Iot monitoring data anomaly early warning method and system based on digital twinning

Abstract

The application relates to the technical field of Internet of Things monitoring, and particularly discloses an Internet of Things monitoring data anomaly early warning method and system based on digital twinning, which comprises the following steps: taking the monitoring data collected by an Internet of Things sensor in a healthy state as a constraint, constructing a digital twin, substituting real-time data in the Internet of Things sensor into a physical control equation of the digital twin, inversely deriving two inverse values of a same target physical parameter, and calculating a consistency residual between the two inverse values; extracting a structural feature of the consistency residual sequence, wherein the structural feature comprises a local curvature mutation point of the consistency residual on a spatial distribution of an oil and gas pipeline; comparing the structural feature with an admissible manifold; and when a topological form of the local curvature mutation point and any subset of the admissible manifold do not satisfy a homeomorphism relationship, determining that the oil and gas pipeline is currently in an abnormal state and outputting early warning information. The application can improve the accuracy and real-time performance of the Internet of Things monitoring data anomaly early warning based on digital twinning.

Description

Technical Field

[0001] This invention relates to the field of Internet of Things (IoT) monitoring technology, and in particular to a method and system for early warning of abnormal IoT monitoring data based on digital twins. Background Technology

[0002] Existing IoT monitoring data anomaly early warning methods typically rely on single sensors or simple statistical models to make threshold judgments on real-time data, and are not good at handling in-depth modeling of the physical processes and spatial distribution characteristics of the system. These methods struggle to accurately capture the nonlinear mapping relationship between monitoring data and system state under complex operating conditions, resulting in low anomaly identification sensitivity, high false alarm rate, and delayed response to early minor anomalies in long-distance, multi-variable coupled monitoring scenarios such as oil and gas pipelines. This makes it difficult to meet the engineering requirements for accurate and real-time early warning.

[0003] In recent years, monitoring methods based on digital twins have improved condition assessment capabilities by constructing virtual mirrors of physical entities. However, existing solutions mostly focus on forward simulation or parameter estimation, failing to fully utilize the consistency characteristics of data under healthy conditions for anomaly detection. Furthermore, traditional methods often neglect the structured topological properties of residual sequences when fusing spatial distribution information from multiple sensors, making it difficult to effectively distinguish manifold differences between different fault modes, resulting in unsatisfactory overall early warning efficiency. Therefore, improving anomaly early warning capabilities based on IoT monitoring data has become an urgent problem to be solved. Summary of the Invention

[0004] This invention provides a method and system for early warning of abnormal IoT monitoring data based on digital twins, in order to solve the problems mentioned in the background art.

[0005] To achieve the above objectives, the present invention provides an IoT monitoring data anomaly early warning method based on digital twins, comprising: P1. Using the monitoring data collected by the IoT sensor in a healthy state as a constraint, construct a digital twin, substitute the real-time data in the IoT sensor into the physical control equation of the digital twin, invert two inversion values ​​of the same target physical parameter, and calculate the consistency residual between the two inversion values. P2. Extract the structural features of the consistency residual sequence, the structural features including the local curvature abrupt change points of the consistency residual in the spatial distribution of the oil and gas pipeline; P3. Compare the structural features with the permissive manifold composed of consistent residual sequences under historical healthy states. When the topology of the local curvature abrupt change point and any subset of the permissive manifold do not satisfy the homeomorphic relationship, determine that the oil and gas pipeline is currently in an abnormal state and output a warning message.

[0006] In a preferred embodiment, constructing a digital twin based on data collected by IoT sensors in a healthy state includes: Multiple sets of monitoring data continuously collected by the IoT sensor in a healthy state are obtained, and the multiple sets of monitoring data are timestamped to obtain the health data sequence of the IoT sensor. Using each set of data in the health data sequence as input boundary values, an initial virtual structure consistent with the geometric contour of the oil and gas pipeline is established in the digital space; All spatial nodes in the initial virtual structure are bound to the corresponding sensor locations in the health data sequence to form a data-space mapping relationship; Based on the data-space mapping relationship, the parameter value range of each node in the initial virtual structure under the healthy state is locked to obtain the digital twin of the oil and gas pipeline.

[0007] In a preferred embodiment, the step of establishing an initial virtual structure in the digital space that conforms to the geometric contour of the oil and gas pipeline, using each set of data in the health data sequence as input boundary values, includes: Read the spatial coordinates of the sensor on the oil and gas pipeline corresponding to each group of data in the health data sequence; Based on the spatial coordinates, a discrete node array matching the external dimensions of the oil and gas pipeline is generated in the digital space. The measured value of each group of data in the health data sequence is assigned to the corresponding position node in the discrete node array, which serves as the activity boundary constraint value of the corresponding position node. Adjacent discrete nodes are connected by edges to form a gridded virtual geometry, which is then used as the initial virtual structure.

[0008] In a preferred embodiment, the step of substituting real-time data from the IoT sensor into the physical control equations of the digital twin to derive two inversion values ​​of the same target physical parameter, and calculating the consistency residual between the two inversion values, includes: The system acquires real-time data from the IoT sensor and simultaneously inputs the real-time data into the first and second physical control equations contained in the digital twin. The first physical control equation outputs the first inversion estimate corresponding to the target physical parameters based on the real-time data. The second physical control equation outputs a second inversion estimate corresponding to the target physical parameters based on the same real-time data; The first inversion estimate is subtracted from the second inversion estimate point by point to obtain the residual value at each time step, and the residual value is used as the consistency residual.

[0009] In a preferred embodiment, the extraction of structural features of the consistency residual sequence, the structural features including local curvature abrupt changes in the spatial distribution of the consistency residuals in the oil and gas pipeline, includes: Repeatedly perform the consistency residual calculation to obtain a consistency residual sequence within a continuous time window, wherein the consistency residual is the difference between two inversion values ​​of the same target physical parameter; Extract the spatial location on the oil and gas pipeline corresponding to each residual value in the consistency residual sequence; According to the spatial location along the extension direction of the oil and gas pipeline, each residual value is arranged into a spatial distribution sequence. The residual values ​​at adjacent positions in the spatial distribution sequence are compared in turn, and the variation between adjacent residual values ​​is calculated. When the variation range between the residual value at any spatial location and the residual value at an adjacent location is greater than the variation range of the adjacent locations before and after the spatial location, the spatial location is determined to be a local curvature abrupt change point.

