A drilling production safety control method based on a knowledge graph

Abstract

The present invention discloses a method for controlling and managing the safety of drilling production based on a knowledge graph. The method includes: constructing a production safety knowledge graph according to the content form of dangerous behaviors or violation behaviors in the drilling production operation specifications, which includes behavior condition nodes, dangerous event nodes, combined conditional relationships and weights. The weights can be defined manually or calculated using algorithms. For the production area, real-time images are collected, processed into images or video frames, and then the YOLOv5 model is used to identify the behaviors therein, obtaining one or more behaviors. Then, according to the relationship between the behavior conditions and dangerous events in the knowledge graph, the possible dangers and risk probabilities are obtained from the associated weights, so as to give an alarm or take other measures. The present invention combines the knowledge graph to control and manage the safety of drilling production, solves the problems of slow response and untimely response of traditional methods, can quickly judge what dangers may occur when risk behaviors occur, and enables the staff to respond in time.

Description

Technical Field

[0001] The present invention belongs to the field of production safety, and particularly relates to a drilling production safety control method based on a knowledge graph. Background Art

[0002] In drilling production operations, traditional safety monitoring methods often rely on manual supervision or automated systems based on fixed rules. However, with the increasing complexity and risk factors of drilling operations, traditional safety monitoring methods gradually show problems such as slow response speed and insufficient accuracy, and are unable to make timely and effective responses to the rapidly changing operating environment. This lag increases potential safety hazards and leads to frequent occurrence of dangerous events, which affects production efficiency and safety.

[0003] In recent years, the rapid development of computer vision and artificial intelligence technologies has provided new solutions for drilling production safety monitoring. By using deep learning technologies, computer vision technologies such as object detection and behavior recognition can analyze images or video streams of the operation site in real time, so as to automatically detect and identify violations or potential hazards in the operation. However, such systems usually rely only on visual information for judgment, lack comprehensive consideration of complex processes and various operating conditions, and are difficult to achieve comprehensive and accurate safety assessment and risk warning.

[0004] To solve this problem, as a tool that can systematically organize and express domain knowledge, the knowledge graph shows broad application prospects in drilling production safety control. By structuring dangerous behaviors and conditions in drilling operations and assigning weight coefficients, the knowledge graph can make more accurate judgments and predictions on complex behaviors and potential hazards in operations. Combined with computer vision technology, the control based on the knowledge graph can realize real-time monitoring and intelligent analysis of the operation site, effectively improving the efficiency and accuracy of safety management. Summary of the Invention

[0005] The purpose of the present invention is to improve the safety management efficiency in drilling production operations, and provide a drilling production safety control method based on a knowledge graph. By combining computer vision and knowledge graph technologies, it automatically identifies behaviors at the operation site, analyzes their potential risks, and thus realizes rapid and accurate danger prediction and response measures.

[0006] To solve the problems of slow response and inaccurate judgment of traditional safety monitoring methods, the present invention constructs an association model of dangerous behaviors and events through a knowledge graph, combines with the YOLOv5 object recognition algorithm, conducts real-time analysis and recognition of videos or images in the operation area, automatically judges the risks of operation behaviors, and gives corresponding risk probabilities, so as to realize intelligent alarm and dynamic safety control.

[0007] To achieve the above object, the present invention adopts the following technical solutions:

[0008] (1) According to the form of dangerous behaviors and violations in the drilling production operation specifications, construct a production safety knowledge graph. The nodes include behavior conditions and dangerous events, and the association relationship is "single cause" where a dangerous event can be caused by one behavior condition, or "joint cause" where a dangerous event is jointly triggered by multiple behavior conditions. The degree of association between the behavior and the event is characterized by a weight coefficient;

[0009] (2) Use a camera device to collect real-time images of the production area to obtain video or image data;

[0010] (3) Use the YOLOv5 object recognition algorithm to identify the operation behaviors in the images or video frames;

[0011] (4) According to the association between the behavior and the dangerous event in the knowledge graph and its weight, infer possible dangers and their risk probabilities;

[0012] (5) According to the calculated dangers and risk probabilities, determine whether to issue an alarm or take corresponding countermeasures.

