Fault prediction method and device, electronic equipment and computer program product

Abstract

The invention discloses a fault prediction method and device, electronic equipment and a computer program product. The method comprises the steps that an equipment data set of an oil and gas transmission system at the first time is obtained, and the oil and gas transmission system comprises a plurality of pieces of equipment with a dependency relationship; a graph structure fault prediction model is used to analyze the operation data corresponding to the target equipment, the equipment state of the target equipment at the first time is obtained, the target equipment is any one of the multiple pieces of equipment, and the graph structure fault prediction model is used to process the graph structure data; determining a propagation state of the target device at the first time according to the device state of at least one adjacent device having a dependency relationship with the target device; and determining a fault prediction result of the target equipment at the second time according to the equipment state and the propagation state. According to the invention, the technical problem of insufficient accuracy of fault prediction in the oil and gas industry in the prior art is solved.

Description

Technical Field

[0001] The present invention relates to the field of data processing, and in particular to a fault prediction method, device, electronic equipment and computer program product. Background Art

[0002] In a complex SCADA system, there are close dependencies between different devices. When a device fails, the failure may spread to other devices along these dependencies, resulting in system-level failures. Most existing fault prediction methods cannot effectively capture and analyze such complex interactions, and thus cannot provide accurate and timely fault prediction.

[0003] For example, in the oil and gas industry, SCADA systems are widely used to monitor and control complex production and transportation processes. The system contains multiple key devices, such as pipeline valves, pump stations, compressors, sensors, and controllers. These devices have complex dependencies, and equipment failures not only affect their own functions, but may also cause system-level fault propagation, resulting in production interruptions and even safety accidents.

[0004] The existing mainstream solutions for SCADA system fault prediction and diagnosis mainly include the following categories:

[0005] The first type is a rule-based fault detection system, which detects abnormal conditions of equipment through predefined rules and thresholds. For example, the system can compare the sensor readings with the preset safety range, and trigger an alarm when a deviation from the range is detected. This method is simple and easy to implement, but it is mainly applicable to known fault types and patterns. The disadvantage of this solution is that the method usually cannot effectively capture the complex interactive relationships between devices. Especially in scenarios where multiple devices are linked, the propagation path of the fault is often very complex, and existing statistical models and rule systems are difficult to accurately predict these complex fault propagation paths.

[0006] The second category is the diagnostic platform based on expert systems, which perform fault diagnosis by encoding the knowledge and experience of domain experts. These systems usually rely on predefined rule bases and logical reasoning processes, and can handle some complex fault scenarios, but are difficult to expand and update to cope with new fault modes.

[0007] However, traditional fault prediction methods usually rely on rule bases or simple statistical models, which are difficult to handle these complex equipment interactions, resulting in insufficient accuracy and real-time performance of fault prediction.

[0008] With respect to the problem that the above-mentioned prior art technologies are insufficiently accurate in predicting faults in the oil and gas industry, no effective solution has yet been proposed. Summary of the invention

[0009] The embodiments of the present invention provide a fault prediction method, device, electronic device and computer program product to at least solve the technical problem that the prior art has insufficient accuracy in fault prediction for the oil and gas industry.

[0010] According to one aspect of an embodiment of the present invention, a fault prediction method is provided, comprising: obtaining a set of equipment data of an oil and gas transmission system at a first time, wherein the oil and gas transmission system comprises: a plurality of equipment having a dependency relationship, wherein the dependency relationship is at least used to represent a physical connection relationship or a data transmission relationship between the equipment, and the set of equipment data comprises: operation data of each equipment in the oil and gas transmission system; using a graph structure fault prediction model to analyze the operation data corresponding to a target equipment to obtain an equipment state of the target equipment at the first time, wherein the target equipment is any one of the plurality of equipment, and the graph structure fault prediction model is used to process graph structure data, wherein nodes in the graph structure data correspond one-to-one to the equipment in the oil and gas transmission system, and edges connecting the nodes in the graph structure data represent the dependency relationship in the oil and gas transmission system; determining a propagation state of the target equipment at the first time according to the equipment state of at least one adjacent equipment having the dependency relationship with the target equipment; and determining a fault prediction result of the target equipment at a second time according to the equipment state and the propagation state, wherein the second time is adjacent to the first time and the second time is later than the first time.

[0011] Optionally, determining, according to the device state and the propagation state, a fault prediction result of the target device at the second time includes: analyzing the device state and the propagation state using a preset propagation model to obtain a fault prediction result of the target device at the second time, wherein the preset propagation model is expressed as: Wherein, t represents the first time, h j (t) represents the device state of the device j at the first time, t+1 represents the second time, and h j (t+1) represents the fault prediction result of the device j at the second time, represents the propagation state at the first time, N(j) is the set of neighboring devices of the device j, i represents the neighboring device of the device j, C ij represents the connection strength between the device i and the device j, h i (t) represents the device state of the adjacent device i at the first time, W and B are learning parameters of the preset propagation probability model, and σ represents a nonlinear activation function.

[0012] Optionally, after determining the fault prediction result of the target device at the second time according to the device status and the propagation status, the method further includes: determining at least one faulty device of the oil and gas transmission system at the second time according to the fault prediction result; in the case where there are multiple faulty devices, analyzing the dependency relationship between the multiple faulty devices based on the graph structure data to obtain a core device, wherein the core device is the faulty device and the core device has the dependency relationship with the multiple faulty devices; and generating maintenance recommendations for the core device.

[0013] Optionally, before obtaining the equipment data set of the oil and gas transmission system at the first time, the method also includes: obtaining structural data of at least one of the oil and gas transmission systems, wherein the structural data is used to represent multiple devices in the oil and gas transmission system, and the dependency relationship between the devices; constructing the graph structure data corresponding to each of the oil and gas transmission systems based on the structural data, wherein, in the graph structure data, the nodes are used to represent the devices in the oil and gas transmission system, the edges connecting the nodes are used to represent the dependency relationship in the oil and gas transmission system, the node attributes of the nodes represent the device status, and the edge attributes of the edges represent the connection strength.

[0014] Optionally, after constructing the graph structure data corresponding to each of the oil and gas transmission systems according to the structure data, the method further includes: obtaining a training data set corresponding to each of the oil and gas transmission systems, wherein the training data set includes: a plurality of correct operation samples, and a plurality of faulty operation samples, the correct operation samples are determined according to the operation data of the device when the device status is a normal status, and the faulty operation samples are determined according to the operation data of the device when the device status is the faulty status; the correct operation samples and the faulty operation samples in the training data set are input into a preset graph structure fault prediction template corresponding to the oil and gas transmission system for training to obtain the graph structure fault prediction model, wherein the preset graph structure fault prediction template is established according to the graph structure data corresponding to the oil and gas transmission system.

[0015] Optionally, obtaining a training data set corresponding to each of the oil and gas transmission systems includes: obtaining an operation data set of each of the oil and gas transmission systems within a preset time period, wherein each of the operation data sets includes: in the same oil and gas transmission system, normal operation data of the equipment whose equipment status is the normal state, and fault operation data of the equipment whose equipment status is the fault state; performing feature engineering processing on multiple normal operation data within the same preset time period to obtain multiple correct operation features indicating that the equipment status is the normal state, wherein the correct operation features are used to determine the correct operation samples; performing feature engineering processing on multiple fault operation data within the same preset time period to obtain multiple fault operation features indicating that the equipment status is the fault state, wherein the fault operation features are used to determine the fault operation samples.

[0016] Optionally, after determining the fault prediction result of the target device at the second time according to the device status and the propagation status, the method further includes: obtaining a fault handling opinion for the oil and gas transmission system based on the fault prediction result, wherein the fault handling opinion is at least used to represent the actual device status of the device in the oil and gas transmission system at the second time; and updating the graph structure fault prediction model based on the actual device status of each device in the oil and gas transmission system at the second time.