[0010] In a preferred embodiment, the permissive manifold includes: During the historical healthy operation of the oil and gas pipeline, multiple sets of monitoring data were collected. For each set of monitoring data, a consistent residual sequence was calculated along the spatial distribution of the oil and gas pipeline, and the corresponding historical local curvature abrupt change point was extracted from each consistent residual sequence. The connection relationships of each historical local curvature abrupt change point on the spatial distribution of oil and gas pipelines are used to form a historical topological subset, and the set of all historical topological subsets is taken as an admissible manifold.

[0011] In a preferred embodiment, the comparison of the structural features with the permissive manifold composed of consistent residual sequences from historical healthy states, and the determination that the oil and gas pipeline is currently in an abnormal state and the output of early warning information when the topology of the local curvature abrupt change point and any subset of the permissive manifold do not satisfy the homeomorphic relation, includes: Extract the connection relationship of the local curvature abrupt change points in the spatial distribution of oil and gas pipelines to form a topological map of the local curvature abrupt change points; Extract the permissive manifold composed of consistent residual sequences under historical health states, and perform continuous deformation comparison between the topological morphology graph and each historical topological subset in the permissive manifold; When the topology diagram cannot be deformed into the shape of any historical topology subset through continuous stretching, compression and bending, it is determined that the topology diagram is different from the historical topology subset, and an anomaly mark of the current state of the oil and gas pipeline is generated. Based on the anomaly marker, an early warning message containing the anomaly location and the topology map category is generated and output to the operation and maintenance terminal.

[0012] In a preferred embodiment, when the topology diagram cannot be deformed into the shape of any historical topology subset through continuous stretching, compression, and bending, determining that the topology diagram is different from the historical topology subset and marking the current state of the oil and gas pipeline as abnormal includes: The homeomorphism degree between the topological graph and the historical topological subset is calculated according to the following formula: ; In the formula, These are the zeroth, first, and second dimensions of the topological graph, respectively; These are the zeroth, first, and second dimensions of the historical topological subset, respectively; The degree of homeomorphism; When the homeomorphism difference is greater than zero, the topological morphology graph is determined to be a different homeomorphism from the historical topological subset.

[0013] In a preferred embodiment, based on the anomaly marker, an early warning message containing the anomaly location and the topology map category is generated and output to the operation and maintenance terminal, including: Extract the spatial coordinates of the local curvature abrupt change point on the oil and gas pipeline, and use the spatial coordinates as the abnormal location; Identify the number of connected branches and the number of loop structures in the topology graph, and combine the number of connected branches and the number of loop structures into the topology graph category; The abnormal location and the topology map category are combined into a warning message, which is then sent to the terminal device of the operation and maintenance personnel.

[0014] To address the aforementioned problems, this invention also provides an IoT monitoring data anomaly early warning system based on digital twins, the system comprising: The consistency residual generation module is used to construct a digital twin based on the monitoring data collected by the IoT sensor in a healthy state. The module substitutes the real-time data from the IoT sensor into the physical control equation of the digital twin to derive two inversion values ​​of the same target physical parameter and calculates the consistency residual between the two inversion values. A curvature abrupt change point extraction module is used to extract structural features of the consistency residual sequence, wherein the structural features include local curvature abrupt change points of the consistency residual in the spatial distribution of the oil and gas pipeline. The topological anomaly determination module is used to compare the structural features with the permissible manifold composed of consistent residual sequences from historical healthy states. When the topological features of a local curvature abrupt change point and any subset of the permissible manifold do not satisfy the homeomorphic relationship, the oil and gas pipeline is determined to be in an abnormal state and an early warning message is output.

[0015] Compared with the prior art, the present invention has the following beneficial effects: 1. This invention constructs a digital twin and inverts two inversion values ​​of the same target physical parameter. The consistent residual sequence between the two values ​​is used to characterize the dynamic deviation of the monitoring data. Based on this, local curvature abrupt change points in the spatial distribution of the residuals in oil and gas pipelines are extracted as structural features. This method effectively preserves the structural information of the monitoring data in the spatial dimension, making anomaly location more accurate. It also avoids dependence on a single threshold or simple statistics, significantly improving the spatial resolution and structural sensitivity of anomaly early warning.

[0016] 2. This invention further compares the extracted topological morphology of local curvature abrupt change points with the admissible manifold formed under historical healthy conditions for homeomorphism. An early warning is triggered only when the topological morphology does not satisfy homeomorphism with any subset of the admissible manifold. This topological invariant-based judgment mechanism can adapt to different operating conditions and multiple fault modes, eliminating the need to set separate rules for each anomaly. This significantly improves the versatility and robustness of the early warning method, while reducing the false alarm rate and ensuring the efficient operation of IoT monitoring data anomaly early warning. Attached Figure Description

[0017] Figure 1 This is a flowchart illustrating an abnormal early warning method for IoT monitoring data based on digital twins, provided in an embodiment of the present invention. Figure 2 A functional block diagram of an IoT monitoring data anomaly early warning system based on digital twin provided in an embodiment of the present invention; The realization of the objective, functional features and advantages of the present invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0018] It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

[0019] This application provides a method for early warning of abnormal IoT monitoring data based on digital twins. The executing entity of this method includes, but is not limited to, at least one of the following electronic devices that can be configured to execute the method provided in this application: a server, a terminal, etc. In other words, the method can be executed by software or hardware installed on a terminal device or a server device. The server includes, but is not limited to, a single server, a server cluster, a cloud server, or a cluster of cloud servers. The server can be an independent server or a cloud server providing basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, content delivery networks (CDNs), and big data and artificial intelligence platforms.

[0020] Reference Figure 1 The diagram shown is a flowchart illustrating an anomaly warning method for IoT monitoring data based on digital twins according to an embodiment of the present invention. In this embodiment, the anomaly warning method for IoT monitoring data based on digital twins includes: P1. Using the monitoring data collected by the IoT sensor in a healthy state as a constraint, construct a digital twin, substitute the real-time data in the IoT sensor into the physical control equation of the digital twin, derive two inversion values ​​of the same target physical parameter, and calculate the consistency residual between the two inversion values.

