Analysis and visualization system and method based on shaft monitoring data

Abstract

The invention belongs to the technical field of industrial data analysis, and discloses an analysis and visualization system and method based on wellbore monitoring data, and the system comprises a data collection and processing module which collects multi-source wellbore monitoring data in the wellbore operation process, carries out the preprocessing of the multi-source wellbore monitoring data, and generates wellbore monitoring feature data in a uniform format; the event disturbance evaluation module is used for carrying out abnormal event detection on the wellbore monitoring characteristic data through a sliding window statistical analysis method, organizing detected abnormal events into a hierarchical event matrix of wellbore anomalies, carrying out priority evaluation on disturbance factors of the abnormal events and obtaining disturbance priorities of the abnormal events; the causal gain fusion module is used for only calculating a closed-loop path between adjacent horizon positions of the shaft by adopting a finite loop causal gain evaluation method based on the hierarchical event matrix and the disturbance priority of the abnormal events, and outputting a closed-loop abnormal gain matrix; and real-time detection, causal tracking and visual display of shaft operation abnormity are realized.

Description

Technical Field

[0001] This invention belongs to the field of industrial data analysis technology, specifically relating to an analysis and visualization system based on wellbore monitoring data, and also to a method for the analysis and visualization system based on wellbore monitoring data. Background Technology

[0002] Existing wellbore monitoring data analysis and visualization systems and methods mainly suffer from the following problems: As a critical channel in oil and gas extraction and mining production, the wellbore's operational status directly impacts production safety and stability. With the development of sensors and monitoring technologies, a large amount of multi-source monitoring data is generated during wellbore operation, providing an important foundation for the analysis and early warning of wellbore anomalies. Existing wellbore monitoring data analysis and visualization methods typically rely on single-point anomaly detection or anomaly identification based on simple correlations. While these methods can detect local anomalies to some extent, they still have the following shortcomings: Traditional methods often perform static analysis on single-point or local anomalies, making it difficult to reflect the transmission path and diffusion pattern of anomalies between different layers and measuring points within the wellbore, resulting in the anomaly propagation mechanism not being effectively revealed. When dealing with anomaly propagation, existing methods often fail to combine the wellbore's layered structural constraints with the temporal sequence control of anomaly generation, easily leading to false or redundant anomaly correlations and affecting the accuracy of the propagation chain.

[0003] Anomalies are typically caused by the combined effects of multiple disturbances. However, traditional methods lack standardized mechanisms for calculating and ranking causal contributions, making it difficult to effectively identify primary disturbance sources and thus impacting the targeted nature of anomaly localization and response. Node layers may exhibit varying responsiveness and importance at different times. Existing methods, often based on static or average indicators, struggle to dynamically reflect the true intensity of a node's influence on anomaly propagation. Loop or closed-loop coupling relationships often exist between anomalies, potentially amplifying or feeding back into the propagation process. Traditional analysis methods often overlook this characteristic, leading to insufficient accuracy in anomaly source localization. Anomaly sources are typically sparsely distributed within wellbore layers. Existing methods fail to fully utilize this prior information, resulting in low anomaly source identification accuracy, high false alarm and false negative rates, and an inability to meet practical monitoring and early warning needs.

[0004] In view of this, the present invention proposes an analysis and visualization system and method based on wellbore monitoring data to solve the above problems. Summary of the Invention

[0005] The purpose of this invention is to provide an analysis and visualization system based on wellbore monitoring data, which solves the problem of insufficient accuracy in locating anomalies in existing wellbore monitoring and analysis.

[0006] Another objective of this invention is to provide a method for an analysis and visualization system based on wellbore monitoring data.

[0007] The technical solution adopted in this invention is a wellbore monitoring data analysis and visualization system, comprising: The data acquisition and processing module collects multi-source wellbore monitoring data during wellbore operation and preprocesses the multi-source wellbore monitoring data to generate wellbore monitoring feature data in a uniform format. The event disturbance assessment module uses a sliding window statistical analysis method to detect abnormal events in wellbore monitoring feature data. It organizes the detected abnormal events into a hierarchical event matrix of wellbore anomalies and prioritizes the disturbance factors of the abnormal events to obtain the disturbance priority of the abnormal events. The causal gain fusion module, based on the hierarchical event matrix and the perturbation priority of abnormal events, adopts the finite loop causal gain evaluation method, calculates only the closed loop path between adjacent layers in the wellbore, and outputs the closed loop abnormal gain matrix. The anomaly source localization module receives the closed-loop anomaly gain matrix, solves for the most likely anomaly source vector using a sparse optimization model, evaluates the uncertainty of the anomaly source vector using a multi-scale Monte Carlo method, and outputs the anomaly source localization result. The anomaly propagation visualization module determines the propagation path and intensity of anomaly events based on the closed-loop anomaly gain matrix and anomaly source localization results, and then visualizes and displays them.

[0008] The invention is further characterized in that, Multi-source wellbore monitoring data includes pressure data, temperature data, flow rate data, vibration and acoustic data, tubing stress data, and fluid physical property parameters.

[0009] Methods for generating wellbore monitoring feature data with a standardized format include: An adaptive filtering method is used to denoise the multi-source wellbore monitoring data. Linear normalization is used to unify the dimensions of the denoised multi-source wellbore monitoring data, mapping different types of parameters to a unified numerical scale. The normalized multi-source monitoring data is processed for time synchronization. An interpolation method is used to align the multi-source monitoring data with different sampling frequencies according to a unified time reference. The denoised, normalized and time-aligned multi-source monitoring data are then integrated to generate wellbore monitoring feature data with a unified format.

[0010] Methods for organizing detected anomalous events into a hierarchical event matrix of wellbore anomalies include: The wellbore monitoring feature data with a uniform format is divided into sliding windows according to the time series. Each sliding window contains continuous sampling points of a fixed length. The length of the sliding window and the sliding step size are set according to the sampling frequency of the multi-source wellbore monitoring data. For wellbore monitoring characteristic data within each sliding window, abnormal indicators are calculated and statistically analyzed. These abnormal indicators include mean deviation, variance, peak value, and energy density change. For each abnormal indicator, a corresponding abnormal threshold is set to judge the abnormal indicators within the sliding window. The abnormal thresholds include preset mean deviation threshold, preset variance threshold, preset peak value threshold, and preset energy density change threshold. Windows that meet the corresponding abnormal thresholds are marked as abnormal events. The detected abnormal events are organized into a hierarchical event matrix according to the wellbore measuring points and time sequence. The rows of the matrix represent different measuring points, and the columns represent different time windows.