[0013] A further improvement of the present invention is that the specific implementation steps of weight assignment in step (1) are as follows:

[0014] (101) For events with fewer samples, or higher importance and more serious consequences that require human intervention, assign weights manually according to the behaviors in the drilling production operation specifications. Connecting edges with higher degrees of association are given higher weights;

[0015] (102) For other events, according to the training set data captured at the drilling site and the actual occurrence frequency of the corresponding dangerous events, calculate their weights using a formula. The weight of the "single cause" relationship edge is the ratio of the number of times the dangerous event occurs when the behavior condition occurs to the number of times the behavior condition occurs; the weight of the "joint cause" relationship edge is N j / N a where N j is the total number of times the corresponding behavior condition occurs when the dangerous event occurs, and N a is the total number of times all associated behavior conditions occur when the dangerous event occurs;

[0016] A further improvement of the present invention is that the specific implementation steps of step (4) are as follows:

[0017] (401) For events that can be caused by a single behavior condition or the existence of one of multiple behaviors, the risk probability is:

[0018]

[0019] Among them, N represents the total number of independent behavior conditions identified as existing, and p n represents the degree of association between the corresponding behavior and the event, that is, the weight of the connection edge between the two nodes;

[0020] (402) For an event jointly caused by multiple behavior conditions, its risk probability is:

[0021]

[0022] Among them, N represents the total number of associated behavior conditions, and w n represents the degree of association between the corresponding behavior and the event, that is, the weight of the connection edge between the two nodes;

[0023] A further improvement of the present invention lies in that the specific implementation steps of step (5) are:

[0024] (501) The danger and risk probability can be controlled by setting a threshold θ to determine whether to give an alarm or take countermeasures:

[0025]

[0026] When R is 1, when the danger probability exceeds the threshold, action is taken at this time, otherwise no action is taken.

[0027] The beneficial effects of the drilling production safety control method based on the knowledge graph proposed by the present invention: It solves the problems of slow response and timeliness of the traditional method, can quickly judge what dangers may occur when a risk behavior occurs, enabling the staff to respond in a timely manner. It can make more accurate judgments and predictions on complex behaviors and potential dangers during operations. Combining computer vision and knowledge graph technologies, it can achieve real-time monitoring and intelligent analysis of the operation site, effectively improving the efficiency and accuracy of safety management. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following will briefly introduce the drawings required for the description of the embodiments. Obviously, the following drawings are only some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0029] Figure 1 It is the framework diagram of the drilling production safety control based on the knowledge graph described in the embodiments of the present invention.

[0030] Figure 2 It is the flow chart of the drilling production safety control method based on the knowledge graph described in the embodiments of the present invention. DETAILED DESCRIPTION OF THE INVENTION

[0031] The embodiments of the present invention will be described in detail below. Examples of the embodiments are shown in the accompanying drawings, where the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the drawings are exemplary and are only used to explain the present invention, and should not be construed as a limitation of the present invention.

[0032] Those skilled in the art of the present technology can understand that, unless otherwise defined, all terms (including technical terms and scientific terms) used herein have the same meaning as the general understanding of those of ordinary skill in the art to which the present invention belongs.

[0033] It should also be understood that terms such as those defined in a general dictionary should be understood to have a meaning consistent with their meaning in the context of the prior art, and will not be interpreted in an idealized or overly formal sense unless defined as here.

[0034] Those skilled in the art of the present technology can understand that, unless specifically stated otherwise, the singular forms "a", "an", "the", and "said" used herein may also include the plural forms. It should be further understood that the term "comprising" used in the description of the present invention means the presence of the described features, integers, steps, operations, elements, and / or components, but does not exclude the presence or addition of one or more other features, integers, steps, operations, elements, and / or groups thereof.

[0035] In the description of this specification, the description with reference to terms such as "one embodiment", "some embodiments", "example", "specific example", or "some examples" means that the specific features, structures, materials, or characteristics described in connection with the embodiment or example are included in at least one embodiment or example of the present invention. Moreover, the specific features, structures, materials, or characteristics described may be combined in a suitable manner in any one or more embodiments or examples. In addition, without conflict, those skilled in the art may combine and combine the different embodiments or examples described in this specification and the features of the different embodiments or examples.

[0036] To facilitate the understanding of the present invention, the present invention will be further explained below with reference to the accompanying drawings in specific embodiments, and the specific embodiments do not constitute a limitation on the embodiments of the present invention.

[0037] Those skilled in the art should understand that the drawings are only schematic diagrams of the embodiments, and the components in the drawings are not necessarily essential for implementing the present invention.

[0038] As Figure 1 shown, the embodiment of the present invention provides a drilling production safety control and management framework diagram based on a knowledge graph, including:

[0039] The video stream acquisition module captures images or videos of the scene at the operation site to obtain on-site scene data.