[0021] In this embodiment of the invention, the step of constructing a digital twin based on data collected by IoT sensors in a healthy state includes: Multiple sets of monitoring data continuously collected by the IoT sensor in a healthy state are obtained, and the multiple sets of monitoring data are timestamped to obtain the health data sequence of the IoT sensor. Using each set of data in the health data sequence as input boundary values, an initial virtual structure consistent with the geometric contour of the oil and gas pipeline is established in the digital space; All spatial nodes in the initial virtual structure are bound to the corresponding sensor locations in the health data sequence to form a data-space mapping relationship; Based on the data-space mapping relationship, the parameter value range of each node in the initial virtual structure under the healthy state is locked to obtain the digital twin of the oil and gas pipeline.

[0022] The step of establishing an initial virtual structure in digital space that matches the geometric contour of the oil and gas pipeline, using each set of data in the health data sequence as input boundary values, includes: Read the spatial coordinates of the sensor on the oil and gas pipeline corresponding to each group of data in the health data sequence; Based on the spatial coordinates, a discrete node array matching the external dimensions of the oil and gas pipeline is generated in the digital space. The measured value of each group of data in the health data sequence is assigned to the corresponding position node in the discrete node array, which serves as the activity boundary constraint value of the corresponding position node. Adjacent discrete nodes are connected by edges to form a gridded virtual geometry, which is then used as the initial virtual structure.

[0023] The step of substituting real-time data from the IoT sensor into the physical control equations of the digital twin to derive two inversion values ​​of the same target physical parameter, and calculating the consistency residual between the two inversion values, includes: The system acquires real-time data from the IoT sensor and simultaneously inputs the real-time data into the first and second physical control equations contained in the digital twin. The first physical control equation outputs the first inversion estimate corresponding to the target physical parameters based on the real-time data. The second physical control equation outputs a second inversion estimate corresponding to the target physical parameters based on the same real-time data; The first inversion estimate is subtracted from the second inversion estimate point by point to obtain the residual value at each time step, and the residual value is used as the consistency residual.

[0024] This process involves acquiring multiple sets of monitoring data continuously collected by IoT sensors while the pipeline is in a healthy operating state, and aligning these sets of data with their timestamps to obtain a health data sequence from the IoT sensors. Specifically, multiple sets of monitoring data continuously recorded by IoT sensors deployed at various locations along the oil and gas pipeline are read when the pipeline is in a healthy operating state. Each set of monitoring data contains values ​​collected by different sensors at the same time. These monitoring data are sorted according to their respective recording timestamps. All sensor values ​​with the same timestamp are grouped together. For data with slightly different timestamps, interpolation is used to adjust the data from different sensors to a unified time point, thus forming a health data sequence arranged in chronological order. Each element in this sequence corresponds to the set of all sensor values ​​at a given time.

[0025] Using each set of data in the health data sequence as input boundary values, an initial virtual structure consistent with the geometric contour of the oil and gas pipeline is established in the digital space. Specifically, for each set of data in the health data sequence, the value measured by each sensor in that set of data is used as the boundary input condition at the corresponding spatial location. Within the computer's digital space, based on the actual geometric dimensions and orientation of the oil and gas pipeline, a virtual geometric shell that is exactly the same as the shape of the real oil and gas pipeline is created. This shell consists of a series of spatial points and line segments connecting these points, and its shape and size are consistent with the diameter, wall thickness, curvature, and total length of the real oil and gas pipeline.

[0026] The initial virtual structure's spatial nodes are bound to their corresponding sensor locations in the health data sequence, forming a data spatial mapping relationship. Specifically, each spatial node in the initial virtual structure is traversed to find its physical coordinates on a real oil and gas pipeline. Then, the output value of the sensor installed at the same physical coordinates is retrieved from the health data sequence, and a one-to-one correspondence record is established between this value and the spatial node. This association record clarifies which sensor's data each spatial node should receive and the storage location of that sensor data within the virtual structure, thus forming a mapping relationship from sensor data to virtual spatial nodes.

[0027] Based on the data spatial mapping relationship, the parameter value ranges of each node in the initial virtual structure under healthy conditions are locked, resulting in a digital twin of the oil and gas pipeline. Specifically, using the established data spatial mapping relationship, sensor values ​​at all times in the health data sequence are assigned to the corresponding spatial nodes in the initial virtual structure. All possible value ranges for each spatial node under healthy conditions are recorded, and these ranges are considered the legal value intervals for that node under healthy conditions, and values ​​exceeding these intervals are not allowed. After locking the value ranges of all nodes, the initial virtual structure is transformed into a digital twin that reflects the health status characteristics of the oil and gas pipeline.

[0028] Read the spatial coordinates of the sensors on the oil and gas pipeline corresponding to each set of data in the health data sequence. Specifically, extract each set of data from the health data sequence, and for each sensor value in each set of data, find the three-dimensional spatial coordinates of the sensor's actual installation location on the oil and gas pipeline, including the mileage value along the pipeline route, the angle on the pipeline cross-section, and the height value above the ground. Organize these coordinate values ​​into a coordinate list according to the sensor number.

[0029] Based on spatial coordinates, a discrete node array matching the dimensions of the oil and gas pipeline is generated in digital space. Specifically, the spatial coordinates of all sensors are input into the digital space, and the overall boundary of the oil and gas pipeline is determined according to the distribution range of these coordinate points. Then, additional spatial points are inserted inside the boundary at fixed intervals, so that these spatial points, together with the original sensor position points, form a discrete node array that covers the entire shape of the oil and gas pipeline and is equidistant from each other or adaptively distributed according to the pipeline curvature. The total number of points in this array is much greater than the number of sensors, but the overall shape is completely consistent with the oil and gas pipeline.

[0030] The measured values ​​of each data set in the health data sequence are assigned to the corresponding nodes in the discrete node array as active boundary constraint values. Specifically, for each node in the discrete node array located at a real sensor location, the measured value of that sensor at the same moment is extracted from the health data sequence and directly written into that node. For nodes in the discrete node array that are not at sensor locations, no value is assigned; instead, indirect constraints are applied in subsequent steps using the values ​​of adjacent sensor nodes. These directly written measured values ​​become the active boundary conditions for the subsequent stress or deformation of the virtual structure.

[0031] Adjacent discrete nodes are connected by edges to form a meshed virtual geometry, which serves as the initial virtual structure. Specifically, all nodes in the discrete node array are traversed, and for each node, the nearest nodes in space are found and connected by straight line segments. Following the geometric characteristics of the oil and gas pipeline, corresponding nodes on adjacent cross-sections are longitudinally connected along the pipeline's length, and nodes are circumferentially connected within each cross-section, ultimately forming a meshed virtual geometry composed of numerous triangular or quadrilateral meshes. The surface mesh of this geometry perfectly conforms to the actual shape of the oil and gas pipeline.