[0011] Methods for obtaining the perturbation priority of abnormal events include: For each abnormal event, the disturbance factors of the abnormal event are extracted based on the multi-source wellbore monitoring characteristic data. The disturbance factors include pressure fluctuations, temperature changes, flow deviations, vibration anomalies, tubing stress anomalies, and changes in fluid physical parameters. For each disturbance factor, quantitative indicators are calculated, including the rate of pressure change, the rate of temperature rise and fall, the proportion of flow rate deviating from the preset reference flow rate, the abnormal amplitude of the frequency of downhole equipment vibration signal, the extent to which the tubing stress exceeds the preset tubing stress threshold, and the degree of change of fluid physical properties such as density, viscosity, gas content, and sand content. The quantitative indicators of each disturbance factor are combined with the preset weighting factors of each disturbance factor to calculate the disturbance contribution of each disturbance factor; the contribution of each disturbance factor is standardized to obtain the priority score of each disturbance factor; based on the priority score, all disturbance factors are ranked to form the disturbance priority of the abnormal event.

[0012] Methods for obtaining the closed-loop abnormal gain matrix include: The abnormal events in the hierarchical event matrix are arranged into an abnormal event node set according to the measurement point and time window order. Each abnormal event is a node. For node pairs that satisfy the adjacent layer constraints and whose time window difference does not exceed the preset maximum lag window, a directed candidate edge is generated and the corresponding time lag is recorded. Assign causal weights to each directed candidate edge. The causal weights are adjusted according to the perturbation priority of abnormal events. Form a directed graph by all nodes and directed candidate edges. In the directed graph, starting from each node, enumerate closed loop paths with a length not exceeding the preset maximum loop length. Only retain closed loop paths of adjacent layers to form a finite set of closed loop paths. For each closed-loop path in the finite set of closed loops, calculate the closed-loop causal gain. Traverse all closed-loop paths. For each pair of nodes involved in a closed-loop path, accumulate the closed-loop causal gain of that closed-loop path to the corresponding row and column elements of the closed-loop abnormal gain matrix. By accumulating all closed-loop paths, a complete closed-loop abnormal gain matrix is ​​formed.

[0013] Methods for finding the most likely anomaly source vector include: Receive the closed-loop anomaly gain matrix and model the propagation process of the anomaly event from the anomaly source node to the global measurement points; normalize the closed-loop anomaly gain matrix by row or column to obtain the anomaly intensity distribution vector of each measurement point, establish an anomaly diffusion model, and regard the observed anomaly intensity distribution vector of the measurement points as the result of the anomaly source vector after diffusion. Based on the anomaly diffusion model, the problem of anomaly source vector localization is transformed into an optimization problem with sparse prior constraints. A sparse optimization model is constructed, and the error between the anomaly intensity distribution vector of the measurement point and the anomaly intensity distribution vector predicted by the sparse optimization model is minimized. The effect of the diffusion operator is calculated by an approximate expansion method, and the optimal solution is obtained by using sparse constraints. The optimal solution obtained is the most likely anomaly source vector.

[0014] Methods for outputting anomaly source localization results include: The multi-scale Monte Carlo method is used for uncertainty assessment. Based on the anomaly source vector, random perturbation is introduced to generate perturbation samples at different scales. The anomaly source vector is repeatedly calculated, and statistical analysis is performed on the perturbation samples to obtain the spatial probability distribution of the anomaly source at each measuring point, and the anomaly source localization result is output.

[0015] Methods for determining the propagation path and intensity of abnormal events and visualizing them include: Receive the closed-loop anomaly gain matrix and anomaly source location results, and starting from each anomaly source node, construct the propagation path of the anomaly event along the wellbore measuring points based on the non-zero elements of the closed-loop anomaly gain matrix. Along each defined propagation path, the propagation intensity on each propagation path is calculated by combining the causal weights of the closed-loop anomaly gain matrix. The propagation intensity is obtained by accumulating the causal weights of each edge along the propagation path. The anomaly source node is represented by the color depth of the node, and the thickness and transparency of the edge represent the anomaly propagation intensity. The propagation path and propagation intensity of the anomaly event are dynamically visualized.

[0016] Another technical solution adopted in this invention is to provide a method for analyzing and visualizing wellbore monitoring data, the steps of which are as follows: S1. Collect multi-source wellbore monitoring data during wellbore operation, and preprocess the multi-source wellbore monitoring data to generate wellbore monitoring feature data with a unified format; S2. Anomaly detection is performed on wellbore monitoring feature data using a sliding window statistical analysis method. The detected anomalies are organized into a hierarchical event matrix of wellbore anomalies, and the perturbation factors of the anomalies are prioritized to obtain the perturbation priority of the anomalies. S3. Based on the hierarchical event matrix and the disturbance priority of abnormal events, the finite loop causal gain evaluation method is adopted to calculate only the closed loop path between adjacent layers in the wellbore and output the closed loop abnormal gain matrix. S4. Receive the closed-loop anomaly gain matrix, combine it with the sparse optimization model to solve for the most likely anomaly source vector, and use the multi-scale Monte Carlo method to evaluate the uncertainty of the anomaly source vector, and output the anomaly source localization result. S5. Based on the closed-loop anomaly gain matrix and the anomaly source localization results, determine the propagation path and propagation intensity of the anomaly event and visualize it.

[0017] Compared with the prior art, the present invention has the following beneficial effects: By forming node sets according to measurement points and time windows, and allowing propagation only between adjacent layers, a reasonable anomaly path structure is ensured, eliminating false connections in cross-layer propagation. A maximum hysteresis window limit is set to control the time delay of anomaly propagation and ensure a reasonable time sequence. Causal weights are weighted and adjusted using a disturbance priority score, and the disturbance response function is calculated in conjunction with node priority, ensuring that the anomaly propagation intensity of each edge reflects the importance of the node. The node priority score, through standardization, clearly distinguishes the contribution of each anomaly factor, providing a basis for anomaly source identification. Enumerating finite closed-loop paths, calculating closed-loop causal gains, and accumulating them into the closed-loop anomaly gain matrix enables a comprehensive evaluation of loop coupling effects. The closed-loop gain matrix allows for more accurate identification of anomaly sources, analysis of anomaly propagation paths and intensity, and provides a scientific basis for wellbore anomaly diagnosis and prediction.

