholds. The preset risk level thresholds are critical values ​​used to distinguish different risk levels. For example, two thresholds can be set: a high-risk threshold and a medium-risk threshold. If the assessment value is below the high-risk threshold, for example, the remaining time window converted to hours is less than 24 hours, a high-risk level is output. If the assessment value is between the medium-risk and high-risk thresholds, for example, between 24 hours and 72 hours, a medium-risk level is output. If the assessment value is above the medium-risk threshold, for example, greater than 72 hours, a low-risk level is output. The preset risk level thresholds can be set based on the safety requirements and experience of on-site operations. The output dynamic assessment value will be accompanied by a timestamp to indicate the assessment time corresponding to the assessment value. The output assessment values ​​and possible risk level information will be used to guide on-site operational decisions.

[0043] Example 2: Figure 2 A schematic diagram of a drill pipe service risk assessment system based on multi-source data streams is provided. This system includes the following modules: The data acquisition module is used to acquire multi-source sensor data of the drill pipe during its service life, and to reconstruct the time-domain load data of the drill pipe based on the multi-source sensor data. The load spectrum analysis module is used to convert load time-domain data into a load spectrum and determine whether the drill pipe is under a load state that is prone to inducing high-cycle fatigue based on the load spectrum. The factor extraction module is used to analyze the load driving factors related to the trend changes in the micro-damage state of the driving drill pipe material from the load spectrum when it is determined that the load state is prone to inducing high-cycle fatigue. The comparison and identification module is used to compare and analyze the statistical characteristics of the load driving factors of the current drill pipe with the historical service data of the same type of drill pipe group to identify whether the current drill pipe has an abnormal response pattern that deviates significantly from the group pattern. The matching and deduction module is used to match the load driving factors and abnormal response patterns of the current drill pipe with the damage evolution path library built based on the historical service data of a group of drill pipes of the same type, and to deduce the potential damage evolution path of the current drill pipe. The evaluation output module is used to determine and output dynamic evaluation values ​​of drill pipe service risk based on potential damage evolution paths.

[0044] All calculations involved in the embodiments are dimensionless numerical calculations, and the preset parameters and thresholds in the calculations are set by those skilled in the art according to the actual situation.

[0045] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented, in whole or in part, in the form of a computer program product.

[0046] Those skilled in the art will recognize that the modules and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and inventive constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0047] In addition, the functional modules in the various embodiments of this application can be integrated into one processing module, or each module can exist physically separately, or two or more modules can be integrated into one module.

[0048] In the several embodiments provided in this application, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of modules is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple modules or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or modules may be electrical, mechanical, or other forms.

[0049] The above are merely specific embodiments of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

[0050] In conclusion, the above are merely preferred embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for assessing the service risk of drill pipes based on multi-source data streams, characterized in that, Includes the following steps: S1. Acquire multi-source sensor data of the drill pipe during its service life, and reconstruct the time-domain load data of the drill pipe based on the multi-source sensor data; S2. Convert the load time domain data into a load spectrum, and determine whether the drill pipe is under a load state that is prone to inducing high cycle fatigue based on the load spectrum; S3. When it is determined that the load is in a load state that is prone to inducing high cycle fatigue, analyze the load driving factors related to the trend change of the micro-damage state of the driving drill pipe material from the load spectrum. S4. Compare and analyze the statistical characteristics of the load driving factor of the current drill pipe with the historical service data of the same type of drill pipe group to identify whether the current drill pipe has an abnormal response pattern that deviates significantly from the group pattern. S5. Match the load driving factor and abnormal response mode of the current drill pipe with the damage evolution path library constructed based on the historical service data of the same type of drill pipe group to deduce the potential damage evolution path of the current drill pipe. S6. Based on the potential damage evolution path, determine and output the dynamic assessment value of the drill pipe service risk.

2. The drill pipe service risk assessment method based on multi-source data stream according to claim 1, characterized in that, Acquire multi-source sensor data of the drill pipe during its service life, and reconstruct the time-domain load data of the drill pipe based on the multi-source sensor data, including: Acquire sensor data including torque, axial force, triaxial vibration, and rotational speed; Obtain drilling operating parameters, including drilling pressure, drilling fluid density, and displacement. Based on the drill string dynamics model, sensor data and drilling condition parameters are used as inputs to calculate the time-domain data of the drill string load. The time-domain data of the load includes the time history of alternating bending stress and the time history of alternating axial stress.

3. The drill pipe service risk assessment method based on multi-source data stream according to claim 1, characterized in that, The load time-domain data is converted into a load spectrum, and the drill pipe is judged based on the load spectrum to determine whether it is under a load state that is prone to inducing high-cycle fatigue, including: Rainflow counting was performed on the alternating bending stress time history and alternating axial stress time history to obtain the load spectrum characterized by stress amplitude and mean value; Extract all load cycles whose stress amplitude exceeds a preset stress amplitude threshold from the load spectrum, and calculate the proportion of load cycles whose stress amplitude exceeds the preset stress amplitude threshold in the total number of cycles; If the ratio continues to exceed the preset ratio threshold, the drill pipe is determined to be under a load state that is prone to inducing high-cycle fatigue.