[0032] Real-time data is acquired from IoT sensors and simultaneously input into the first and second control equations contained within the digital twin. Specifically, during the operation of the oil and gas pipeline, real-time values ​​collected by all sensors are read through the IoT communication interface and organized into a dataset with the same format as the health data sequence. This dataset is then simultaneously transmitted to the two pre-set control equations within the digital twin, with each equation receiving the data independently.

[0033] The first governing equation, based on real-time data, outputs the first inversion estimate corresponding to the target parameters. Specifically, the first governing equation is a set of equations describing the flow or structural deformation relationships within the oil and gas pipeline. The received real-time data is substituted into these equations as known boundary conditions, and the target parameters are obtained through inverse solving. This value is the first inversion estimate. The target parameter is a specific internal state variable of the oil and gas pipeline, such as the pressure or temperature value at a certain point in the pipeline.

[0034] The second governing equation, based on the same real-time data, outputs a second inversion estimate corresponding to the target parameter. Specifically, the second governing equation is a different set of equations than the first governing equation. It also uses the same set of real-time data as known boundary conditions, and through inverse solving, obtains another value for the same target parameter; this value is the second inversion estimate. Because the first and second governing equations differ mathematically, they will produce different output estimates for the same input data.

[0035] The first inversion estimate is subtracted from the second inversion estimate point by point to obtain the residual value at each time step, and this residual value is used as the consistent residual. Specifically, for the current time step, the first inversion estimate is subtracted from the second inversion estimate to obtain a difference. For each subsequent time step, the above process is repeated to obtain the first and second inversion estimates at each time step, and then the difference at each time step is calculated. These differences are arranged in chronological order, and each difference is a residual value. These residual values ​​together constitute a consistent residual sequence.

[0036] The beneficial effect is that by aligning multiple sets of monitoring data collected under healthy conditions according to timestamps to form a health data sequence, and using each set of data as boundary values ​​to establish an initial virtual structure in digital space consistent with the geometric contour of the oil and gas pipeline, spatial nodes are then bound to sensor locations to form a data space mapping relationship, thereby locking the parameter value range of each node under healthy conditions, resulting in a digital twin that can faithfully reflect health characteristics. By reading the spatial coordinates of the sensors to generate a discrete node array, the measured values ​​are assigned to the corresponding location nodes as active boundary constraint values, and the edges of adjacent nodes are connected to form a gridded virtual geometry, ensuring a high degree of consistency between the initial virtual structure and the actual pipeline shape. After acquiring real-time data from IoT sensors, the first and second control equations are simultaneously input, and two inversion estimates of the same target parameter are output respectively. The two are subtracted point by point to obtain the consistency residual. This process does not rely on a single threshold judgment, but uses the inversion difference of two different control equations for the same input to construct residuals, so that the residuals themselves contain the inherent inconsistency information between the physical control equations, eliminating the dependence on prior fault samples, improving the sensitivity and reliability of anomaly warning, and avoiding false alarms, thus providing an accurate data foundation for subsequent spatial distribution feature analysis based on residual structure.

[0037] P2. Extract the structural features of the consistency residual sequence, wherein the structural features include the local curvature abrupt change points of the consistency residual in the spatial distribution of the oil and gas pipeline.

[0038] In this embodiment of the invention, the extraction of structural features of the consistency residual sequence, wherein the structural features include local curvature abrupt change points in the spatial distribution of the consistency residuals in oil and gas pipelines, includes: Repeatedly perform the consistency residual calculation to obtain a consistency residual sequence within a continuous time window, wherein the consistency residual is the difference between two inversion values ​​of the same target physical parameter; Extract the spatial location on the oil and gas pipeline corresponding to each residual value in the consistency residual sequence; According to the spatial location along the extension direction of the oil and gas pipeline, each residual value is arranged into a spatial distribution sequence. The residual values ​​at adjacent positions in the spatial distribution sequence are compared in turn, and the variation between adjacent residual values ​​is calculated. When the variation range between the residual value at any spatial location and the residual value at an adjacent location is greater than the variation range of the adjacent locations before and after the spatial location, the spatial location is determined to be a local curvature abrupt change point.

[0039] Repeatedly perform the consistency residual calculation to obtain a consistency residual sequence within a continuous time window. The consistency residual is the difference between two inversion values ​​of the same target physical parameter. Specifically, starting from the first moment, calculate the consistency residual for each moment according to the aforementioned steps. Continue this process until a preset time length, such as ten minutes or one hour, is accumulated. Arrange the consistency residuals of all moments within this time period in chronological order to form a consistency residual sequence within a continuous time window. Each element in this sequence corresponds to a residual value at a moment, and these residual values ​​come from the subtraction result between two inversion values ​​of the same target physical parameter.

[0040] Extracting the spatial location on the oil and gas pipeline corresponding to each residual value in the consistent residual sequence involves: traversing each residual value in the consistent residual sequence, identifying which IoT sensors were used to generate the real-time data, and then determining the location of each residual value on the oil and gas pipeline based on the pre-recorded installation coordinates of these sensors. If a residual value involves the locations of multiple sensors, the center point of these locations is taken as the spatial location of the residual value. Finally, the spatial location points on the oil and gas pipeline corresponding to each residual value are obtained.

[0041] According to the spatial location along the extension direction of the oil and gas pipeline, each residual value is arranged into a spatial distribution sequence. The residual values ​​at adjacent positions in the spatial distribution sequence are compared sequentially to calculate the variation between adjacent residual values. Specifically, each residual value is associated with its corresponding spatial location point. Then, along the direction of the oil and gas pipeline from the starting point to the ending point, all residual values ​​are rearranged according to the chronological order of their spatial locations, forming a spatial distribution sequence. The order of the residual values ​​in this sequence is no longer based on time but on their position along the pipeline. Starting from the first residual value in the sequence, the residual values ​​at two adjacent positions are taken sequentially. The previous residual value is subtracted from the subsequent residual value, and the positive value of the difference is taken as the variation between these two adjacent positions.