[0018] By receiving the closed-loop anomaly gain matrix, the propagation process of anomaly events from the anomaly source node to all measurement points is modeled. The closed-loop anomaly gain matrix is ​​then normalized to obtain the anomaly intensity distribution vector for each measurement point, establishing an anomaly diffusion model. The anomaly observations at the measurement points are considered as the result of the diffusion effect of the anomaly source, thus quantifying the propagation characteristics of the anomaly event. Furthermore, the anomaly source localization problem is transformed into an optimization problem with sparse prior constraints. A sparse optimization model is constructed, minimizing the error between the predicted and observed anomaly intensity values ​​at the measurement points. Simultaneously, sparse constraints are introduced to ensure that the anomaly source for most measurement points is zero, retaining only the most probable anomaly source, thus ensuring high accuracy in anomaly source localization. The diffusion operator effect is calculated using an approximate expansion method, effectively handling the complexity of matrix exponential operations and improving solution efficiency. Attached Figure Description

[0019] Figure 1 This is a schematic diagram of the structure of the wellbore monitoring data analysis and visualization system of the present invention; Figure 2 This is a schematic diagram of the analysis and visualization method based on wellbore monitoring data according to the present invention. Detailed Implementation

[0020] 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 skilled in the art without creative effort are within the scope of protection of the present invention.

[0021] Example 1 The wellbore monitoring data analysis and visualization system of the present invention has the following structure: Figure 1 As shown, it includes: The data acquisition and processing module collects multi-source wellbore monitoring data during wellbore operation and preprocesses the multi-source wellbore monitoring data to generate wellbore monitoring feature data in a uniform format. The event disturbance assessment module uses a sliding window statistical analysis method to detect abnormal events in wellbore monitoring feature data. It organizes the detected abnormal events into a hierarchical event matrix of wellbore anomalies and prioritizes the disturbance factors of the abnormal events to obtain the disturbance priority of the abnormal events. The causal gain fusion module, based on the hierarchical event matrix and the perturbation priority of abnormal events, adopts the finite loop causal gain evaluation method, calculates only the closed loop path between adjacent layers in the wellbore, and outputs the closed loop abnormal gain matrix. The anomaly source localization module receives the closed-loop anomaly gain matrix, solves for the most likely anomaly source vector using a sparse optimization model, evaluates the uncertainty of the anomaly source vector using a multi-scale Monte Carlo method, and outputs the anomaly source localization result. The anomaly propagation visualization module determines the propagation path and intensity of anomaly events based on the closed-loop anomaly gain matrix and anomaly source localization results, and then visualizes and displays them.

[0022] Example 2 Based on Example 1, the multi-source wellbore monitoring data includes pressure data, temperature data, flow rate data, vibration acoustic data, tubing stress data, and fluid physical property parameters.

[0023] Methods for generating wellbore monitoring feature data with a standardized format include: An adaptive filtering method is used to denoise the multi-source wellbore monitoring data. Linear normalization is used to unify the dimensions of the denoised multi-source wellbore monitoring data, mapping different types of parameters to a unified numerical scale. The normalized multi-source monitoring data is processed for time synchronization. An interpolation method is used to align the multi-source monitoring data with different sampling frequencies according to a unified time reference. The denoised, normalized and time-aligned multi-source monitoring data are then integrated to generate wellbore monitoring feature data with a unified format.

[0024] Example 3 Based on Example 2, the method for organizing detected abnormal events into a hierarchical event matrix of wellbore anomalies includes: The wellbore monitoring feature data with a uniform format is divided into sliding windows according to the time series. Each sliding window contains continuous sampling points of a fixed length. The length of the sliding window and the sliding step size are set according to the sampling frequency of the multi-source wellbore monitoring data. For wellbore monitoring characteristic data within each sliding window, abnormal indicators are calculated and statistically analyzed. These abnormal indicators include mean deviation, variance, peak value, and energy density change. For each abnormal indicator, a corresponding abnormal threshold is set to judge the abnormal indicators within the sliding window. The abnormal thresholds include preset mean deviation threshold, preset variance threshold, preset peak value threshold, and preset energy density change threshold. Windows that meet the corresponding abnormal thresholds are marked as abnormal events. The detected abnormal events are organized into a hierarchical event matrix according to the wellbore measuring points and time sequence. The rows of the matrix represent different measuring points, and the columns represent different time windows.

[0025] Example 4 Based on Example 3, the method for obtaining the disturbance priority of abnormal events includes: For each abnormal event, the disturbance factors of the abnormal event are extracted based on the multi-source wellbore monitoring characteristic data. The disturbance factors include pressure fluctuations, temperature changes, flow deviations, vibration anomalies, tubing stress anomalies, and changes in fluid physical parameters. For each disturbance factor, quantitative indicators are calculated, including the rate of pressure change, the rate of temperature rise and fall, the proportion of flow rate deviating from the preset reference flow rate, the abnormal amplitude of the frequency of downhole equipment vibration signal, the extent to which the tubing stress exceeds the preset tubing stress threshold, and the degree of change of fluid physical properties such as density, viscosity, gas content, and sand content. The quantitative indicators of each disturbance factor are combined with the preset weighting factors of each disturbance factor to calculate the disturbance contribution of each disturbance factor; the contribution of each disturbance factor is standardized to obtain the priority score of each disturbance factor; based on the priority score, all disturbance factors are ranked to form the disturbance priority of the abnormal event.

[0026] Example 5 Based on Example 4, the method for obtaining the closed-loop abnormal gain matrix includes: The abnormal events in the hierarchical event matrix are arranged into an abnormal event node set according to the measurement point and time window order. Each abnormal event is a node. For node pairs that satisfy the adjacent layer constraints and whose time window difference does not exceed the preset maximum lag window, a directed candidate edge is generated and the corresponding time lag is recorded. Assign causal weights to each directed candidate edge. The causal weights are adjusted according to the perturbation priority of abnormal events. Form a directed graph by all nodes and directed candidate edges. In the directed graph, starting from each node, enumerate closed loop paths with a length not exceeding the preset maximum loop length. Only retain closed loop paths of adjacent layers to form a finite set of closed loop paths. For each closed-loop path in the finite set of closed loops, calculate the closed-loop causal gain. Traverse all closed-loop paths. For each pair of nodes involved in a closed-loop path, accumulate the closed-loop causal gain of that closed-loop path to the corresponding row and column elements of the closed-loop abnormal gain matrix. By accumulating all closed-loop paths, a complete closed-loop abnormal gain matrix is ​​formed.