[0017] According to another aspect of an embodiment of the present invention, a fault prediction device is further provided, comprising: an acquisition module, used to acquire a set of equipment data of an oil and gas transmission system at a first time, wherein the oil and gas transmission system comprises: a plurality of equipment having a dependency relationship, wherein the dependency relationship is at least used to represent a physical connection relationship or a data transmission relationship between the equipment, and the set of equipment data comprises: operation data of each equipment in the oil and gas transmission system; an analysis module, used to analyze the operation data corresponding to a target equipment using a graph structure fault prediction model to obtain an equipment state of the target equipment at the first time, wherein the target equipment is any one of the plurality of equipment, and the graph structure fault prediction model is used to process graph structure data, wherein nodes in the graph structure data correspond one-to-one to the equipment in the oil and gas transmission system, and edges connecting the nodes in the graph structure data represent the dependency relationship in the oil and gas transmission system; a first determination module, used to determine a propagation state of the target equipment at the first time according to the equipment state of at least one adjacent equipment having the dependency relationship with the target equipment; and a second determination module, used to determine a fault prediction result of the target equipment at a second time according to the equipment state and the propagation state, wherein the second time is adjacent to the first time and the second time is later than the first time.

[0018] According to another aspect of an embodiment of the present invention, there is further provided an electronic device, including a memory and a processor, wherein a computer program is stored in the memory, and the processor is configured to execute the above-mentioned fault prediction method through the computer program.

[0019] According to another aspect of an embodiment of the present invention, a non-volatile storage medium is provided, in which a computer program is stored, wherein the above-mentioned fault prediction method is executed by running the computer program in the device where the non-volatile storage medium is located.

[0020] In an embodiment of the present invention, a device data set of an oil and gas transmission system at a first time is obtained, wherein the oil and gas transmission system includes: a plurality of devices having a dependency relationship, the dependency relationship is used to at least represent a physical connection relationship or a data transmission relationship between the devices, and the device data set includes: operating data of each device in the oil and gas transmission system; using a graph structure fault prediction model to analyze the operating data corresponding to the target device, and obtain the device state of the target device at the first time, wherein the target device is any one of the plurality of devices, and the graph structure fault prediction model is used to process graph structure data, the nodes in the graph structure data correspond one-to-one to the devices in the oil and gas transmission system, and the edges connecting the nodes in the graph structure data Representing the dependency relationship in the oil and gas transmission system; determining the propagation state of the target device at the first time according to the device state of at least one adjacent device having a dependency relationship with the target device; determining the fault prediction result of the target device at the second time according to the device state and the propagation state, wherein the second time is adjacent to the first time and the second time is later than the first time, so that according to the operating data of each device in the oil and gas transmission system at the first time, combined with the dependency relationship between the devices, the fault prediction result of the target device at the second time can be more accurately predicted, thereby achieving the technical effect of improving the accuracy of fault prediction, and further solving the technical problem of insufficient accuracy of fault prediction in the oil and gas industry in the prior art. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] The drawings described herein are used to provide a further understanding of the present invention and constitute a part of this application. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute an improper limitation of the present invention. In the drawings:

[0022] Figure 1 is a flow chart of a fault prediction method according to an embodiment of the present invention;

[0023] Figure 2 is a schematic diagram of an implementation process of a fault prediction model based on a graph neural network according to an embodiment of the present invention;

[0024] Figure 3is a schematic diagram of a process for predicting and maintaining oil and gas pipeline valve faults according to an embodiment of the present invention;

[0025] Figure 4 is a schematic diagram of a graph structure of each device in a SCADA system according to an embodiment of the present invention;

[0026] Figure 5 is a schematic diagram of a fault prediction device according to an embodiment of the present invention;

[0027] Figure 6 It is a structural block diagram of a computer terminal according to an embodiment of the present invention. DETAILED DESCRIPTION

[0028] In order to enable those skilled in the art to better understand the scheme of the present invention, the technical scheme in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present invention.

[0029] It should be noted that the terms "first", "second", etc. in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments of the present invention described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0030] First, some nouns or terms that appear in the description of the embodiments of the present application are subject to the following explanations:

[0031] SCADA system: Supervisory control and data acquisition system used for industrial process automation control.

[0032] Graph Neural Network (GNN): A neural network model specifically designed to process graph-structured data. In many practical applications, data naturally exists in the form of graphs, such as social networks, molecular structures, and transportation networks. GNN can effectively capture the complex relationships between nodes in the graph and learn the high-dimensional representation (feature vector) of each node through the connection relationship between nodes. This makes GNN particularly suitable for solving problems such as node classification, graph classification, and link prediction.

[0033] Graph data structure: Very suitable for representing and processing data sets with complex relationships and interactions. In many fields, data does not only exist in isolation, but is connected to each other through various types of relationships. These relationships themselves often carry important information and are crucial for understanding and analyzing the entire data system.

[0034] Node: An element in a graph, which can be any entity, such as a device, site, molecule, etc.

[0035] Edge: A line connecting two nodes in a graph, representing the relationship or interaction between the two nodes, such as friendship, chemical bonds, etc.

[0036] Graph: An overall data structure consisting of nodes and edges, which can be directed or undirected.

[0037] Propagation probability: The propagation probability is usually calculated based on the edge attributes in the graph and the states of the connected nodes. In the context of fault propagation, the propagation probability P ij It can be defined as the probability of fault propagation from node i to node j, where i and j are connected by an edge.

[0038] According to an embodiment of the present invention, a fault prediction method embodiment is provided. It should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and although a logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0039] Figure 1 is a flow chart of a fault prediction method according to an embodiment of the present invention. Figure 1 As shown, the method comprises the following steps:

[0040] Step S102, obtaining a device data set of the oil and gas transmission system at the first time, wherein the oil and gas transmission system includes: a plurality of devices having a dependency relationship, the dependency relationship is at least used to indicate a physical connection relationship or a data transmission relationship between the devices, and the device data set includes: operation data of each device in the oil and gas transmission system;

[0041] Step S104, using the graph structure fault prediction model to analyze the operation data corresponding to the target device to obtain the device status of the target device at the first time, wherein the target device is any one of the multiple devices, and the graph structure fault prediction model is used to process the graph structure data, the nodes in the graph structure data correspond one-to-one to the devices in the oil and gas transmission system, and the edges connecting the nodes in the graph structure data represent the dependency relationship in the oil and gas transmission system;

[0042] Step S106, determining a propagation state of the target device at a first time according to a device state of at least one neighboring device having a dependency relationship with the target device;

[0043] Step S108, determining a fault prediction result of the target device at a second time according to the device state and the propagation state, wherein the second time is adjacent to the first time and the second time is later than the first time.

[0044] In an embodiment of the present invention, a device data set of an oil and gas transmission system at a first time is obtained, wherein the oil and gas transmission system includes: a plurality of devices having a dependency relationship, the dependency relationship is used to at least represent a physical connection relationship or a data transmission relationship between the devices, and the device data set includes: operating data of each device in the oil and gas transmission system; using a graph structure fault prediction model to analyze the operating data corresponding to the target device, and obtain the device state of the target device at the first time, wherein the target device is any one of the plurality of devices, and the graph structure fault prediction model is used to process graph structure data, the nodes in the graph structure data correspond one-to-one to the devices in the oil and gas transmission system, and the edges connecting the nodes in the graph structure data Representing the dependency relationship in the oil and gas transmission system; determining the propagation state of the target device at the first time according to the device state of at least one adjacent device having a dependency relationship with the target device; determining the fault prediction result of the target device at the second time according to the device state and the propagation state, wherein the second time is adjacent to the first time and the second time is later than the first time, so that according to the operating data of each device in the oil and gas transmission system at the first time, combined with the dependency relationship between the devices, the fault prediction result of the target device at the second time can be more accurately predicted, thereby achieving the technical effect of improving the accuracy of fault prediction, and further solving the technical problem of insufficient accuracy of fault prediction in the oil and gas industry in the prior art.

[0045] In the above step S102, the operation data in the equipment data set may be acquired through a SCADA system, and the SCADA system may be a supervisory control and data acquisition system for realizing automated control of industrial processes.

[0046] In the above step S104, the graph structure data may use nodes to represent the equipment in the oil and gas transmission system, and use edges to represent the dependency relationship between the equipment.

[0047] In the above step S104, the graph structure fault prediction model is a pre-trained fault prediction model based on a graph neural network (GNN), and the graph structure fault prediction model is used to process graph structure data.

[0048] In the above step S104, the device state at least includes: a fault state; the dependency relationship is also used to represent the propagation path of the fault state and the connection strength between the devices, and the propagation state is determined according to the connection strength.

[0049] In the above step S106, the propagation state indicates that the target device is affected by the device state of the neighboring device, resulting in that the device state of the target device may change at the second time.

[0050] In the above step S108, the fault prediction result may represent the device state of the target device at the second time.