[0042] When the magnitude of change between the residual value at any spatial location and the residual values ​​at adjacent locations is greater than the magnitude of change between the spatial locations before and after that location, that spatial location is identified as a local curvature abrupt change point. Specifically: For a specific spatial location in the spatial distribution sequence, first calculate the magnitude of change between that location and its next adjacent location, then calculate the magnitude of change between that location and its previous adjacent location. The larger of these two magnitudes is taken as the first comparison value for that location. Then, calculate the magnitude of change between the pair of adjacent locations before and after that location, and take the larger of these two magnitudes as the second comparison value. Compare the first comparison value with the second comparison value. If the first comparison value is greater than the second comparison value, then that specific spatial location is marked as a local curvature abrupt change point; otherwise, it is not marked. This process is repeated for all spatial locations in the spatial distribution sequence except for the beginning and end, until all local curvature abrupt change points are obtained.

[0043] The beneficial effect is that by repeatedly performing consistent residual calculations to obtain a consistent residual sequence within a continuous time window, and then extracting the spatial location on the oil and gas pipeline corresponding to each residual value, the original time-varying residual sequence is transformed into a spatial distribution sequence arranged along the pipeline extension direction, thus preserving the spatial continuity and positional information of the residuals on the pipeline. By sequentially comparing the residual values ​​at adjacent positions in the spatial distribution sequence and calculating the magnitude of change, spatial positions with a change magnitude greater than that of the adjacent positions are identified as local curvature abrupt change points. This process does not rely on any preset thresholds or statistical distribution assumptions, but directly judges based on the relative change relationship between adjacent positions, eliminating interference from the time dimension. This allows the extracted local curvature abrupt change points to accurately reflect the drastic changes in the consistent residuals in the pipeline space. These local curvature abrupt change points, as core components of structural features, provide clear spatial topological basis for subsequent homeomorphic comparison with the allowable manifold under historical healthy conditions, thereby improving the ability of anomaly early warning to identify local structural anomalies in pipelines.

[0044] P3. Compare the structural features with the permissive manifold composed of consistent residual sequences under historical healthy states. When the topology of the local curvature abrupt change point and any subset of the permissive manifold do not satisfy the homeomorphic relationship, determine that the oil and gas pipeline is currently in an abnormal state and output a warning message.

[0045] In this embodiment of the invention, the permissible manifold includes: During the historical healthy operation of the oil and gas pipeline, multiple sets of monitoring data were collected. For each set of monitoring data, a consistent residual sequence was calculated along the spatial distribution of the oil and gas pipeline, and the corresponding historical local curvature abrupt change point was extracted from each consistent residual sequence. The connection relationships of each historical local curvature abrupt change point on the spatial distribution of oil and gas pipelines are used to form a historical topological subset, and the set of all historical topological subsets is taken as an admissible manifold.

[0046] The process of comparing the structural features with the permissive manifold composed of consistent residual sequences from historical healthy states, and determining that the oil and gas pipeline is currently in an abnormal state and outputting early warning information when the topology of the local curvature abrupt change point and any subset of the permissive manifold do not satisfy the homeomorphic relation, includes: Extract the connection relationship of the local curvature abrupt change points in the spatial distribution of oil and gas pipelines to form a topological map of the local curvature abrupt change points; Extract the permissive manifold composed of consistent residual sequences under historical health states, and perform continuous deformation comparison between the topological morphology graph and each historical topological subset in the permissive manifold; When the topology diagram cannot be deformed into the shape of any historical topology subset through continuous stretching, compression and bending, it is determined that the topology diagram is different from the historical topology subset, and an anomaly mark of the current state of the oil and gas pipeline is generated. Based on the anomaly marker, an early warning message containing the anomaly location and the topology map category is generated and output to the operation and maintenance terminal.

[0047] When the topology diagram cannot be deformed into the shape of any historical topology subset through continuous stretching, compression, and bending, it is determined that the topology diagram is different from the historical topology subset, and the current state of the oil and gas pipeline is marked as abnormal, including: The homeomorphism degree between the topological graph and the historical topological subset is calculated according to the following formula: ; In the formula, These are the zeroth, first, and second dimensions of the topological graph, respectively; These are the zeroth, first, and second dimensions of the historical topological subset, respectively; The degree of homeomorphism; When the homeomorphism difference is greater than zero, the topological morphology graph is determined to be a different homeomorphism from the historical topological subset.

[0048] Based on the anomaly marker, an early warning message containing the anomaly location and the topology map category is generated and output to the operation and maintenance terminal, including: Extract the spatial coordinates of the local curvature abrupt change point on the oil and gas pipeline, and use the spatial coordinates as the abnormal location; Identify the number of connected branches and the number of loop structures in the topology graph, and combine the number of connected branches and the number of loop structures into the topology graph category; The abnormal location and the topology map category are combined into a warning message, which is then sent to the terminal device of the operation and maintenance personnel.

[0049] During the historical healthy operation of the oil and gas pipeline, multiple sets of monitoring data are collected. For each set of monitoring data, a consistent residual sequence spatially distributed along the oil and gas pipeline is calculated, and the corresponding historical local curvature abrupt change points are extracted from each consistent residual sequence. Specifically, multiple sets of monitoring data recorded by IoT sensors when the oil and gas pipeline was in a healthy operating state are retrieved from the historical archives of the oil and gas pipeline. Each set of monitoring data corresponds to a different healthy time period. For each set of monitoring data, the consistent residual sequence corresponding to the set of data is obtained according to the aforementioned method of calculating consistent residuals. Then, according to the aforementioned method of extracting local curvature abrupt change points, all historical local curvature abrupt change points within the healthy time period are extracted from the consistent residual sequence. This process is repeated until each set of monitoring data produces a set of historical local curvature abrupt change points.

[0050] The connectivity of each historical local curvature abrupt change point in the spatial distribution of the oil and gas pipeline is used to form a historical topological subset. The set of all historical topological subsets is considered the admissible manifold. Specifically, for all historical local curvature abrupt changes points extracted from a set of health monitoring data, adjacent abrupt changes points are connected by a virtual line segment according to their actual spatial location on the oil and gas pipeline. If multiple abrupt changes points are located within the same cross-section of the pipeline, they are considered coincident points and not connected repeatedly. All abrupt changes points and their connecting line segments together constitute a plane or spatial figure, which is a historical topological subset. This connection operation is performed on each set of historical local curvature abrupt changes points generated from each set of health monitoring data, resulting in multiple historical topological subsets. All these historical topological subsets are put together to form a large set, which is called the admissible manifold.