[0027] Example 6 Based on Example 5, the methods for solving the most likely anomaly source vector include: Receive the closed-loop anomaly gain matrix and model the propagation process of the anomaly event from the anomaly source node to the global measurement points; normalize the closed-loop anomaly gain matrix by row or column to obtain the anomaly intensity distribution vector of each measurement point, establish an anomaly diffusion model, and regard the observed anomaly intensity distribution vector of the measurement points as the result of the anomaly source vector after diffusion. Based on the anomaly diffusion model, the problem of anomaly source vector localization is transformed into an optimization problem with sparse prior constraints. A sparse optimization model is constructed, and the error between the anomaly intensity distribution vector of the measurement point and the anomaly intensity distribution vector predicted by the sparse optimization model is minimized. The effect of the diffusion operator is calculated by an approximate expansion method, and the optimal solution is obtained by using sparse constraints. The optimal solution obtained is the most likely anomaly source vector.

[0028] Methods for outputting anomaly source localization results include: The multi-scale Monte Carlo method is used for uncertainty assessment. Based on the anomaly source vector, random perturbation is introduced to generate perturbation samples at different scales. The anomaly source vector is repeatedly calculated, and statistical analysis is performed on the perturbation samples to obtain the spatial probability distribution of the anomaly source at each measuring point, and the anomaly source localization result is output.

[0029] Methods for determining the propagation path and intensity of abnormal events and visualizing them include: Receive the closed-loop anomaly gain matrix and anomaly source location results, and starting from each anomaly source node, construct the propagation path of the anomaly event along the wellbore measuring points based on the non-zero elements of the closed-loop anomaly gain matrix. Along each defined propagation path, the propagation intensity on each propagation path is calculated by combining the causal weights of the closed-loop anomaly gain matrix. The propagation intensity is obtained by accumulating the causal weights of each edge along the propagation path. The anomaly source node is represented by the color depth of the node, and the thickness and transparency of the edge represent the anomaly propagation intensity. The propagation path and propagation intensity of the anomaly event are dynamically visualized.

[0030] Example 7 Please see Figure 1 As shown, this embodiment provides an analysis and visualization system based on wellbore monitoring data, specifically including the following steps: The data acquisition and processing module collects multi-source wellbore monitoring data during wellbore operation and preprocesses the multi-source wellbore monitoring data to generate wellbore monitoring feature data in a uniform format. The event disturbance assessment module uses a sliding window statistical analysis method to detect abnormal events in wellbore monitoring feature data. It organizes the detected abnormal events into a hierarchical event matrix of wellbore anomalies and prioritizes the disturbance factors of the abnormal events to obtain the disturbance priority of the abnormal events. The causal gain fusion module, based on the hierarchical event matrix and the perturbation priority of abnormal events, adopts the finite loop causal gain evaluation method, calculates only the closed loop path between adjacent layers in the wellbore, and outputs the closed loop abnormal gain matrix. The anomaly source localization module receives the closed-loop anomaly gain matrix, solves for the most likely anomaly source vector using a sparse optimization model, evaluates the uncertainty of the anomaly source vector using a multi-scale Monte Carlo method, and outputs the anomaly source localization result. The anomaly propagation visualization module determines the propagation path and intensity of anomaly events based on the closed-loop anomaly gain matrix and anomaly source localization results, and then visualizes and displays them.

[0031] Multi-source wellbore monitoring data includes pressure data, temperature data, flow rate data, vibration and acoustic data, tubing stress data, and fluid physical property parameters.

[0032] The pressure data is collected by pressure sensors at monitoring points at different depths within the wellbore, including fluid pressure, casing pressure, oil layer pressure, and gas layer pressure at each monitoring point within the wellbore. This data reflects the pressure distribution of fluids in different layers within the wellbore. The pressure data acquisition frequency can be set to 1–10 Hz to capture short-cycle pressure fluctuations. During steady-state production, a low-frequency sampling mode of 0.1–1 Hz can also be used to reduce communication load. Temperature data is collected through wellhead temperature sensors and downhole distributed temperature sensors to continuously obtain temperature changes at different depths in the wellbore, specifically including wellhead temperature and temperature at each downhole monitoring point, to reflect the temperature gradient changes along the depth direction of the wellbore; the temperature data acquisition frequency can be set to 0.1 to 1 Hz. For operating conditions that require analysis of temperature abrupt changes or gas-liquid interface changes, the acquisition frequency can be temporarily increased to 1 to 5 Hz for high-frequency sampling. Flow data is acquired through flow meters installed at the wellhead or downhole, including instantaneous flow velocity, cumulative flow, and flow rate of change of liquid or gas, used to characterize wellbore production capacity and production stability; the acquisition frequency of instantaneous flow velocity can be set to 0.5 to 5 Hz, and the cumulative flow is periodically accumulated and updated according to time windows, such as once every 1 minute or 5 minutes; the flow rate of change can be calculated by the flow difference between adjacent time windows. Vibration acoustic data is acquired by vibration sensors (such as triaxial accelerometers) and acoustic sensors installed in downhole equipment, tubing, and pumps. The acquisition frequency is usually set to 1kHz to 10kHz. Vibration acoustic data specifically includes downhole equipment vibration signals, tubing vibration signals, pump vibration signals, and wellbore acoustic signals. It is used to reflect the operating status of mechanical equipment and the acoustic characteristics caused by fluid flow in the wellbore, thereby identifying abnormal operating conditions, tubing collisions, or gas locks and other fault signs. The stress data of the tubing includes axial tensile force, torque and bending stress of the tubing; strain gauges or distributed fiber optic strain sensors are deployed on the surface of the tubing to collect changes in axial tensile strain, torsional load and bending stress in real time, and convert them into corresponding force values ​​through a mechanical model; the tubing stress data acquisition frequency can be set to 1 to 50 Hz to capture transient force changes caused by pump start-up and shutdown, impact load or jamming, etc. Fluid physical properties include density, viscosity, gas content, and sand content. Online sampling and physical property analysis devices are installed at the wellhead to monitor the fluid's density, viscosity, gas content, and sand content in real time. The sampling period for fluid physical properties can be set from 20 seconds to 5 minutes, and can be dynamically adjusted according to production conditions. During periods of fluctuating operating conditions, the sampling frequency can be increased to enhance monitoring sensitivity.

[0033] Methods for generating wellbore monitoring feature data with a standardized format include: An adaptive filtering method is used to denoise the multi-source wellbore monitoring data. Linear normalization is used to unify the dimensions of the denoised multi-source wellbore monitoring data, mapping different types of parameters to a unified numerical scale. The normalized multi-source monitoring data is processed for time synchronization. An interpolation method is used to align the multi-source monitoring data with different sampling frequencies according to a unified time reference. The denoised, normalized and time-aligned multi-source monitoring data are then integrated to generate wellbore monitoring feature data with a unified format.