[0040] In this embodiment, the video stream acquisition module continuously captures the scene of the drilling operation site. For the on-site data used to initialize the system of this embodiment, the behaviors that appear therein are manually marked, including the name of the behavior, the position and scope where the behavior appears, whether the behavior has caused a dangerous event to occur, and these data are stored. For the captured images during operation, the directly captured video stream is stored for backup and transmitted to the video image processing module for processing.

[0041] The knowledge graph construction module constructs a production safety knowledge graph according to the list of dangerous behaviors and violation behaviors in the drilling production operation specifications.

[0042] In this embodiment, the knowledge graph construction module includes independent association construction and joint association construction.

[0043] The independent association construction is to construct an "independently cause" relationship and assign a weight to the edge. The weight is the ratio of the number of times the danger occurs when the behavior condition occurs to the number of times the behavior condition occurs;

[0044] The joint association construction is to construct a "jointly cause" relationship and assign a weight to the edge. The weight is the total number of times the corresponding behavior condition occurs when the danger occurs to the total number of times all associated behavior conditions occur when the danger occurs.

[0045] Specifically, the knowledge graph construction module first reads the list of violation behaviors, lists the dangerous events and one or more action behaviors that cause the events to occur in the list, and creates action behavior nodes and dangerous event nodes to form an initial knowledge graph, which is stored using a knowledge graph storage structure, such as a Neo4j graph database. For a dangerous event with only one pre-action behavior, the two are connected with an "independently cause" relationship edge, and according to the on-site data of the drilling operation, the weight of this edge is assigned as: the ratio of the number of times the danger occurs when the behavior condition occurs to the number of times the behavior condition occurs; for a dangerous event with multiple pre-action behaviors, the two are connected with a "jointly cause" relationship edge, and according to the on-site data of the drilling operation, the weight of this edge is assigned as N j / N a where N j is the total number of times the corresponding behavior condition occurs when the danger occurs, and N a is the total number of times all associated behavior conditions occur when the danger occurs.

[0046] The video image processing module performs frame-by-frame processing on the video data of the operation site to obtain image data, identifies the behavior targets in the images, and outputs the action behaviors existing in the frame;

[0047] In this embodiment, the video image processing module frames the incoming video to obtain consecutive images, normalizes the image resolution to 1280*1280, and uses a pre-trained YOLOv5 target recognition algorithm model to perform target recognition on the behaviors in the images. The model is trained until the recognition accuracy reaches over 90%. After recognizing the behavioral actions in the consecutive frames, the module outputs them. The behavioral actions that occur within 15 consecutive frames are regarded as occurring simultaneously, and the module outputs the behavioral actions that have appeared in the past 15 frames in real time.

[0048] The logical deduction module calculates various possible dangerous events and their occurrence probabilities based on the constructed knowledge graph and the recognized behavioral actions.

[0049] In this embodiment, the logical deduction module includes independent behavior deduction and joint behavior deduction.

[0050] The independent behavior deduction is used to calculate the occurrence probability of a dangerous event when a behavior that can independently cause a dangerous event appears.

[0051] The joint behavior deduction is used to calculate the occurrence probability of a dangerous event when some or all of the behaviors that can jointly cause a dangerous event appear.

[0052] Specifically, the logical deduction module searches in the production safety knowledge graph based on the recognized behavioral actions, finds all the dangerous events connected to these behavioral actions, and then calculates the probabilities for these dangerous events. If the connection mode between a dangerous event and the existing behavioral actions is "independently cause", the occurrence probability P of this event is calculated as , where N is all the currently existing behavioral actions connected to this event in an "independently cause" relationship, and p n is the weight of the corresponding connection edge; if the connection mode between a dangerous event and the existing behavioral actions is "jointly cause", the occurrence probability P of this event is calculated as , where N is all the currently existing behavioral actions connected to this event in a "jointly cause" relationship, and w n is the weight of the corresponding connection edge. After calculating all the above dangerous events, a list of dangerous events and their occurrence rates is output.

[0053] The danger handling module gives an alarm or takes corresponding measures for behaviors with a danger probability exceeding the threshold.

[0054] Specifically, a probability threshold for each dangerous event to be responded to is manually set. After the logic deduction module calculates the dangerous events and probabilities at the current work site, it determines whether the probability of each event exceeds the specified threshold for it. If it exceeds the threshold, preset processing is performed; otherwise, no response is made. For dangerous events that require an alarm, an alarm is issued after their probability exceeds the threshold. For example, standing under the working area can cause the danger of falling objects from a height. When the danger probability exceeds 12%, an alarm needs to be issued to remind the on-site workers to leave the dangerous area in time. For dangerous events that require other corresponding measures, the on-site management personnel are notified. For example, the behavior of rising with the hook can lead to a dangerous event of being trapped at a high place. When the danger probability is higher than 3%, the management personnel need to be informed to conduct an on-site inspection and carry out rescue if necessary.