[0051] As an optional embodiment, determining the fault prediction result of the target device at the second time according to the device state and the propagation state includes: analyzing the device state and the propagation state using a preset propagation model to obtain the fault prediction result of the target device at the second time, wherein the preset propagation model is expressed as: Where t represents the first time h j (t) represents the device state of device j at the first time, t+1 represents the second time, and h j (t+1) represents the fault prediction result of device j at the second time, represents the propagation state at the first time, N(j) is the set of neighboring devices of device j, i represents the neighboring devices of device j, C ij represents the connection strength between device i and device j, h i (t) represents the device state of the adjacent device i at the first time, W and B are the learning parameters of the preset propagation probability model, and σ represents the nonlinear activation function.

[0052] In the above-mentioned embodiments of the present application, the device state of the target device at the second time will be affected by the device state of the neighboring devices of the target device at the first time, and the influence can be represented by the propagation state, and then the fault prediction result of the target device at the second time can be determined by a preset propagation model. The preset propagation model can determine the propagation state of the target device at the first time based on the device state of the neighboring devices of the target device at the first time, and combine the device state of the target device at the first time to jointly determine the device state of the target device at the second time, and obtain the fault prediction result of the target device.

[0053] As an optional embodiment, after determining the fault prediction result of the target device at the second time according to the device status and the propagation status, the method also includes: determining at least one faulty device in the oil and gas transmission system at the second time according to the fault prediction result; in the case where there are multiple faulty devices, analyzing the dependency relationship between the multiple faulty devices based on the graph structure data to obtain the core device, wherein the core device is the faulty device and the core device has a dependency relationship with multiple faulty devices; and generating maintenance recommendations for the core device.

[0054] In the above-mentioned embodiment of the present application, in the oil and gas transmission system, if a certain device has a dependency relationship with multiple devices, then when the device fails, the devices that have a dependency relationship with the device may also fail. Therefore, after determining the fault prediction result of each device in the oil and gas transmission system at the second time, the faulty device predicted by the oil and gas transmission system at the second time can be obtained, and then the faulty device can be analyzed according to the dependency relationship indicated in the graph structure data. Among the faulty devices with dependencies, a core device that has a dependency relationship with multiple faulty devices can be determined, and it is predicted that multiple faulty devices in the oil and gas transmission system may be caused by the failure of the core device, and then a maintenance suggestion for the core device is generated. By maintaining the core device, the core device can be restored to normal, and then the faulty devices that have a dependency relationship with the core device may also be restored to normal.

[0055] As an optional embodiment, before obtaining the equipment data set of the oil and gas transmission system at the first time, the method also includes: obtaining structural data of at least one oil and gas transmission system, wherein the structural data is used to represent multiple devices in the oil and gas transmission system and the dependency relationship between the devices; constructing graph structure data corresponding to each oil and gas transmission system based on the structural data, wherein in the graph structure data, nodes are used to represent the equipment in the oil and gas transmission system, edges connecting the nodes are used to represent the dependency relationship in the oil and gas transmission system, the node attributes of the nodes represent the equipment status, and the edge attributes of the edges represent the connection strength.

[0056] In the above-mentioned embodiments of the present application, the graph structure data may be determined based on the structure between the devices in the oil and gas transmission system. In the graph structure data, the devices in the oil and gas transmission system may be represented by nodes, the node attributes of each node may represent the device status of the corresponding device, the edges connecting the nodes may represent the dependency relationship between the devices in the oil and gas transmission system, and the edge attributes may represent the connection strength of the devices over time, so that the graph structure data may be used to represent the structure between the devices in the oil and gas transmission system.

[0057] It should be noted that different oil and gas transmission systems may correspond to different graph structure data. Based on each graph structure data, a graph structure fault prediction model corresponding to each oil and gas transmission system may be trained.

[0058] As an optional embodiment, after constructing the graph structure data corresponding to each oil and gas transmission system according to the structural data, the method also includes: obtaining a training data set corresponding to each oil and gas transmission system, wherein the training data set includes: multiple correct operation samples, and multiple fault operation samples, the correct operation samples are determined according to the operation data of the equipment in a normal state, and the fault operation samples are determined according to the operation data of the equipment in a fault state; the correct operation samples and the fault operation samples in the training data set are input into the preset graph structure fault prediction template corresponding to the oil and gas transmission system for training to obtain a graph structure fault prediction model, wherein the preset graph structure fault prediction template is established according to the graph structure data corresponding to the oil and gas transmission system.

[0059] In the above-mentioned embodiments of the present application, a preset graph structure fault prediction template can be established according to the graph structure data corresponding to the oil and gas transmission system. Different oil and gas transmission systems can correspond to different preset graph structure fault prediction templates. The training data set of each oil and gas transmission system is input into the corresponding preset graph structure fault prediction template, and the parameters in the preset graph structure fault prediction template can be optimized to obtain the graph structure fault prediction model corresponding to each oil and gas transmission system, thereby realizing the training of the graph structure fault prediction model.

[0060] As an optional embodiment, obtaining a training data set corresponding to each oil and gas transmission system includes: obtaining an operation data set for each oil and gas transmission system within a preset time period, wherein each operation data set includes: normal operation data of equipment in a normal state, and fault operation data of equipment in a fault state in the same oil and gas transmission system; performing feature engineering processing on multiple normal operation data within the same preset time period to obtain multiple correct operation features indicating that the equipment state is a normal state, wherein the correct operation features are used to determine correct operation samples; performing feature engineering processing on multiple fault operation data within the same preset time period to obtain multiple fault operation features indicating that the equipment state is a fault state, wherein the fault operation features are used to determine fault operation samples.

[0061] Optionally, the preset time period may be a historical time period that has occurred.

[0062] In the above-mentioned embodiments of the present application, the training data set corresponding to each oil and gas transmission system can be determined according to the operation data set of the oil and gas transmission system within a preset time period. By analyzing the operation data of the oil and gas transmission system within the preset time period, the normal operation data that puts the equipment in a normal state and the fault operation data that puts the equipment in a faulty state can be determined. Then, the normal operation data and the fault operation data are processed using feature engineering respectively. Based on multiple normal operation data, the correct operation features that can put the equipment in a normal state can be determined, and positive samples in the training data set, that is, correct operation samples, can be constructed based on the correct operation features; and fault operation features that can put the equipment in a faulty state can be determined based on multiple fault operation data, and negative samples in the training data set, that is, fault operation samples, can be constructed based on the fault operation features, thereby realizing the construction of the training data set.

[0063] As an optional embodiment, after determining the fault prediction result of the target device at the second time according to the device status and the propagation status, the method also includes: obtaining a fault handling opinion for the oil and gas transmission system based on the fault prediction result, wherein the fault handling opinion is at least used to represent the actual device status of the equipment in the oil and gas transmission system at the second time; and updating the graph structure fault prediction model based on the actual device status of the target device in the oil and gas transmission system at the second time.

[0064] In the above-mentioned embodiment of the present application, after the fault prediction result of the target device at the second time is obtained, the staff processes and verifies it according to the fault prediction result, and can obtain fault handling opinions for the oil and gas transmission system. Then, the actual device status of the target device at the second time fed back in the fault handling opinions is used as training data for the graph structure fault prediction model, and the graph structure fault prediction model is updated to realize the updating and maintenance of the graph structure fault prediction model.

[0065] The present invention also provides a preferred embodiment, which provides a fault prediction model based on graph neural network, which can automatically capture the topological structure and interaction relationship between devices in the SCADA system, predict possible fault propagation paths by analyzing the interaction relationship between devices, and realize dynamic prediction of fault propagation paths and generation of preventive maintenance suggestions.

[0066] Figure 2 is a schematic diagram of an implementation process of a fault prediction model based on a graph neural network according to an embodiment of the present invention, such as Figure 2 As shown, the specific steps include:

[0067] Step S21, graph structure modeling.

[0068] Optionally, graph structure modeling includes at least: defining nodes, selecting features and defining edges, wherein defining nodes specifically includes: defining each device (such as valves, sensors, controllers) as a node in the graph structure; the selected features include at least: device type, historical fault data, normal operating parameters (such as temperature, pressure) and device operation history; the defined edges represent physical connections or data dependencies between devices, for example, a controller may directly control multiple valves, and these control relationships can be represented by edges in the graph structure; wherein the features of the edges may include: connection type, signal strength or data transmission frequency.