[0034] Methods for organizing detected anomalous events into a hierarchical event matrix of wellbore anomalies include: The wellbore monitoring feature data with a uniform format is divided into sliding windows according to the time series. Each sliding window contains continuous sampling points of a fixed length. The length of the sliding window and the sliding step size are set according to the sampling frequency of the multi-source wellbore monitoring data. For wellbore monitoring characteristic data within each sliding window, abnormal indicators are calculated and statistically analyzed. These abnormal indicators include mean deviation, variance, peak value, and energy density change. For each abnormal indicator, a corresponding abnormal threshold is set to judge the abnormal indicators within the sliding window. The abnormal thresholds include preset mean deviation threshold, preset variance threshold, preset peak value threshold, and preset energy density change threshold. Windows that meet the corresponding abnormal thresholds are marked as abnormal events. The detected abnormal events are organized into a hierarchical event matrix according to the wellbore measuring points and time sequence. The rows of the matrix represent different measuring points, and the columns represent different time windows.

[0035] Methods for obtaining the perturbation priority of abnormal events include: For each abnormal event, the disturbance factors of the abnormal event are extracted based on the multi-source wellbore monitoring characteristic data. The disturbance factors include pressure fluctuations, temperature changes, flow deviations, vibration anomalies, tubing stress anomalies, and changes in fluid physical parameters. For each disturbance factor, quantitative indicators are calculated, including the rate of pressure change, the rate of temperature rise and fall, the proportion of flow rate deviating from the preset reference flow rate, the abnormal amplitude of the frequency of downhole equipment vibration signal, the extent to which the tubing stress exceeds the preset tubing stress threshold, and the degree of change of fluid physical properties such as density, viscosity, gas content, and sand content. By combining the quantitative indicators of each disturbance factor with its preset weighting factor, the disturbance contribution of each disturbance factor is calculated; the disturbance contribution of each disturbance factor is as follows: ;in, Indicates the first The contribution of each disturbance factor to the disturbance; Indicates the first Quantitative indicators of each disturbance factor; Indicates the preset number Preset weighting factors for each disturbance factor; An index representing the disturbance factors; The contribution of each disturbance factor is standardized to obtain a priority score for each disturbance factor. Based on the priority score, all disturbance factors are ranked to form the disturbance priority of the abnormal event.

[0036] The priority scores for each disturbance factor are as follows: ;in, Indicates the first Priority rating of each disturbance factor; This represents the sum of the contributions of all disturbance factors; Indicates the first The contribution of each disturbance factor; This represents the index used to iterate through all perturbation factors in the abnormal event during summation.

[0037] Methods for obtaining the closed-loop abnormal gain matrix include: The abnormal events in the hierarchical event matrix are arranged into an abnormal event node set according to the measurement point and time window order. Each abnormal event is a node. For node pairs that satisfy the adjacent layer constraints and whose time window difference does not exceed the preset maximum lag window, a directed candidate edge is generated and the corresponding time lag is recorded. The wellbore is divided into multiple layers. The adjacent layer constraint means that nodes connected to each other in the same abnormal event propagation path must be located in adjacent layers. For example, if the wellbore has layers 1, 2, and 3, an abnormal event of a node in layer 1 can only point to a node in layer 2, and cannot directly point to layer 3 or further layers. Each node corresponds to a time window (sliding window). The time window difference cannot exceed the preset maximum lag window, which means that the time interval between the starting node and the target node of the candidate edge cannot exceed a threshold. For example, the maximum lag window is set to 3 sampling windows. If the starting node is in time window 5, it can only point to a node in time window 6, 7, or 8, and cannot point to a node in time window 9 or later.

[0038] It should be noted that: based on the actual locations of the sensors deployed downhole, the wellbore is discretized into layers, and the monitoring points corresponding to each sensor are defined as layer units within the wellbore. Specifically: the installation depth of various sensors within the wellbore is used as the basis for layer division. When different types of sensors are set at the same depth, this depth is defined as the same layer, and the multi-source monitoring data collected by different types of sensors at this depth is fused to form a state description of the current layer. When different types of sensors differ in depth, they are numbered as different layers, thereby constructing a discretized set of layer measurement points. The number of wellbore layers is determined by the number of monitoring points with effective depth locations, denoted as N layers. The layers are sorted and numbered in order from the wellhead to the bottom of the well, forming an ordered layer sequence L1, L2, ..., LN. After the layer sorting and numbering are completed, adjacent layers are defined by the numbering relationship in the ordered layer sequence. When two layers have adjacent numbers in the ordered layer sequence, they are determined to be adjacent layers.

[0039] Assign causal weights to each directed candidate edge. The causal weights are adjusted according to the perturbation priority of abnormal events. Form a directed graph by all nodes and directed candidate edges. In the directed graph, starting from each node, enumerate closed loop paths with a length not exceeding the preset maximum loop length. Only retain closed loop paths of adjacent layers to form a finite set of closed loop paths. For each closed-loop path in the finite set of closed loops, calculate the closed-loop causal gain. Traverse all closed-loop paths. For each pair of nodes involved in a closed-loop path, accumulate the closed-loop causal gain of that closed-loop path to the corresponding row and column elements of the closed-loop abnormal gain matrix. By accumulating all closed-loop paths, a complete closed-loop abnormal gain matrix is ​​formed.

[0040] Closed-loop causal gain of the closed-loop path: ;in, Indicates the closed-loop path At the point of time Closed-loop causal gain; Indicates the index of the closed-loop path; Indicates the closed-loop path A directed edge from node (Starting node) to node (Target node); This represents the causal weight of a directed edge; Representing an edge The time lag indicates that the abnormal event originates from the node. propagation to nodes Required delay time; This indicates a lag in the preset reference time; This represents the adjustment coefficient, used to balance the contribution of causal gain to the disturbance response; The disturbance response function is represented by the time point. At that time, node For nodes The degree of disturbance response, through the node and nodes The average of the sum of priority scores is obtained; The propagation strength of an edge is influenced by the nodes at both ends of the edge. This is addressed by averaging the priority scores of the nodes at both ends to reflect the edge's propagation strength. The comprehensive response capability to disturbances means that the disturbance propagation intensity is simultaneously affected by the source node (node). ) and target node (node) Common constraints on importance; and Indicates the index of the node; In the closed-loop abnormal gain matrix, the rows represent the starting nodes of the closed-loop path, corresponding to a certain measuring point (layer) in the wellbore, and the columns represent the ending nodes of the closed-loop path, also corresponding to a certain measuring point (layer) in the wellbore.