4. The drill pipe service risk assessment method based on multi-source data stream according to claim 1, characterized in that, When the load condition is determined to be prone to inducing high-cycle fatigue, the load driving factors related to the trend of micro-damage changes in the driving drill pipe material are analyzed from the load spectrum, including: Based on the stress amplitude and mean in the load spectrum, the equivalent bending stress amplitude and equivalent axial stress amplitude are calculated by modifying the Goodman equation; For the equivalent bending stress amplitude and the equivalent axial stress amplitude, calculate their mean and coefficient of variation over the continuous monitoring period, respectively. The mean value of the equivalent bending stress amplitude, the coefficient of variation of the equivalent bending stress amplitude, the mean value of the equivalent axial stress amplitude, and the coefficient of variation of the equivalent axial stress amplitude are used as load driving factors.

5. The drill pipe service risk assessment method based on multi-source data stream according to claim 4, characterized in that, The equivalent bending stress amplitude and equivalent axial stress amplitude are calculated by modifying the Goodman equation, including: Based on the stress amplitude and mean, and combined with the tensile strength and fatigue limit of the drill pipe material, the equivalent stress amplitude under bending load and axial load is calculated respectively. The equivalent stress amplitude is used to characterize the stress level equivalent to symmetrical cyclic fatigue damage under the influence of mean stress.

6. The drill pipe service risk assessment method based on multi-source data stream according to claim 1, characterized in that, By comparing and analyzing the statistical characteristics of the current drill pipe's load driving factors with historical service data of a group of drill pipes of the same type, we can identify whether the current drill pipe exhibits any abnormal response patterns that significantly deviate from the group's patterns, including: Based on historical service data of the same type of drill pipe group, a reference distribution of the load driving factor vector is established, which consists of the mean of the equivalent bending stress amplitude, the coefficient of variation of the equivalent bending stress amplitude, the mean of the equivalent axial stress amplitude, and the coefficient of variation of the equivalent axial stress amplitude. Calculate the Mahalanobis distance between the current drill pipe load driving factor vector and the reference distribution; If the Mahalanobis distance exceeds the preset anomaly detection threshold, then an abnormal response pattern that significantly deviates from the group pattern is identified in the current drill pipe.

7. The drill pipe service risk assessment method based on multi-source data stream according to claim 1, characterized in that, By matching the current drill pipe's load driving factors and abnormal response patterns with a damage evolution path database constructed based on historical service data of a group of drill pipes of the same type, the potential damage evolution paths of the current drill pipe are deduced, including: The historical service data of the same type of drill pipe group are organized into a time window sequence for each complete record from initial service to damage or decommissioning. Each time window contains a load driving factor vector consisting of the mean of the equivalent bending stress amplitude, the coefficient of variation of the equivalent bending stress amplitude, the mean of the equivalent axial stress amplitude, and the coefficient of variation of the equivalent axial stress amplitude, thereby constructing a damage evolution path library. Calculate the dynamic time warping distance between the load driving factor vector sequence of the latest consecutive time windows of the current drill pipe and the vector sequences of each path in the damage evolution path library at the same time stage; Based on the dynamic time warping distance, several historical damage evolution paths with the highest similarity are selected as potential damage evolution paths for the current drill pipe.

8. The drill pipe service risk assessment method based on multi-source data stream according to claim 7, characterized in that, When constructing the damage evolution path library, the damage status in the complete record of damage or decommissioning is quantified and calibrated by matching the changing trend of the continuous load driving factor vector of the drill pipe within the corresponding time window with the preset damage evolution mode.

9. The drill pipe service risk assessment method based on multi-source data stream according to claim 1, characterized in that, Based on the potential damage evolution path, a dynamic assessment value of the drill pipe service risk is determined and output, including: For each potential damage evolution path, extract the number of remaining time windows experienced from the time window that matches the current drill pipe state until the path terminates. Based on the statistical characteristics of the number of remaining time windows corresponding to all potential damage evolution paths, an assessment value characterizing the remaining safe service life of the drill pipe is determined. The assessment value is output as a dynamic assessment value of the service risk of the drill pipe.

10. A drill pipe service risk assessment system based on multi-source data streams, used to implement the drill pipe service risk assessment method based on multi-source data streams as described in any one of claims 1-9, characterized in that, Includes the following modules: The data acquisition module is used to acquire multi-source sensor data of the drill pipe during its service life, and to reconstruct the time-domain load data of the drill pipe based on the multi-source sensor data. The load spectrum analysis module is used to convert load time-domain data into a load spectrum and determine whether the drill pipe is under a load state that is prone to inducing high-cycle fatigue based on the load spectrum. The factor extraction module is used to analyze the load driving factors related to the trend changes in the micro-damage state of the driving drill pipe material from the load spectrum when it is determined that the load state is prone to inducing high-cycle fatigue. The comparison and identification module is used to compare and analyze the statistical characteristics of the load driving factors of the current drill pipe with the historical service data of the same type of drill pipe group to identify whether the current drill pipe has an abnormal response pattern that deviates significantly from the group pattern. The matching and deduction module is used to match the load driving factors and abnormal response patterns of the current drill pipe with the damage evolution path library built based on the historical service data of a group of drill pipes of the same type, and to deduce the potential damage evolution path of the current drill pipe. The evaluation output module is used to determine and output dynamic evaluation values ​​of drill pipe service risk based on potential damage evolution paths.

Images