[0069] Step S22: data preprocessing.

[0070] Optionally, the specific process of data preprocessing includes at least: data collection, data cleaning and feature engineering.

[0071] Optionally, data collection specifically includes: ensuring that the data collection system can be seamlessly integrated with other components of SCADA such as control systems, alarm systems, etc., to ensure the integrity and real-time nature of the data, and using the built-in data logger of the SCADA system to collect real-time operating data of key equipment such as temperature, pressure, flow rate, etc.

[0072] Optionally, data cleaning includes: missing value processing and outlier processing. For missing value processing, the missing points in the data can be identified, the pattern and cause of the missing data can be analyzed, and then the missing values ​​can be filled using historical averages or time series prediction methods (such as linear interpolation or regression methods) based on the equipment type and historical operating data. For outlier processing, statistical methods (such as Z-score, IQR) can be used to identify outliers in the data, and then, based on the nature of the outliers, it can be decided whether to correct or delete these data points to avoid misleading the model training.

[0073] Optionally, feature engineering includes: creating derived features and obtaining time window features, where creating derived features includes: calculating equipment operating efficiency based on raw data, such as energy consumption ratio (ratio of energy consumption to output). These derived features can help the model better understand the relationship between equipment operating status and fault occurrence; obtaining time window features includes: calculating statistical values ​​(average, maximum, minimum) for data within the past 24 hours, which can help the model capture short-term trends and periodic changes in equipment operation.

[0074] Step S23, GNN model construction and training.

[0075] Optionally, the specific process of GNN model construction and training includes at least: preparing a training data set and model training.

[0076] Optionally, prepare a training data set, specifically including: data labeling, labeling the training set using historical fault records and normal operation data, and ensuring that the data set contains sufficient fault cases and diversity to improve the generalization ability of the model.

[0077] Optionally, model training specifically includes: batch training, hyperparameter optimization and model validation and testing, wherein batch training specifically includes: using batch processing methods to process large-scale graph data to optimize computing resources and improve training efficiency; hyperparameter optimization specifically includes: adjusting the model's hyperparameters (such as learning rate, number of layers, number of hidden units) through cross-validation and other techniques to find the best model configuration; model validation and testing specifically includes: testing model performance on an independent validation set to ensure that the model can maintain good prediction accuracy on unseen data.

[0078] Step S24: fault prediction and real-time monitoring.

[0079] Optionally, the specific process of fault prediction and real-time monitoring includes at least: real-time data input, fault prediction and propagation probability analysis.

[0080] Optionally, the real-time data input specifically includes: inputting the real-time collected equipment data into the model for fault prediction.

[0081] Optionally, fault prediction specifically includes: using a model to predict the possible fault state of each device and the corresponding confidence level, and automatically marking high-risk devices and possible fault propagation paths based on the prediction results.

[0082] Optionally, the propagation probability analysis specifically includes: when calculating the propagation probability, factors considered include at least: node status, edge attributes, and node and edge characteristics.

[0083] Optionally, the node state, which may be the current state of the node (eg, healthy, faulty, etc.), may directly affect the probability of fault propagation. Typically, a node's fault state increases its probability of propagating the fault to neighboring nodes.

[0084] Optionally, edge attributes, which may indicate the strength or type of connections between nodes, may affect how easily faults propagate.

[0085] Optionally, the characteristics of the nodes and edges may include: historical fault data of the equipment, operating conditions, environmental factors, etc.

[0086] As an alternative example, in GNN, the calculation of propagation probability is usually closely related to the feature update of the node. Consider a simplified model, where the state h of node j is j (t+1)At time t+1, it is determined by the state of itself and neighbor node i at time t. The specific expression formula can be:

[0087]

[0088] Among them, N(j) is the set of neighbor nodes of node j, C ij is the connection strength between nodes i and j, W and B are learning parameters, and σ is a nonlinear activation function.

[0089] Under the above setting, the fault propagation probability P ij By changing h j (t+1) The calculation result of is interpreted as the probability that node j is in a faulty state due to the influence of node i in the next time step.

[0090] Step S25: Maintenance suggestions and responses.

[0091] Optionally, the specific process of maintenance suggestion and response includes at least: an early warning system and maintenance suggestion.

[0092] Optionally, the early warning system specifically includes: when the model predicts a high-risk failure, automatically triggering an early warning to notify the maintenance team.

[0093] Optionally, maintenance recommendations include: generating specific maintenance recommendations for each fault type, such as replacing parts, adjusting parameters, etc. Providing priority sorting to guide the maintenance team to prioritize the most serious or most impactful issues.

[0094] Step S26: model iteration and feedback.

[0095] Optionally, the specific process of model iteration and feedback includes at least: feedback mechanism and continuous optimization.

[0096] Optionally, the feedback mechanism specifically includes: feeding back the results of fault handling into the model for retraining and fine-tuning the model.

[0097] Optionally, continuous optimization, including: regular evaluation of model performance, adjustment of models based on new data and failure modes, and use of online learning techniques to respond to changes in equipment and system configurations.

[0098] As an optional example, the above-mentioned fault prediction model based on graph neural network can be used for oil and gas pipeline valve fault prediction and maintenance.

[0099] In the process of oil and gas transportation, pipeline valves are crucial control equipment. Their failure may lead to fluid leakage, transportation interruption, and even cause serious problems such as environmental pollution. Assuming that a pipeline valve (node ​​A) in a SCADA system has an abnormal operation, the system uses the GNN model to analyze the relationship between the device and its surrounding equipment and predict the possible impact of the failure.

[0100] Figure 3 is a schematic diagram of a process for predicting and maintaining oil and gas pipeline valve faults according to an embodiment of the present invention, such as Figure 3 As shown, it specifically includes: modeling the equipment according to the actual physical relationship of the equipment to obtain the GNN model; identifying node anomalies through the GNN model and performing message transmission; then, alarming the remote maintenance and diagnosis platform of the SCADA system can be performed through message transmission, and key equipment can be analyzed through message transmission, and maintenance suggestions can be generated. Then, the GNN model is adaptively optimized according to the maintenance suggestions. The GNN model can be continuously optimized, analyze the fault propagation path, realize dynamic prediction of fault propagation probability, and automatically generate maintenance suggestions. Only parameters can be adjusted.

[0101] Optionally, the devices in the SCADA system include: pipeline valves, controllers, sensors, pump stations and other devices. These devices interact with each other through data transmission and control signals, and the actual physical relationship between the devices can be obtained. The system models these devices into a graph structure (such as the graph structure used in the GNN model) according to the actual physical relationship between the devices, where each device is a node in the graph, and the physical connection and data dependency between the devices are the edges in the graph structure.

[0102] Figure 4 is a schematic diagram of a graph structure of each device in a SCADA system according to an embodiment of the present invention, such as Figure 4 As shown in the figure, the pipeline valve (node ​​A) is connected to the controller (node ​​B) through a data link, and the controller is connected to a series of sensors (nodes C and D). The actual physical relationship between the devices is represented by a graph structure, so that the GNN model can accurately capture the interaction relationship between the devices.

[0103] Optionally, after the graph structure modeling is completed, the GNN model performs diffusion analysis on the fault information of the pipeline valve through a multi-round message passing mechanism. Assuming that the valve (node ​​A) is abnormal, the GNN model will analyze the paths through which the abnormality may propagate to other key devices. For example, the abnormality of the valve may affect the operation of the controller (node ​​B), thereby causing the sensors (nodes C and D) connected to the controller to be affected. By analyzing these relationships, the model identifies potential paths for fault propagation and predicts the probability of fault propagation for each path.

[0104] Optionally, through the analysis of the GNN model, the system identifies the most critical nodes in the fault propagation path. For example, the controller (node ​​B) may be the core node of the fault propagation. Once the controller is affected, all sensors connected to it may fail. Therefore, the system generates maintenance recommendations, suggesting that the operator should prioritize the inspection and maintenance of the controller (node ​​B) while monitoring the connected sensors (nodes C and D). This preventive maintenance recommendation can help the operator take measures before the fault spreads and avoid system-level failures.