[0041] This approach addresses the following technical problems of existing technologies: Traditional methods typically analyze only single-point anomalies or localized events, making it difficult to reflect the transmission relationship of anomalies between different layers in the wellbore. They lack structural constraints and temporal sequence control over anomaly propagation paths, easily leading to false or redundant anomaly correlations. Existing methods struggle to standardize and rank the contribution of different anomaly factors, failing to clearly identify the main disturbance source. Anomaly response intensity cannot dynamically reflect the actual importance and responsiveness of nodes. Anomaly propagation is usually influenced by node layer structure and time delays; traditional methods struggle to simultaneously consider node layer constraints and time lags, resulting in ambiguous anomaly causal relationships. Loop coupling relationships exist between anomaly events; traditional methods struggle to capture the amplification or feedback effect of closed-loop paths on anomaly propagation, affecting the accuracy of anomaly source localization.

[0042] Compared to existing technologies, the advantages are as follows: Anomalies are organized into node sets according to measurement points (layers) and time windows, and propagation is only allowed between adjacent layers, ensuring a reasonable anomaly path structure and eliminating false connections in cross-layer propagation. A maximum hysteresis window limit is set to control the time delay of anomaly propagation, ensuring a reasonable time sequence. Causal weights are weighted and adjusted through perturbation priority scoring, and the perturbation response function is calculated in conjunction with node priority, ensuring that the anomaly propagation intensity of each edge reflects the importance of the node. Node priority scoring, through standardization, clearly distinguishes the contribution of each anomaly factor, providing a basis for anomaly source identification. Enumerating finite closed-loop paths, calculating closed-loop causal gains, and accumulating them into the closed-loop anomaly gain matrix enables a comprehensive evaluation of loop coupling effects. Through the closed-loop gain matrix, anomaly sources can be identified more accurately, and anomaly propagation paths and intensities can be analyzed, providing a scientific basis for wellbore anomaly diagnosis and prediction.

[0043] Methods for finding the most likely anomaly source vector include: Receive the closed-loop anomaly gain matrix and model the propagation process of the anomaly event from the anomaly source node to the global measurement points; normalize the closed-loop anomaly gain matrix by row or column to obtain the anomaly intensity distribution vector of each measurement point, establish an anomaly diffusion model, and regard the observed anomaly intensity distribution vector of the measurement points as the result of the anomaly source vector after diffusion. The abnormal diffusion model is as follows: in, This represents the observed distribution vector of anomaly intensity at the measuring points; This represents the anomaly source vector in the wellbore; This represents the adjacency matrix between measurement points; This indicates the diffusion scale, controlling the degree of diffusion of the anomalous source vector; This represents the diffusion operator, used to describe the process of anomalies propagating along wellbore measuring points or stratigraphic networks; This represents the noise term, indicating unavoidable random interference during the observation process; Based on the anomaly diffusion model, the problem of anomaly source vector localization is transformed into an optimization problem with sparse prior constraints. A sparse optimization model is constructed, and the error between the anomaly intensity distribution vector of the measurement point and the anomaly intensity distribution vector predicted by the sparse optimization model is minimized. The effect of the diffusion operator is calculated by an approximate expansion method, and the optimal solution is obtained by using sparse constraints. The optimal solution obtained is the most likely anomaly source vector.

[0044] It should be noted that anomalies typically occur only at a few measurement points, thus the anomaly source vector is sparse. The anomaly source localization problem is transformed into an optimization problem: while ensuring that the predicted anomalies at the measurement points are consistent with the observed anomalies, the number of anomaly sources is limited by sparsity constraints, ensuring that the anomaly source for most measurement points is zero, and only the most likely anomaly source is retained.

[0045] The following technical problems of existing technologies have been solved: the source of abnormal events is difficult to locate accurately, and the propagation path and intensity of abnormal events are difficult to quantify; existing methods mostly rely on single-point anomaly detection or simple correlation analysis, and cannot comprehensively consider the propagation law of abnormal events between different layers and measurement points; in addition, anomaly sources are usually sparsely distributed, but traditional methods cannot make use of this prior information, resulting in low accuracy of anomaly source location and high false alarm and false alarm rates.

[0046] The advantages over existing technologies are as follows: By receiving the closed-loop anomaly gain matrix, the propagation process of anomaly events from the anomaly source node to all measurement points is modeled. The closed-loop anomaly gain matrix is ​​normalized to obtain the anomaly intensity distribution vector for each measurement point, establishing an anomaly diffusion model. The anomaly observations at measurement points are considered as the result of the diffusion effect of the anomaly source, thus quantifying the propagation characteristics of anomaly events. Furthermore, the anomaly source localization problem is transformed into an optimization problem with sparse prior constraints. A sparse optimization model is constructed, minimizing the error between the predicted and observed anomaly intensity values ​​at measurement points. Simultaneously, sparse constraints are introduced to ensure that the anomaly source for most measurement points is zero, retaining only the most probable anomaly source, thus ensuring high accuracy in anomaly source localization. The diffusion operator effect is calculated using an approximate expansion method, effectively handling the complexity of matrix exponential operations and improving solution efficiency.

[0047] Methods for outputting anomaly source localization results include: A multi-scale Monte Carlo method is employed for uncertainty assessment. Based on the anomaly source vector, random disturbances are introduced to simulate the uncertainties in the distribution and diffusion process of anomaly intensity at measurement points, such as random variations in measurement point noise and propagation delay. Disturbance samples are generated at different scales, the anomaly source vector is repeatedly calculated, and statistical analysis is performed on the disturbance samples to obtain the spatial probability distribution of the anomaly source at each measurement point, outputting the anomaly source localization result.

[0048] Methods for determining the propagation path and intensity of abnormal events and visualizing them include: Receive the closed-loop anomaly gain matrix and anomaly source location results. Starting from each anomaly source node, construct the propagation path of the anomaly event along the wellbore measuring points based on the non-zero elements of the closed-loop anomaly gain matrix, and record the feasible paths of the anomaly event from the anomaly source node to each measuring point. Along each defined propagation path, the propagation intensity on each propagation path is calculated by combining the causal weights of the closed-loop anomaly gain matrix. The propagation intensity is obtained by accumulating the causal weights of each edge along the propagation path. The anomaly source node is represented by the color depth of the node, and the thickness and transparency of the edge represent the anomaly propagation intensity. The propagation path and propagation intensity of the anomaly event are dynamically visualized.

[0049] The preset mean deviation threshold is set by the staff. By collecting different mean deviations, the average of multiple mean deviations is taken as the preset mean deviation threshold. Similarly, preset variance threshold, preset peak threshold, preset energy density change threshold, and preset tubing stress threshold are set and can be adjusted by the staff according to the actual situation during system operation.