[0051] The connection relationships of local curvature abrupt change points in the spatial distribution of oil and gas pipelines are extracted to form a topological diagram of local curvature abrupt change points. Specifically, starting from all local curvature abrupt change points calculated at the current moment, the spatial coordinates of each local curvature abrupt change point on the oil and gas pipeline are obtained. According to the front-to-back order of these coordinates along the pipeline, adjacent local curvature abrupt change points are connected with a virtual line segment. If there are no other abrupt change points between two abrupt change points and the pipeline is continuous in this segment, they are directly connected. If two abrupt change points are located at the bend of the pipeline, they are connected along the actual path of the pipeline centerline. Finally, a graph is obtained with abrupt change points as vertices and connecting line segments as edges. This graph is called the topological diagram.

[0052] Extract the permissible manifold composed of consistent residual sequences from historical healthy states. Perform continuous deformation comparisons between the topological morphology graph and each historical topological subset within the permissible manifold. Specifically: Take the first historical topological subset from the permissible manifold, place the currently obtained topological morphology graph and this historical topological subset in the same space, and attempt to change the shape of the topological morphology graph by stretching the distance between vertices, compressing the length of certain edges, and bending the direction of edges. During the shape change process, continuously compare the changed graph with the historical topological subset. If, after a series of stretching, compression, and bending operations, all vertices and edges of the topological morphology graph correspond one-to-one with the vertices and edges of the historical topological subset without breakage or adhesion, the comparison stops. If they do not correspond, take the next historical topological subset from the permissible manifold and repeat the above continuous deformation comparison process until all historical topological subsets have been compared.

[0053] When the topological shape diagram cannot be deformed into the shape of any historical topological subset through continuous stretching, compression, and bending, it is determined that the topological shape diagram is not a different embryo from the historical topological subset, and an anomaly marker for the current state of the oil and gas pipeline is generated. Specifically, after completing the continuous deformation comparison between the topological shape diagram and each historical topological subset in the allowable manifold, if the topological shape diagram does not achieve a one-to-one correspondence between vertices and edges with any historical topological subset, that is, no stretching, compression, or bending operation can make the topological shape diagram into the shape of that historical topological subset, then it is determined that the topological shape diagram is not a different embryo from that historical topological subset. At this time, a marker representing that the oil and gas pipeline is currently in an abnormal state is generated. This marker is an independent identification symbol used to trigger subsequent early warning processes.

[0054] Based on the anomaly markers, an early warning message containing the anomaly location and topology category is generated and output to the maintenance terminal. Specifically, after obtaining the anomaly markers, the spatial coordinates of all local curvature abrupt change points used to generate the topology map are first extracted on the oil and gas pipeline. These coordinates are then organized into an anomaly location list. Next, the number of isolated vertices not connected to other vertices in the topology map is identified as the number of connected components, and the number of closed loops formed by edges and vertices in the topology map is identified as the number of loop structures. The number of connected components and the number of loop structures are combined into a text string as the topology map category. This text string is then merged with the anomaly location list to form a complete early warning message. Finally, the early warning message is sent to the screen of the maintenance personnel's handheld terminal device via wired or wireless communication.

[0055] Calculate the homeomorphism difference between the topological morphology diagram and the historical topological subset. When the homeomorphism difference is greater than zero, determine that the topological morphology diagram and the historical topological subset are not homeomorphic, and mark the current state of the oil and gas pipeline as abnormal.

[0056] Extract the spatial coordinates of local curvature abrupt change points on the oil and gas pipeline, and use the spatial coordinates as anomaly locations. Specifically, iterate through each local curvature abrupt change point extracted at the current moment, look up the pipeline mileage value or three-dimensional spatial coordinate value corresponding to the abrupt change point from the pre-stored sensor position lookup table, and arrange these coordinate values ​​into a coordinate list according to the order in which the abrupt change points appear on the pipeline. This coordinate list is the anomaly location.

[0057] To identify the number of connected components and loops in a topological graph, combine these numbers to form a topological graph category. Specifically, in the topological graph, starting from any vertex, move along connecting lines to all reachable vertices, grouping these vertices and lines into a connected component. Repeat this process until all vertices have been visited, counting the total number of connected components. Next, find a closed loop formed by a number of lines, where the loop does not contain other lines. Each such loop is counted as a loop, and the total number of loops is counted. Combine the total number of connected components and the total number of loops, listing them first and then the number of loops, into a single name; this combined name is the topological graph category.

[0058] The abnormal location and topology category are combined into an early warning message and sent to the terminal device of the operation and maintenance personnel. Specifically, each coordinate value in the abnormal location coordinate list is written into the beginning of the early warning message text, and the topology category is written into the end of the early warning message text, separated by a fixed separator to form a complete text message. This text message is then sent to the mobile phone or tablet carried by the operation and maintenance personnel via wireless network and displayed in the form of a pop-up window or notification bar message.

[0059] The sources and methods for obtaining the zero-dimensional Betty number, first-dimensional Betty number, and second-dimensional Betty number of a topological graph are as follows: The zero-dimensional Betty number is derived from the count of all independent connected segments in the topological graph. To obtain it, start from any vertex in the topological graph and move along connecting lines to all reachable vertices, grouping these vertices and lines into a single connected segment. Repeat this operation until all vertices have been visited; the total number of segments counted is the zero-dimensional Betty number. The first-dimensional Betty number is derived from the count of independently existing closed loops in the topological graph. To obtain it, search for closed loops formed by several line segments in the topological graph, where the loop does not contain other line segments. Count each such loop found; the number of all non-repeating loops is the first-dimensional Betty number. The second-dimensional Betty number is derived from the count of internal voids in the topological graph. To obtain it, find blank areas completely surrounded by the graph and not connected to the outside within the space occupied by the topological graph; count each such blank area. The total number of blank areas is the second-dimensional Betty number. The zero-dimensional, first-dimensional, and second-dimensional Betty numbers of the historical topological subset are obtained using the exact same method, except that the operation object is replaced from the topological morphology graph to the historical topological subset.

[0060] The significance of this calculation formula lies in using a unified numerical value to express the degree of homeomorphism between a topological graph and a historical topological subset. The formula subtracts the corresponding three Betti numbers from the three Betti numbers in the historical topological subset from the three Betti numbers of the topological graph. Then, it squares each difference to eliminate the influence of the sign, adds the three squared values, and takes the square root. The final value directly reflects the overall difference in topological structure between the two graphs. A value of zero indicates that the two graphs are completely identical in the three topological features: the number of zero-dimensional connected segments, the number of first-dimensional loops, and the number of second-dimensional holes; therefore, the two graphs are homeomorphic. A value greater than zero indicates that at least one topological feature is different; therefore, the two graphs are not homeomorphic. In this way, the formula transforms the complex problem of comparing continuous graph deformations into a comparison problem of three integers, allowing the determination of homeomorphism using a single numerical value.