[0105] Optionally, during the implementation process, the operating status of the SCADA system is dynamically changing, new data is constantly pouring in, and the status of the equipment may change at any time. The GNN model has real-time learning and adaptive adjustment capabilities, and can continuously update the prediction results of the fault propagation path based on the latest equipment data and operating status. For example, if the fault of node A is found to have further deteriorated during the monitoring process, the model can update the predicted path in time, prompting the operator to take more urgent maintenance measures immediately. At the same time, the model can also be optimized based on the long-term operating data of the system to improve the accuracy and response speed of future fault predictions.

[0106] As another optional example, the above-mentioned fault prediction model based on graph neural network can be used for fault prediction and maintenance applications of oil pipeline PLC control systems.

[0107] Alternatively, assume that in a long-distance oil pipeline, the PLC control system is responsible for managing multiple key devices, including: pump stations, pressure sensors, valves, compressors, and flow meters along the pipeline. The core PLC (node ​​A) in the system is connected to these devices through signal lines to monitor and control the oil transportation process. If a device in the system fails, it may cause crude oil transportation to be interrupted, affecting the operation of the entire pipeline. Therefore, it is crucial to predict these failures in advance and take corresponding maintenance measures.

[0108] As an optional example, in the PLC control system of an oil pipeline, the core PLC (node ​​A) forms a complex control network with the pump station (node ​​B), pressure sensor (node ​​C), flow meter (node ​​D), valve (node ​​E), and compressor (node ​​F) along the pipeline. The system models this network as a graph structure, in which each device (such as PLC, sensor, valve) is a node in the graph, and the signal transmission or control relationship between different devices is an edge. For example, the PLC (node ​​A) is connected to the pressure sensor (node ​​C) and flow meter (node ​​D) through control signals, and these edges represent the transmission path of data or control signals. Through this graph structure representation, the system can clearly capture the complex interaction relationship between devices.

[0109] As an optional example, when an abnormality occurs in a device in the system, such as a pressure sensor (node ​​C) or a pump station (node ​​B), the GNN model predicts along which paths the fault may propagate by analyzing the dependencies between the device and other devices. For example, if pressure sensor C detects abnormal pressure, the GNN model will evaluate the impact of the abnormality on the control logic of the PLC (node ​​A) and further analyze whether the fault will spread to other key devices, such as flow meters (node ​​D) and valves (node ​​E). Through a multi-round message passing mechanism, the model can dynamically predict the fault propagation path and give the fault propagation probability of each path.

[0110] As an optional example, based on the prediction results of the fault propagation path, the system can identify the most critical equipment in the fault propagation path. For example, if the GNN model analysis concludes that the pump station (node ​​B) is the core node of the fault propagation, and the fault is likely to spread further to other equipment (such as flow meter D and compressor F) through the PLC (node ​​A), the system will generate maintenance recommendations, prompting the operator to prioritize the inspection and maintenance of the pump station and PLC. At the same time, the system will also recommend status monitoring of affected equipment (such as pressure sensor C and valve E) to prevent further spread of the fault.

[0111] As an optional example, in the actual operation of the oil pipeline, the status of the equipment and the operating conditions of the system are changing dynamically. The GNN model continuously updates the prediction results of the fault propagation path by receiving the operating data of the PLC and other equipment in real time. If the system detects new abnormal signals during operation (such as further deterioration of the status of pump station B and compressor F), the GNN model will immediately update the fault propagation path and adjust the priority of the maintenance recommendations to ensure that the operator can take timely countermeasures. In addition, the model optimizes the fault prediction algorithm by continuously accumulating and learning historical data, thereby improving the accuracy and efficiency of future fault predictions.

[0112] Alternatively, assume that during a certain oil transportation process, the pressure sensor (node ​​C) located in the middle of the pipeline detects abnormal pressure fluctuations. By analyzing the anomaly, the GNN model finds that this may cause the pump station (node ​​B) to overload and further affect the downstream flow meter (node ​​D) and compressor (node ​​F). The system generates maintenance recommendations, prompting the operator to check the data transmission line of pressure sensor C and perform a load check on pump station B to ensure its normal operation. In addition, the system also recommends checking downstream equipment, especially flow meter D and compressor F, to prevent system failure due to abnormal pressure. According to the system's recommendations, the operator found that the signal line of pressure sensor C was mechanically damaged, resulting in unstable data transmission. By repairing the sensor's signal line, the operator successfully avoided overloading of the pump station and potential failure of downstream equipment, ensuring the safety and stability of the oil transportation process.

[0113] As an optional example, the GNN model mentioned above in this application can not only handle currently known fault modes, but also make adaptive adjustments when equipment or systems change. For example, when an oil pipeline is expanded or new control equipment is added, the system can retrain and optimize the fault prediction algorithm by updating the graph structure and GNN model to ensure that the model adapts to the new pipeline layout and control environment. With the support of this adaptive learning, the system always maintains the ability to efficiently predict faults in complex control environments.

[0114] The above embodiments of the present application can capture the complex interactions between devices. The use of graph neural networks (GNNs) has some unique advantages in dealing with fault propagation and other graph structure problems, which may be difficult to achieve or less effective than expected when using other models. The core of this advantage lies in the design concept and processing mechanism of GNN, which are naturally suitable for processing graph structure data and can effectively simulate complex interactions between nodes. The GNN model is used to process graph structure data in the SCADA system, fully explore the correlation between devices, and accurately predict the fault propagation path.

[0115] The above embodiments of the present application can improve the real-time and accuracy of fault prediction. Through dynamic analysis of real-time data of the equipment, the GNN model can predict possible system failures in a timely manner and make maintenance recommendations in advance to avoid the spread of faults. By performing multiple rounds of updates, information can be transmitted over longer distances in the graph, allowing the model to capture remote dependencies. This means that the system can predict potential fault propagation even between indirectly connected devices, which is often difficult to achieve when using traditional neural network models.

[0116] The above embodiments of the present application can enhance the adaptability of the system. GNN can adapt to dynamic changes in the graph structure, such as the addition or deletion of nodes and edges, which is particularly important for continuously changing systems such as SCADA systems. This flexibility may require complex modifications and retraining in other types of models.

[0117] According to an embodiment of the present invention, a fault prediction device embodiment is also provided. It should be noted that the fault prediction device can be used to execute the fault prediction method in the embodiment of the present invention, and the fault prediction method in the embodiment of the present invention can be executed in the fault prediction device.

[0118] Figure 5 is a schematic diagram of a fault prediction device according to an embodiment of the present invention. Figure 5As shown, the device may include: an acquisition module 52, which is used to acquire a set of equipment data of the oil and gas transmission system at the first time, wherein the oil and gas transmission system includes: a plurality of devices with a dependency relationship, the dependency relationship is at least used to represent the physical connection relationship or data transmission relationship between the devices, and the equipment data set includes: operation data of each device in the oil and gas transmission system; an analysis module 54, which is used to use a graph structure fault prediction model to analyze the operation data corresponding to the target device to obtain the device state of the target device at the first time, wherein the target device is any one of the plurality of devices, the graph structure fault prediction model is used to process graph structure data, the nodes in the graph structure data correspond one-to-one to the devices in the oil and gas transmission system, and the edges connecting the nodes in the graph structure data represent the dependency relationship in the oil and gas transmission system; a first determination module 56, which is used to determine the propagation state of the target device at the first time according to the device state of at least one adjacent device having a dependency relationship with the target device; a second determination module 58, which is used to determine the fault prediction result of the target device at the second time according to the device state and the propagation state, wherein the second time is adjacent to the first time and the second time is later than the first time.

[0119] It should be noted that the acquisition module 52 in this embodiment can be used to execute step S102 in the embodiment of the present application, the analysis module 54 in this embodiment can be used to execute step S104 in the embodiment of the present application, the first determination module 56 in this embodiment can be used to execute step S106 in the embodiment of the present application, and the second determination module 58 in this embodiment can be used to execute step S108 in the embodiment of the present application. The examples and application scenarios implemented by the above modules and corresponding steps are the same, but are not limited to the contents disclosed in the above embodiments.