[0050] In this embodiment, by forming a node set according to the measurement point and time window sequence of abnormal events, and allowing propagation only between adjacent layers, a reasonable abnormal path structure is ensured, and false connections propagating across layers are eliminated. A maximum hysteresis window limit is set to control the time delay of anomaly propagation and ensure a reasonable time sequence. Causal weights are weighted and adjusted through disturbance priority scoring, and the disturbance response function is calculated in conjunction with node priority, ensuring that the anomaly propagation intensity of each edge reflects the importance of the node. The node priority scoring, through standardization, clearly distinguishes the contribution of each anomaly factor, providing a basis for anomaly source identification. Enumerating finite loop closed-loop paths, calculating the closed-loop causal gain, and accumulating it into the closed-loop anomaly gain matrix achieves a comprehensive evaluation of loop coupling effects. Through the closed-loop gain matrix, anomaly sources can be identified more accurately, and anomaly propagation paths and intensity can be analyzed, providing a scientific basis for wellbore anomaly diagnosis and prediction.

[0051] By receiving the closed-loop anomaly gain matrix, the propagation process of anomaly events from the anomaly source node to all measurement points is modeled. The closed-loop anomaly gain matrix is ​​then normalized to obtain the anomaly intensity distribution vector for each measurement point, establishing an anomaly diffusion model. The anomaly observations at the measurement points are considered as the result of the diffusion effect of the anomaly source, thus quantifying the propagation characteristics of the anomaly event. Furthermore, the anomaly source localization problem is transformed into an optimization problem with sparse prior constraints. A sparse optimization model is constructed, minimizing the error between the predicted and observed anomaly intensity values ​​at the measurement points. Simultaneously, sparse constraints are introduced to ensure that the anomaly source for most measurement points is zero, retaining only the most probable anomaly source, thus ensuring high accuracy in anomaly source localization. The diffusion operator effect is calculated using an approximate expansion method, effectively handling the complexity of matrix exponential operations and improving solution efficiency.

[0052] Example 8 Please see Figure 2 As shown, the parts not described in detail in this embodiment are described in Embodiment 7. A method for analyzing and visualizing wellbore monitoring data is provided, including: S1. Collect multi-source wellbore monitoring data during wellbore operation, and preprocess the multi-source wellbore monitoring data to generate wellbore monitoring feature data with a unified format; S2. Anomaly detection is performed on wellbore monitoring feature data using a sliding window statistical analysis method. The detected anomalies are organized into a hierarchical event matrix of wellbore anomalies, and the perturbation factors of the anomalies are prioritized to obtain the perturbation priority of the anomalies. S3. Based on the hierarchical event matrix and the disturbance priority of abnormal events, the finite loop causal gain evaluation method is adopted to calculate only the closed loop path between adjacent layers in the wellbore and output the closed loop abnormal gain matrix. S4. Receive the closed-loop anomaly gain matrix, combine it with the sparse optimization model to solve for the most likely anomaly source vector, and use the multi-scale Monte Carlo method to evaluate the uncertainty of the anomaly source vector, and output the anomaly source localization result. S5. Based on the closed-loop anomaly gain matrix and the anomaly source localization results, determine the propagation path and propagation intensity of the anomaly event and visualize it.

[0053] Since the electronic device described in this embodiment is the electronic device used to implement the wellbore monitoring data analysis and visualization system and method described in this application, those skilled in the art can understand the specific implementation methods and various variations of the electronic device in this embodiment based on the wellbore monitoring data analysis and visualization system and method described in this application. Therefore, how the electronic device implements the method in this application will not be described in detail here. Any electronic device used by those skilled in the art to implement the wellbore monitoring data analysis and visualization system and method described in this application falls within the scope of protection of this application.

[0054] The above formulas are all dimensionless calculations. The formulas are derived from software simulations based on a large amount of collected data to obtain the most recent real-world results. The preset parameters and thresholds in the formulas are set by those skilled in the art according to the actual situation.

[0055] The above description is merely a preferred embodiment of the present invention, and the scope of protection of the present invention is not limited to the above embodiments. All technical solutions falling within the scope of the present invention's concept are within the scope of protection of the present invention. It should be noted that for users of ordinary technical skills, any improvements and modifications made without departing from the principles of the present invention should also be considered within the scope of protection of the present invention.

Claims

1. A system for analyzing and visualizing wellbore monitoring data, characterized in that, include: The data acquisition and processing module collects multi-source wellbore monitoring data during wellbore operation and preprocesses the multi-source wellbore monitoring data to generate wellbore monitoring feature data in a uniform format. The event disturbance assessment module uses a sliding window statistical analysis method to detect abnormal events in wellbore monitoring feature data. It organizes the detected abnormal events into a hierarchical event matrix of wellbore anomalies and prioritizes the disturbance factors of the abnormal events to obtain the disturbance priority of the abnormal events. The causal gain fusion module, based on the hierarchical event matrix and the perturbation priority of abnormal events, adopts the finite loop causal gain evaluation method, calculates only the closed loop path between adjacent layers in the wellbore, and outputs the closed loop abnormal gain matrix. The anomaly source localization module receives the closed-loop anomaly gain matrix, solves for the most likely anomaly source vector using a sparse optimization model, evaluates the uncertainty of the anomaly source vector using a multi-scale Monte Carlo method, and outputs the anomaly source localization result. The anomaly propagation visualization module determines the propagation path and intensity of anomaly events based on the closed-loop anomaly gain matrix and anomaly source localization results, and then visualizes and displays them.

2. The analysis and visualization system based on wellbore monitoring data according to claim 1, characterized in that, The multi-source wellbore monitoring data includes pressure data, temperature data, flow rate data, vibration and acoustic data, tubing stress data, and fluid physical property parameters.

3. The analysis and visualization system based on wellbore monitoring data according to claim 2, characterized in that, The method for generating wellbore monitoring feature data in a uniform format includes: An adaptive filtering method is used to denoise the multi-source wellbore monitoring data. Linear normalization is used to unify the dimensions of the denoised multi-source wellbore monitoring data, mapping different types of parameters to a unified numerical scale. The normalized multi-source monitoring data is processed for time synchronization. An interpolation method is used to align the multi-source monitoring data with different sampling frequencies according to a unified time reference. The denoised, normalized and time-aligned multi-source monitoring data are then integrated to generate wellbore monitoring feature data with a unified format.