[0061] The trend of this calculation formula is that as the difference in Betti numbers between the topological graph and the historical topological subset increases, the homeomorphism dissimilarity also increases monotonically. Specifically: when the Betti numbers of two graphs are equal in all three dimensions, the homeomorphism dissimilarity is zero, and they are considered homeomorphic. When the Betti numbers of two graphs differ in any dimension, the square of the difference is positive, the sum of the three squares is greater than zero, and the square root of the difference results in a positive homeomorphism dissimilarity. If the Betti numbers differ by one unit in only one dimension, the homeomorphism dissimilarity is equal to the absolute value of that unit difference. If the Betti numbers differ simultaneously in two or three dimensions, the homeomorphism dissimilarity increases according to the square root of the sum of squares, and the contributions of the differences in each dimension to the final value are additive. The homeomorphism dissimilarity has no negative value; its minimum value is fixed at zero, and its maximum value depends on the actual range of Betti numbers.

[0062] The beneficial effects are as follows: By collecting multiple sets of monitoring data during the historical healthy operation of oil and gas pipelines, calculating the consistency residual sequence distributed along the pipeline space, and extracting historical local curvature abrupt change points, the spatial connectivity of each historical local curvature abrupt change point forms a historical topological subset. The set of all historical topological subsets is used as the allowable manifold, providing a complete healthy topological reference benchmark for the current state. The spatial connectivity of the current local curvature abrupt change points is extracted to form a topological morphology map. This map is then continuously deformed and compared with each historical topological subset in the allowable manifold. When the topological morphology map cannot be deformed into the shape of any historical topological subset through stretching, compression, and bending, it is determined to be a different embryo, generating an anomaly marker. This process does not rely on any preset thresholds or fault samples; it is based solely on the topological characteristics of the graphic that remain unchanged under continuous deformation, eliminating the influence of operating condition changes on the early warning results. By calculating the homeomorphism difference degree and comparing its relationship with zero, a clear determination of abnormal states is achieved. Then, the abnormal location and topology morphology category are extracted and combined to form early warning information, which is output to the operation and maintenance terminal. This makes the early warning result include both the specific spatial coordinates of the abnormal occurrence and the description of the abnormal topology, which makes it easier for operation and maintenance personnel to quickly locate the fault location and understand the structural characteristics of the abnormality, thereby improving the availability of early warning information and the efficiency of on-site handling.

[0063] like Figure 2 The diagram shown is a functional block diagram of an IoT monitoring data anomaly early warning system based on digital twins provided in an embodiment of the present invention.

[0064] The IoT monitoring data anomaly early warning system 100 based on digital twins described in this invention can be installed in an electronic device. Depending on the functions implemented, the IoT monitoring data anomaly early warning system 100 based on digital twins may include a consistency residual generation module 101, a curvature abrupt change point extraction module 102, and a topological anomaly determination module 103. The module described in this invention can also be referred to as a unit, which refers to a series of computer program segments that can be executed by the processor of an electronic device and can perform a fixed function, and which are stored in the memory of the electronic device.

[0065] In this embodiment, the functions of each module / unit are as follows: The consistency residual generation module 101 is used to construct a digital twin by using the monitoring data collected by the IoT sensor in a healthy state as a constraint, substitute the real-time data in the IoT sensor into the physical control equation of the digital twin, invert two inversion values ​​of the same target physical parameter, and calculate the consistency residual between the two inversion values. The curvature abrupt change point extraction module 102 is used to extract the structural features of the consistency residual sequence, the structural features including the local curvature abrupt change points of the consistency residual in the spatial distribution of the oil and gas pipeline. The topological anomaly determination module 103 is used to compare the structural features with the permissible manifold composed of consistent residual sequences under historical healthy states. When the topological features of a local curvature abrupt change point and any subset of the permissible manifold do not satisfy the homeomorphic relationship, the oil and gas pipeline is determined to be in an abnormal state and an early warning message is output.

[0066] In the several embodiments provided by this invention, it should be understood that the disclosed methods and systems can be implemented in other ways. For example, the system embodiments described above are merely illustrative; for instance, the division of modules is only a logical functional division, and other division methods may be used in actual implementation.

[0067] The modules described as separate components may or may not be physically separate. The components shown as modules may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs.

[0068] Furthermore, the functional modules in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or in the form of hardware plus software functional modules.

[0069] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0070] This application embodiment can acquire and process relevant data based on artificial intelligence technology. Artificial intelligence is the theory, method, technology, and application system that uses digital computers or machines controlled by digital computers to simulate, extend, and expand human intelligence, perceive the environment, acquire knowledge, and use that knowledge to obtain optimal results.

[0071] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. A method for monitoring data anomaly early warning based on digital twinning, characterized in that, The method includes: P1. Using the monitoring data collected by the IoT sensor in a healthy state as a constraint, construct a digital twin, substitute the real-time data in the IoT sensor into the physical control equation of the digital twin, invert two inversion values ​​of the same target physical parameter, and calculate the consistency residual between the two inversion values. P2. Extract the structural features of the consistency residual sequence, the structural features including the local curvature abrupt change points of the consistency residual in the spatial distribution of the oil and gas pipeline; P3. Compare the structural features with the permissive manifold composed of consistent residual sequences under historical healthy states. When the topology of the local curvature abrupt change point and any subset of the permissive manifold do not satisfy the homeomorphic relationship, determine that the oil and gas pipeline is currently in an abnormal state and output a warning message.

2. The method for early warning of abnormal IoT monitoring data based on digital twins as described in claim 1, characterized in that, The construction of a digital twin, constrained by data collected by IoT sensors in a healthy state, includes: Multiple sets of monitoring data continuously collected by the IoT sensor in a healthy state are obtained, and the multiple sets of monitoring data are timestamped to obtain the health data sequence of the IoT sensor. Using each set of data in the health data sequence as input boundary values, an initial virtual structure consistent with the geometric contour of the oil and gas pipeline is established in the digital space; All spatial nodes in the initial virtual structure are bound to the corresponding sensor locations in the health data sequence to form a data-space mapping relationship; Based on the data-space mapping relationship, the parameter value range of each node in the initial virtual structure under the healthy state is locked to obtain the digital twin of the oil and gas pipeline.