[0120] In an embodiment of the present invention, a device data set of an oil and gas transmission system at a first time is obtained, wherein the oil and gas transmission system includes: a plurality of devices having a dependency relationship, the dependency relationship is used to at least represent a physical connection relationship or a data transmission relationship between the devices, and the device data set includes: operating data of each device in the oil and gas transmission system; using a graph structure fault prediction model to analyze the operating data corresponding to the target device, and obtain the device state of the target device at the first time, wherein the target device is any one of the plurality of devices, and the graph structure fault prediction model is used to process graph structure data, the nodes in the graph structure data correspond one-to-one to the devices in the oil and gas transmission system, and the edges connecting the nodes in the graph structure data Representing the dependency relationship in the oil and gas transmission system; determining the propagation state of the target device at the first time according to the device state of at least one adjacent device having a dependency relationship with the target device; determining the fault prediction result of the target device at the second time according to the device state and the propagation state, wherein the second time is adjacent to the first time and the second time is later than the first time, so that according to the operating data of each device in the oil and gas transmission system at the first time, combined with the dependency relationship between the devices, the fault prediction result of the target device at the second time can be more accurately predicted, thereby achieving the technical effect of improving the accuracy of fault prediction, and further solving the technical problem of insufficient accuracy of fault prediction in the oil and gas industry in the prior art.

[0121] As an optional embodiment, the second determination module includes: an analysis unit, configured to analyze the device state and the propagation state using a preset propagation model to obtain a fault prediction result of the target device at the second time, wherein the preset propagation model is expressed as: t means the first time, h j (t) represents the device state of device j at the first time, t+1 represents the second time, and h j (t+1) represents the fault prediction result of device j at the second time, represents the propagation state at the first time, N(j) is the set of neighboring devices of device j, i represents the neighboring devices of device j, C ij represents the connection strength between device i and device j, h i (t) represents the device state of the adjacent device i at the first time, W and B are the learning parameters of the preset propagation probability model, and σ represents the nonlinear activation function.

[0122] As an optional embodiment, the device also includes: a first determination submodule, which is used to determine at least one faulty device in the oil and gas transmission system at the second time based on the fault prediction result after determining the fault prediction result of the target device at the second time according to the device status and the propagation status; an analysis submodule, which is used to analyze the dependency relationship between multiple faulty devices based on graph structure data when there are multiple faulty devices, and obtain a core device, wherein the core device is a faulty device and there is a dependency relationship between the core device and multiple faulty devices; and a generation submodule, which is used to generate maintenance recommendations for the core device.

[0123] As an optional embodiment, the device also includes: a first acquisition submodule, used to acquire structural data of at least one oil and gas transmission system before acquiring a set of equipment data of the oil and gas transmission system at a first time, wherein the structural data is used to represent multiple devices in the oil and gas transmission system and the dependency relationship between the devices; a construction submodule, used to construct graph structure data corresponding to each oil and gas transmission system based on the structural data, wherein in the graph structure data, nodes are used to represent equipment in the oil and gas transmission system, edges connecting nodes are used to represent dependency relationships in the oil and gas transmission system, node attributes of nodes represent equipment status, and edge attributes of edges represent connection strength.

[0124] As an optional embodiment, the device also includes: a second acquisition submodule, which is used to acquire a training data set corresponding to each oil and gas transmission system after constructing the graph structure data corresponding to each oil and gas transmission system according to the structural data, wherein the training data set includes: multiple correct operation samples, and multiple fault operation samples, the correct operation samples are determined according to the operation data of the equipment in a normal state, and the fault operation samples are determined according to the operation data of the equipment in a fault state; a training submodule, which is used to input the correct operation samples and fault operation samples in the training data set into a preset graph structure fault prediction template corresponding to the oil and gas transmission system for training to obtain a graph structure fault prediction model, wherein the preset graph structure fault prediction template is established according to the graph structure data corresponding to the oil and gas transmission system.

[0125] As an optional embodiment, the second acquisition submodule includes: an acquisition subunit, used to acquire an operation data set of each oil and gas transmission system within a preset time period, wherein each operation data set includes: normal operation data of equipment in a normal state, and fault operation data of equipment in a fault state in the same oil and gas transmission system; a first processing subunit, used to perform feature engineering processing on multiple normal operation data within the same preset time period, and obtain multiple correct operation features indicating that the equipment state is in a normal state, wherein the correct operation features are used to determine correct operation samples; a second processing subunit, used to perform feature engineering processing on multiple fault operation data within the same preset time period, and obtain multiple fault operation features indicating that the equipment state is in a fault state, wherein the fault operation features are used to determine fault operation samples.

[0126] As an optional embodiment, the device also includes: a second determination submodule, which is used to obtain a fault handling opinion for the oil and gas transmission system based on the fault prediction result after determining the fault prediction result of the target device at the second time according to the device status and the propagation status, wherein the fault handling opinion is at least used to represent the actual device status of the equipment in the oil and gas transmission system at the second time; an updating submodule, which is used to update the graph structure fault prediction model based on the actual device status of the target device in the oil and gas transmission system at the second time.

[0127] An embodiment of the present invention may provide an electronic device, which may be a computer terminal, and the computer terminal may be any computer terminal device in a computer terminal group. Optionally, in this embodiment, the computer terminal may also be replaced by a terminal device such as a mobile terminal.

[0128] Optionally, in this embodiment, the computer terminal may be located in at least one network device among a plurality of network devices of the computer network.

[0129] In this embodiment, the above-mentioned computer terminal can execute the program code of the following steps in the fault prediction method: obtaining a set of equipment data of the oil and gas transmission system at the first time, wherein the oil and gas transmission system includes: multiple devices with a dependency relationship, the dependency relationship is at least used to represent the physical connection relationship or data transmission relationship between the devices, and the equipment data set includes: operation data of each device in the oil and gas transmission system; using a graph structure fault prediction model to analyze the operation data corresponding to the target device to obtain the device state of the target device at the first time, wherein the target device is any one of the multiple devices, and the graph structure fault prediction model is used to process graph structure data, the nodes in the graph structure data correspond one-to-one to the devices in the oil and gas transmission system, and the edges connecting the nodes in the graph structure data represent the dependency relationship in the oil and gas transmission system; determining the propagation state of the target device at the first time according to the device state of at least one adjacent device having a dependency relationship with the target device; determining the fault prediction result of the target device at the second time according to the device state and the propagation state, wherein the second time is adjacent to the first time and the second time is later than the first time.

[0130] Figure 6 is a structural block diagram of a computer terminal according to an embodiment of the present invention. Figure 6 As shown, the computer terminal 60 may include: one or more (only one is shown in the figure) processors 62 and a memory 64 .

[0131] Among them, the memory can be used to store software programs and modules, such as program instructions / modules corresponding to the fault prediction method and device in the embodiment of the present invention. The processor executes various functional applications and data processing by running the software programs and modules stored in the memory, that is, realizing the above-mentioned fault prediction method. The memory may include a high-speed random access memory, and may also include a non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some instances, the memory may further include a memory remotely arranged relative to the processor, and these remote memories may be connected to the terminal 60 via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0132] The processor can call the information and application stored in the memory through the transmission device to perform the following steps: obtain a set of equipment data of the oil and gas transmission system at the first time, wherein the oil and gas transmission system includes: multiple devices with a dependency relationship, the dependency relationship is at least used to represent the physical connection relationship or data transmission relationship between the devices, and the equipment data set includes: operation data of each device in the oil and gas transmission system; use a graph structure fault prediction model to analyze the operation data corresponding to the target device to obtain the device state of the target device at the first time, wherein the target device is any one of the multiple devices, and the graph structure fault prediction model is used to process graph structure data, the nodes in the graph structure data correspond one-to-one to the devices in the oil and gas transmission system, and the edges connecting the nodes in the graph structure data represent the dependency relationship in the oil and gas transmission system; determine the propagation state of the target device at the first time based on the device state of at least one adjacent device with a dependency relationship with the target device; determine the fault prediction result of the target device at the second time based on the device state and the propagation state, wherein the second time is adjacent to the first time and the second time is later than the first time.

[0133] Optionally, the processor may further execute a program code of the following steps: using a preset propagation model to analyze the device state and the propagation state to obtain a fault prediction result of the target device at the second time, wherein the preset propagation model is expressed as: Among them, t represents the first time, h j (t) represents the device state of device j at the first time, t+1 represents the second time, and h j (t+1) represents the fault prediction result of device j at the second time, represents the propagation state at the first time, N(j) is the set of neighboring devices of device j, i represents the neighboring devices of device j, C ij represents the connection strength between device i and device j, h i (t) represents the device state of the adjacent device i at the first time, W and B are the learning parameters of the preset propagation probability model, and σ represents the nonlinear activation function.