4. The analysis and visualization system based on wellbore monitoring data according to claim 3, characterized in that, The method for organizing detected abnormal events into a hierarchical event matrix of wellbore anomalies includes: The wellbore monitoring feature data with a uniform format is divided into sliding windows according to the time series. Each sliding window contains continuous sampling points of a fixed length. The length of the sliding window and the sliding step size are set according to the sampling frequency of the multi-source wellbore monitoring data. For wellbore monitoring characteristic data within each sliding window, abnormal indicators are calculated and statistically analyzed. These abnormal indicators include mean deviation, variance, peak value, and energy density change. For each abnormal indicator, a corresponding abnormal threshold is set to judge the abnormal indicators within the sliding window. The abnormal thresholds include preset mean deviation threshold, preset variance threshold, preset peak value threshold, and preset energy density change threshold. Windows that meet the corresponding abnormal thresholds are marked as abnormal events. The detected abnormal events are organized into a hierarchical event matrix according to the wellbore measuring points and time sequence. The rows of the matrix represent different measuring points, and the columns represent different time windows.

5. The analysis and visualization system based on wellbore monitoring data according to claim 4, characterized in that, The method for obtaining the perturbation priority of abnormal events includes: For each abnormal event, the disturbance factors of the abnormal event are extracted based on the multi-source wellbore monitoring characteristic data. The disturbance factors include pressure fluctuations, temperature changes, flow deviations, vibration anomalies, tubing stress anomalies, and changes in fluid physical parameters. For each disturbance factor, quantitative indicators are calculated, including the rate of pressure change, the rate of temperature rise and fall, the proportion of flow rate deviating from the preset reference flow rate, the abnormal amplitude of the frequency of downhole equipment vibration signal, the extent to which the tubing stress exceeds the preset tubing stress threshold, and the degree of change of fluid physical properties such as density, viscosity, gas content, and sand content. The quantitative indicators of each disturbance factor are combined with the preset weighting factors of each disturbance factor to calculate the disturbance contribution of each disturbance factor; the contribution of each disturbance factor is standardized to obtain the priority score of each disturbance factor; based on the priority score, all disturbance factors are ranked to form the disturbance priority of the abnormal event.

6. The analysis and visualization system based on wellbore monitoring data according to claim 5, characterized in that, The method for obtaining the closed-loop abnormal gain matrix includes: The abnormal events in the hierarchical event matrix are arranged into an abnormal event node set according to the measurement point and time window order. Each abnormal event is a node. For node pairs that satisfy the adjacent layer constraints and whose time window difference does not exceed the preset maximum lag window, a directed candidate edge is generated and the corresponding time lag is recorded. Assign causal weights to each directed candidate edge. The causal weights are adjusted according to the perturbation priority of abnormal events. Form a directed graph by all nodes and directed candidate edges. In the directed graph, starting from each node, enumerate closed loop paths with a length not exceeding the preset maximum loop length. Only retain closed loop paths of adjacent layers to form a finite set of closed loop paths. For each closed-loop path in the finite set of closed loops, calculate the closed-loop causal gain. Traverse all closed-loop paths. For each pair of nodes involved in a closed-loop path, accumulate the closed-loop causal gain of that closed-loop path to the corresponding row and column elements of the closed-loop abnormal gain matrix. By accumulating all closed-loop paths, a complete closed-loop abnormal gain matrix is ​​formed.

7. The analysis and visualization system based on wellbore monitoring data according to claim 6, characterized in that, The methods for finding the most likely anomaly source vector include: Receive the closed-loop anomaly gain matrix and model the propagation process of the anomaly event from the anomaly source node to the global measurement points; normalize the closed-loop anomaly gain matrix by row or column to obtain the anomaly intensity distribution vector of each measurement point, establish an anomaly diffusion model, and regard the observed anomaly intensity distribution vector of the measurement points as the result of the anomaly source vector after diffusion. Based on the anomaly diffusion model, the problem of anomaly source vector localization is transformed into an optimization problem with sparse prior constraints. A sparse optimization model is constructed, and the error between the anomaly intensity distribution vector of the measurement point and the anomaly intensity distribution vector predicted by the sparse optimization model is minimized. The effect of the diffusion operator is calculated by an approximate expansion method, and the optimal solution is obtained by using sparse constraints. The optimal solution obtained is the most likely anomaly source vector.

8. The analysis and visualization system based on wellbore monitoring data according to claim 7, characterized in that, The method for outputting the anomaly source localization result includes: The multi-scale Monte Carlo method is used for uncertainty assessment. Based on the anomaly source vector, random perturbation is introduced to generate perturbation samples at different scales. The anomaly source vector is repeatedly calculated, and statistical analysis is performed on the perturbation samples to obtain the spatial probability distribution of the anomaly source at each measuring point, and the anomaly source localization result is output.

9. The analysis and visualization system based on wellbore monitoring data according to claim 8, characterized in that, The method for determining the propagation path and intensity of abnormal events and visualizing them includes: Receive the closed-loop anomaly gain matrix and anomaly source location results, and starting from each anomaly source node, construct the propagation path of the anomaly event along the wellbore measuring points based on the non-zero elements of the closed-loop anomaly gain matrix. Along each defined propagation path, the propagation intensity on each propagation path is calculated by combining the causal weights of the closed-loop anomaly gain matrix. The propagation intensity is obtained by accumulating the causal weights of each edge along the propagation path. The anomaly source node is represented by the color depth of the node, and the thickness and transparency of the edge represent the anomaly propagation intensity. The propagation path and propagation intensity of the anomaly event are dynamically visualized.

10. A method for analyzing and visualizing wellbore monitoring data, comprising the wellbore monitoring data analysis and visualization system as described in any one of claims 1 to 9, characterized in that... The steps are as follows: S1. Collect multi-source wellbore monitoring data during wellbore operation, and preprocess the multi-source wellbore monitoring data to generate wellbore monitoring feature data with a unified format; S2. Anomaly detection is performed on wellbore monitoring feature data using a sliding window statistical analysis method. The detected anomalies are organized into a hierarchical event matrix of wellbore anomalies, and the perturbation factors of the anomalies are prioritized to obtain the perturbation priority of the anomalies. S3. Based on the hierarchical event matrix and the disturbance priority of abnormal events, the finite loop causal gain evaluation method is adopted to calculate only the closed loop path between adjacent layers in the wellbore and output the closed loop abnormal gain matrix. S4. Receive the closed-loop anomaly gain matrix, combine it with the sparse optimization model to solve for the most likely anomaly source vector, and use the multi-scale Monte Carlo method to evaluate the uncertainty of the anomaly source vector, and output the anomaly source localization result. S5. Based on the closed-loop anomaly gain matrix and the anomaly source localization results, determine the propagation path and propagation intensity of the anomaly event and visualize it.

Images