3. The method for early warning of abnormal IoT monitoring data based on digital twins as described in claim 2, characterized in that, The step of establishing an initial virtual structure in digital space that matches the geometric contour of the oil and gas pipeline, using each set of data in the health data sequence as input boundary values, includes: Read the spatial coordinates of the sensor on the oil and gas pipeline corresponding to each group of data in the health data sequence; Based on the spatial coordinates, a discrete node array matching the external dimensions of the oil and gas pipeline is generated in the digital space. The measured value of each group of data in the health data sequence is assigned to the corresponding position node in the discrete node array, which serves as the activity boundary constraint value of the corresponding position node. Adjacent discrete nodes are connected by edges to form a gridded virtual geometry, which is then used as the initial virtual structure.

4. The method for early warning of abnormal IoT monitoring data based on digital twins as described in claim 1, characterized in that, The step of substituting real-time data from the IoT sensor into the physical control equations of the digital twin to derive two inversion values ​​of the same target physical parameter, and calculating the consistency residual between the two inversion values, includes: The system acquires real-time data from the IoT sensor and simultaneously inputs the real-time data into the first and second physical control equations contained in the digital twin. The first physical control equation outputs the first inversion estimate corresponding to the target physical parameters based on the real-time data. The second physical control equation outputs a second inversion estimate corresponding to the target physical parameters based on the same real-time data; The first inversion estimate is subtracted from the second inversion estimate point by point to obtain the residual value at each time step, and the residual value is used as the consistency residual.

5. The method for early warning of abnormal IoT monitoring data based on digital twins as described in claim 1, characterized in that, The structural features extracted from the consistency residual sequence include local curvature abrupt change points in the spatial distribution of the consistency residuals in oil and gas pipelines, including: Repeatedly perform the consistency residual calculation to obtain a consistency residual sequence within a continuous time window, wherein the consistency residual is the difference between two inversion values ​​of the same target physical parameter; Extract the spatial location on the oil and gas pipeline corresponding to each residual value in the consistency residual sequence; According to the spatial location along the extension direction of the oil and gas pipeline, each residual value is arranged into a spatial distribution sequence. The residual values ​​at adjacent positions in the spatial distribution sequence are compared in turn, and the variation between adjacent residual values ​​is calculated. When the variation range between the residual value at any spatial location and the residual value at an adjacent location is greater than the variation range of the adjacent locations before and after the spatial location, the spatial location is determined to be a local curvature abrupt change point.

6. The method for early warning of abnormal IoT monitoring data based on digital twins as described in claim 1, characterized in that, The permissible manifold includes: During the historical healthy operation of the oil and gas pipeline, multiple sets of monitoring data were collected. For each set of monitoring data, a consistent residual sequence was calculated along the spatial distribution of the oil and gas pipeline, and the corresponding historical local curvature abrupt change point was extracted from each consistent residual sequence. The connection relationships of each historical local curvature abrupt change point on the spatial distribution of oil and gas pipelines are used to form a historical topological subset, and the set of all historical topological subsets is taken as an admissible manifold.

7. The method for early warning of abnormal IoT monitoring data based on digital twins as described in claim 1, characterized in that, The process of comparing the structural features with the permissive manifold composed of consistent residual sequences from historical healthy states, and determining that the oil and gas pipeline is currently in an abnormal state and outputting early warning information when the topology of the local curvature abrupt change point and any subset of the permissive manifold do not satisfy the homeomorphic relation, includes: Extract the connection relationship of the local curvature abrupt change points in the spatial distribution of oil and gas pipelines to form a topological map of the local curvature abrupt change points; Extract the permissive manifold composed of consistent residual sequences under historical health states, and perform continuous deformation comparison between the topological morphology graph and each historical topological subset in the permissive manifold; When the topology diagram cannot be deformed into the shape of any historical topology subset through continuous stretching, compression and bending, it is determined that the topology diagram is different from the historical topology subset, and an anomaly mark of the current state of the oil and gas pipeline is generated. Based on the anomaly marker, an early warning message containing the anomaly location and the topology map category is generated and output to the operation and maintenance terminal.

8. The method for early warning of abnormal IoT monitoring data based on digital twins as described in claim 7, characterized in that, When the topology diagram cannot be deformed into the shape of any historical topology subset through continuous stretching, compression, and bending, it is determined that the topology diagram is different from the historical topology subset, and the current state of the oil and gas pipeline is marked as abnormal, including: The homeomorphism degree between the topological graph and the historical topological subset is calculated according to the following formula: ； In the formula, These are the zeroth, first, and second dimensions of the topological graph, respectively; These are the zeroth, first, and second dimensions of the historical topological subset, respectively; The degree of homeomorphism; When the homeomorphism difference is greater than zero, the topological morphology graph is determined to be a different homeomorphism from the historical topological subset.

9. The method for early warning of abnormal IoT monitoring data based on digital twins as described in claim 7, characterized in that, Based on the anomaly marker, an early warning message containing the anomaly location and the topology map category is generated and output to the operation and maintenance terminal, including: Extract the spatial coordinates of the local curvature abrupt change point on the oil and gas pipeline, and use the spatial coordinates as the abnormal location; Identify the number of connected branches and the number of loop structures in the topology graph, and combine the number of connected branches and the number of loop structures into the topology graph category; The abnormal location and the topology map category are combined into a warning message, which is then sent to the terminal device of the operation and maintenance personnel.

10. An IoT monitoring data anomaly early warning system based on digital twins, characterized in that, The system for implementing the IoT monitoring data anomaly early warning method based on digital twins as described in claim 1, the system comprising: The consistency residual generation module is used to construct a digital twin based on the monitoring data collected by the IoT sensor in a healthy state. The module substitutes the real-time data from the IoT sensor into the physical control equation of the digital twin to derive two inversion values ​​of the same target physical parameter and calculates the consistency residual between the two inversion values. A curvature abrupt change point extraction module is used to extract structural features of the consistency residual sequence, wherein the structural features include local curvature abrupt change points of the consistency residual in the spatial distribution of the oil and gas pipeline. The topological anomaly determination module is used to compare the structural features with the permissible manifold composed of consistent residual sequences from historical healthy states. When the topological features of a local curvature abrupt change point and any subset of the permissible manifold do not satisfy the homeomorphic relationship, the oil and gas pipeline is determined to be in an abnormal state and an early warning message is output.

Images