[0134] Optionally, the processor may also execute the program code of the following steps: determining at least one faulty device in the oil and gas transmission system at the second time based on the fault prediction result; in the case where there are multiple faulty devices, analyzing the dependency relationship between the multiple faulty devices based on the graph structure data to obtain a core device, wherein the core device is a faulty device and there is a dependency relationship between the core device and multiple faulty devices; and generating maintenance recommendations for the core device.

[0135] Optionally, the processor may also execute the program code of the following steps: obtaining structural data of at least one oil and gas transmission system, wherein the structural data is used to represent multiple devices in the oil and gas transmission system and the dependency relationships between the devices; constructing graph structure data corresponding to each oil and gas transmission system based on the structural data, wherein, in the graph structure data, nodes are used to represent devices in the oil and gas transmission system, edges connecting nodes are used to represent the dependency relationships in the oil and gas transmission system, the node attributes of the nodes represent the device status, and the edge attributes of the edges represent the connection strength.

[0136] Optionally, the processor may also execute the following program code: obtaining a training data set corresponding to each oil and gas transmission system, wherein the training data set includes: a plurality of correct operation samples, and a plurality of faulty operation samples, the correct operation samples are determined based on the operation data of a device in a normal state, and the faulty operation samples are determined based on the operation data of a device in a faulty state; inputting the correct operation samples and the faulty operation samples in the training data set into a preset graph structure fault prediction template corresponding to the oil and gas transmission system for training to obtain a graph structure fault prediction model, wherein the preset graph structure fault prediction template is established based on the graph structure data corresponding to the oil and gas transmission system.

[0137] Optionally, the processor may also execute the program code of the following steps: obtaining an operation data set of each oil and gas transmission system within a preset time period, wherein each operation data set includes: normal operation data of equipment in a normal state, and fault operation data of equipment in a fault state in the same oil and gas transmission system; performing feature engineering processing on multiple normal operation data within the same preset time period to obtain multiple correct operation features indicating that the equipment state is in a normal state, wherein the correct operation features are used to determine correct operation samples; performing feature engineering processing on multiple fault operation data within the same preset time period to obtain multiple fault operation features indicating that the equipment state is in a fault state, wherein the fault operation features are used to determine fault operation samples.

[0138] Optionally, the processor may also execute program codes for the following steps: obtaining fault handling opinions for the oil and gas transmission system based on the fault prediction results, wherein the fault handling opinions are at least used to represent the actual equipment status of the equipment in the oil and gas transmission system at the second time; and updating the graph structure fault prediction model based on the actual equipment status of the target equipment in the oil and gas transmission system at the second time.

[0139] In an embodiment of the present invention, a device data set of an oil and gas transmission system at a first time is obtained, wherein the oil and gas transmission system includes: a plurality of devices having a dependency relationship, the dependency relationship is used to at least represent a physical connection relationship or a data transmission relationship between the devices, and the device data set includes: operating data of each device in the oil and gas transmission system; using a graph structure fault prediction model to analyze the operating data corresponding to the target device, and obtain the device state of the target device at the first time, wherein the target device is any one of the plurality of devices, and the graph structure fault prediction model is used to process graph structure data, the nodes in the graph structure data correspond one-to-one to the devices in the oil and gas transmission system, and the edges connecting the nodes in the graph structure data Representing the dependency relationship in the oil and gas transmission system; determining the propagation state of the target device at the first time according to the device state of at least one adjacent device having a dependency relationship with the target device; determining the fault prediction result of the target device at the second time according to the device state and the propagation state, wherein the second time is adjacent to the first time and the second time is later than the first time, so that according to the operating data of each device in the oil and gas transmission system at the first time, combined with the dependency relationship between the devices, the fault prediction result of the target device at the second time can be more accurately predicted, thereby achieving the technical effect of improving the accuracy of fault prediction, and further solving the technical problem of insufficient accuracy of fault prediction in the oil and gas industry in the prior art.

[0140] It can be understood by those skilled in the art that Figure 6 The structure shown is for illustration only, and the computer terminal may also be a smart phone (such as an Android phone, an iOS phone, etc.), a tablet computer, a handheld computer, a mobile Internet device (Mobile Internet Devices, MID), a PAD, or other terminal devices. Figure 6 The structure of the electronic device is not limited. For example, the computer terminal 60 may also include Figure 6 More or fewer components (such as network interfaces, display devices, etc.) shown in, or having Figure 6 Different configurations shown.

[0141] A person of ordinary skill in the art can understand that all or part of the steps in the various methods of the above embodiments can be completed by instructing the hardware related to the terminal device through a computer program. The computer program can be stored in a non-volatile medium. The non-volatile storage medium may include: a flash drive, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, etc.

[0142] The embodiment of the present invention further provides a non-volatile storage medium. Optionally, in this embodiment, the non-volatile storage medium can be used to store the program code executed by the fault prediction method provided by the above embodiment.

[0143] Optionally, in this embodiment, the non-volatile storage medium may be located in any computer terminal in a computer terminal group in a computer network, or in any mobile terminal in a mobile terminal group.

[0144] Optionally, in this embodiment, the non-volatile storage medium is configured to store program code for executing the following steps: obtaining a set of equipment data of the oil and gas transmission system at the first time, wherein the oil and gas transmission system includes: multiple devices with a dependency relationship, the dependency relationship is at least used to represent the physical connection relationship or data transmission relationship between the devices, and the equipment data set includes: operating data of each device in the oil and gas transmission system; using a graph structure fault prediction model to analyze the operating data corresponding to the target device to obtain the device state of the target device at the first time, wherein the target device is any one of the multiple devices, and the graph structure fault prediction model is used to process graph structure data, the nodes in the graph structure data correspond one-to-one to the devices in the oil and gas transmission system, and the edges connecting the nodes in the graph structure data represent the dependency relationship in the oil and gas transmission system; determining the propagation state of the target device at the first time based on the device state of at least one adjacent device having a dependency relationship with the target device; determining the fault prediction result of the target device at the second time based on the device state and the propagation state, wherein the second time is adjacent to the first time and the second time is later than the first time.

[0145] Optionally, in this embodiment, the non-volatile storage medium is configured to store a program code for executing the following steps: using a preset propagation model to analyze the device state and the propagation state to obtain a fault prediction result of the target device at the second time, wherein the preset propagation model is expressed as: Among them, t represents the first time, h j (t) represents the device state of device j at the first time, t+1 represents the second time, and h j (t+1) represents the fault prediction result of device j at the second time, represents the propagation state at the first time, N(j) is the set of neighboring devices of device j, i represents the neighboring devices of device j, C ij represents the connection strength between device i and device j, h i (t) represents the device state of the adjacent device i at the first time, W and B are the learning parameters of the preset propagation probability model, and σ represents the nonlinear activation function.

[0146] Optionally, in this embodiment, the non-volatile storage medium is configured to store program codes for executing the following steps: determining at least one faulty device in the oil and gas transmission system at the second time based on the fault prediction result; in the case where there are multiple faulty devices, analyzing the dependency relationship between the multiple faulty devices based on the graph structure data to obtain a core device, wherein the core device is a faulty device and the core device has a dependency relationship with multiple faulty devices; and generating maintenance recommendations for the core device.

[0147] Optionally, in this embodiment, the non-volatile storage medium is configured to store program code for executing the following steps: obtaining structural data of at least one oil and gas transmission system, wherein the structural data is used to represent multiple devices in the oil and gas transmission system and the dependencies between the devices; constructing graph structure data corresponding to each oil and gas transmission system based on the structural data, wherein, in the graph structure data, nodes are used to represent devices in the oil and gas transmission system, edges connecting nodes are used to represent dependencies in the oil and gas transmission system, node attributes of the nodes represent device status, and edge attributes of the edges represent connection strength.

[0148] Optionally, in this embodiment, the non-volatile storage medium is configured to store program code for executing the following steps: obtaining a training data set corresponding to each oil and gas transmission system, wherein the training data set includes: multiple correct operation samples, and multiple faulty operation samples, the correct operation samples are determined based on the operation data of the device in a normal state, and the faulty operation samples are determined based on the operation data of the device in a faulty state; the correct operation samples and the faulty operation samples in the training data set are input into a preset graph structure fault prediction template corresponding to the oil and gas transmission system for training to obtain a graph structure fault prediction model, wherein the preset graph structure fault prediction template is established based on the graph structure data corresponding to the oil and gas transmission system.

[0149] Optionally, in this embodiment, the non-volatile storage medium is configured to store program code for executing the following steps: obtaining an operation data set for each oil and gas transmission system within a preset time period, wherein each operation data set includes: normal operation data of equipment in a normal state, and fault operation data of equipment in a fault state in the same oil and gas transmission system; performing feature engineering processing on multiple normal operation data within the same preset time period to obtain multiple correct operation features indicating that the equipment state is in a normal state, wherein the correct operation features are used to determine correct operation samples; performing feature engineering processing on multiple fault operation data within the same preset time period to obtain multiple fault operation features indicating that the equipment state is in a fault state, wherein the fault operation features are used to determine fault operation samples.

[0150] Optionally, in this embodiment, the non-volatile storage medium is configured to store program code for executing the following steps: obtaining fault handling opinions for the oil and gas transmission system based on the fault prediction results, wherein the fault handling opinions are at least used to represent the actual equipment status of the equipment in the oil and gas transmission system at the second time; and updating the graph structure fault prediction model based on the actual equipment status of the target equipment in the oil and gas transmission system at the second time.

[0151] The embodiment of the present invention further provides a computer program product, including a computer program. Optionally, in this embodiment, when the computer program is executed by a processor, the steps of the fault prediction method provided in the above embodiment are implemented.

[0152] The serial numbers of the above embodiments of the present invention are only for description and do not represent the advantages or disadvantages of the embodiments.

[0153] In the above embodiments of the present invention, the description of each embodiment has its own emphasis. For parts that are not described in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0154] In the several embodiments provided in this application, it should be understood that the disclosed technical content can be implemented in other ways. Among them, the device embodiments described above are only schematic. For example, the division of the units can be a logical function division. There may be other division methods in actual implementation. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of units or modules, which can be electrical or other forms.

[0155] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple units. Some or all of the units may be selected according to actual needs to achieve the purpose of the present embodiment.

[0156] In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0157] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a non-volatile storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product. The computer software product is stored in a non-volatile storage medium, including several instructions for a computer device (which can be a personal computer, server or network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned non-volatile storage medium includes: U disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), mobile hard disk, magnetic disk or optical disk and other media that can store program codes.

[0158] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principle of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. A fault prediction method, characterized in that: include: Acquire a device data set of an oil and gas transmission system at a first time, wherein the oil and gas transmission system includes: a plurality of devices having a dependency relationship, wherein the dependency relationship is at least used to represent a physical connection relationship or a data transmission relationship between the devices, and the device data set includes: operation data of each device in the oil and gas transmission system; Using a graph structure fault prediction model to analyze the operation data corresponding to the target device, to obtain the device state of the target device at the first time, wherein the target device is any one of the multiple devices, the graph structure fault prediction model is used to process graph structure data, the nodes in the graph structure data correspond one-to-one to the devices in the oil and gas transmission system, and the edges connecting the nodes in the graph structure data represent the dependency relationship in the oil and gas transmission system; determining a propagation state of the target device at the first time according to the device state of at least one neighboring device having the dependency relationship with the target device; A fault prediction result of the target device at a second time is determined according to the device state and the propagation state, wherein the second time is adjacent to the first time and the second time is later than the first time.

2. The method according to claim 1, characterized in that Determining, according to the device state and the propagation state, a fault prediction result of the target device at a second time includes: The device state and the propagation state are analyzed using a preset propagation model to obtain a fault prediction result of the target device at the second time, wherein the preset propagation model is expressed as: Wherein, t represents the first time, h j (t) represents the device state of the device j at the first time, t+1 represents the second time, and h j (t+1) represents the fault prediction result of the device j at the second time, represents the propagation state at the first time, N(j) is the set of neighboring devices of the device j, i represents the neighboring device of the device j, C ij represents the connection strength between the device i and the device j, h i (t) represents the device state of the adjacent device i at the first time, W and B are learning parameters of the preset propagation probability model, and σ represents a nonlinear activation function.

3. The method according to claim 1, characterized in that After determining the fault prediction result of the target device at the second time according to the device state and the propagation state, the method further includes: determining, according to the fault prediction result, at least one faulty device of the oil and gas transmission system at the second time; In the case where there are multiple faulty devices, analyzing the dependency relationship between the multiple faulty devices based on the graph structure data to obtain a core device, wherein the core device is the faulty device, and the core device has the dependency relationship with the multiple faulty devices; A maintenance recommendation for the core equipment is generated.

4. The method according to claim 1, characterized in that: Before obtaining the equipment data set of the oil and gas transmission system at the first time, the method further includes: Acquire structural data of at least one of the oil and gas transmission systems, wherein the structural data is used to represent a plurality of the devices in the oil and gas transmission system and dependencies between the devices; According to the structural data, the graph structure data corresponding to each of the oil and gas transmission systems is constructed, wherein, in the graph structure data, the nodes are used to represent the equipment in the oil and gas transmission system, the edges connecting the nodes are used to represent the dependency relationships in the oil and gas transmission system, the node attributes of the nodes represent the equipment status, and the edge attributes of the edges represent the connection strength.

5. The method according to claim 4, characterized in that After constructing the graph structure data corresponding to each of the oil and gas transmission systems according to the structure data, the method further includes: Acquire a training data set corresponding to each of the oil and gas transmission systems, wherein the training data set includes: a plurality of correct operation samples, and a plurality of fault operation samples, the correct operation samples are determined according to the operation data of the device when the device state is a normal state, and the fault operation samples are determined according to the operation data of the device when the device state is the fault state; The correct operation samples and the faulty operation samples in the training data set are input into the preset graph structure fault prediction template corresponding to the oil and gas transmission system for training to obtain the graph structure fault prediction model, wherein the preset graph structure fault prediction template is established based on the graph structure data corresponding to the oil and gas transmission system.

6. The method according to claim 5, characterized in that Acquiring a training data set corresponding to each of the oil and gas transmission systems includes: Acquire an operation data set of each of the oil and gas transmission systems within a preset time period, wherein each of the operation data sets includes: normal operation data of the equipment whose equipment status is the normal state, and fault operation data of the equipment whose equipment status is the fault state in the same oil and gas transmission system; Performing feature engineering processing on the plurality of normal operation data within the same preset time period to obtain a plurality of correct operation features indicating that the device state is the normal state, wherein the correct operation features are used to determine the correct operation samples; Feature engineering processing is performed on the multiple fault operation data within the same preset time period to obtain multiple fault operation features indicating that the device state is the fault state, wherein the fault operation features are used to determine the fault operation samples.

7. The method according to claim 1, characterized in that According to the device state and the propagation state, After determining the fault prediction result of the target device at the second time, the method further includes: Obtaining a fault handling opinion for the oil and gas transmission system based on the fault prediction result, wherein the fault handling opinion is at least used to indicate the actual device state of the device in the oil and gas transmission system at the second time; Based on the actual device status of each device in the oil and gas transmission system at the second time, the graph structure fault prediction model is updated.

8. A fault prediction device, characterized in that: include: an acquisition module, used for acquiring a device data set of an oil and gas transmission system at a first time, wherein the oil and gas transmission system comprises: a plurality of devices having a dependency relationship, wherein the dependency relationship is at least used to represent a physical connection relationship or a data transmission relationship between the devices, and the device data set comprises: operation data of each device in the oil and gas transmission system; an analysis module, configured to analyze operation data corresponding to a target device using a graph structure fault prediction model to obtain a device state of the target device at the first time, wherein the target device is any one of the multiple devices, the graph structure fault prediction model is configured to process graph structure data, the nodes in the graph structure data correspond one-to-one to the devices in the oil and gas transmission system, and the edges connecting the nodes in the graph structure data represent the dependency relationships in the oil and gas transmission system; A first determining module, configured to determine a propagation state of the target device at the first time according to the device state of at least one neighboring device having the dependency relationship with the target device; The second determination module is used to determine the fault prediction result of the target device at a second time according to the device state and the propagation state, wherein the second time is adjacent to the first time and the second time is later than the first time.

9. An electronic device, comprising a memory and a processor, characterized in that: The memory stores a computer program, and the processor is configured to execute the fault prediction method according to any one of claims 1 to 7 through the computer program.

10. A computer program product comprising computer instructions, characterized in that: When the computer instructions are executed by a processor, the steps of the fault prediction method according to any one of claims 1 to 7 are implemented.

Images