[0055] The overall process of the drilling production safety control method based on the knowledge graph provided by the present invention is as Figure 2 shown, and the specific operation steps of this process are as follows:

[0056] (1) According to the form of dangerous behaviors and violation behaviors in the drilling production operation specifications, construct a production safety knowledge graph, which includes behavior condition nodes and dangerous event nodes, and combines condition relationships and weight coefficients;

[0057] (2) Collect image or video data of the production area, and after processing, obtain the image of the operation area and construct it into a data set;

[0058] (3) Label the collected data and train the YOLOv5 model. After completion, it can identify the behavior actions in the visual objects.

[0059] (4) Obtain the video stream of the work site, use the YOLOv5 target recognition algorithm to identify the targets in the image or video frame, and determine one or more existing behaviors;

[0060] (5) According to the one or more behaviors obtained, infer through the dangers that the actions in the knowledge graph may lead to and the possible behavior sources of the corresponding dangers, and give the possible dangers and risk probabilities;

[0061] (6) According to the given danger and risk probability information, for behaviors or combinations of behaviors with probabilities higher than the set threshold, give an alarm or take corresponding measures;

[0062] Those skilled in the art should understand that the embodiments of the present invention can be provided as a method, a system, or a computer program product. Therefore, the present invention can take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present invention can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that contain computer-usable program code.

[0063] The present invention is described with reference to the flowcharts and / or block diagrams of methods, apparatuses (systems), and computer program products according to embodiments of the present invention. It should be understood that each flow and / or block in the flowchart and / or block diagram, and the combination of flows and / or blocks in the flowchart and / or block diagram, can be realized by computer program instructions. These computer program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing devices to generate a machine, such that the instructions executed by the processor of the computer or other programmable data processing devices generate a device for realizing the functions specified in Figure 1 one flow or multiple flows and / or blocks Figure 1 one block or multiple blocks.

[0064] These computer program instructions can also be stored in a computer-readable memory that can direct a computer or other programmable data processing device to work in a specific manner, such that the instructions stored in the computer-readable memory generate a manufactured article including an instruction device that realizes the functions specified in Figure 1 one flow or multiple flows and / or blocks Figure 1 one block or multiple blocks.

[0065] These computer program instructions can also be loaded onto a computer or other programmable data processing device, and a series of operation steps are executed on the computer or other programmable device to generate a computer-implemented process. Thus, the instructions executed on the computer or other programmable device provide steps for realizing the functions specified in Figure 1 one flow or multiple flows and / or blocks Figure 1 one block or multiple blocks.

[0066] The above is only the preferred embodiment of the present disclosure and is not used to limit the present disclosure. For those skilled in the art, various changes and modifications can be made to the present disclosure. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present disclosure should be included within the protection scope of the present disclosure.

Claims

1. A drilling production safety management and control method based on knowledge graph, characterized in that: include: (1) Based on the dangerous behaviors and illegal behaviors in the drilling production operation specifications, a production safety knowledge graph is constructed, which includes behavior condition nodes and dangerous event nodes, combined condition relationships and weight coefficients; (2) Use image acquisition equipment in the production area to perform real-time image acquisition and obtain videos or images of the operation area; (3) Use the YOLOv5 target recognition algorithm to identify targets in images or video frames and determine the presence of one or more behaviors; (4) Based on the acquired one or more behaviors, the possible dangers that the actions may lead to and the possible behavioral sources of the corresponding dangers, as well as the association weights between the behaviors and the dangers, are inferred to give the possible dangers and risk probabilities; (5) Issue warnings or take countermeasures based on the given danger and risk probability information; (a) Behavior condition nodes, which include behaviors that may cause danger or violation in drilling production operation specifications. Each behavior is an independent node in the knowledge graph. The behavior here is composed of a single judgment condition or a combination of judgment conditions; (b) Dangerous event nodes, which include the consequences or risks of the behaviors mentioned in the drilling production operation specifications, where each consequence or risk is an independent node in the knowledge graph; The relationships and weights of the production safety knowledge graph include: (a) Combination conditional relationship: one or several behavior conditions may lead to consequences or risks. This relationship is used as an edge to form the knowledge graph. Specifically, when one behavior condition or multiple behaviors can lead to an event, the relationship between nodes, that is, the connecting edge, is "independently caused". When an event is caused by several behavior conditions, the connecting edge is "jointly caused". (b) Weight coefficient. For a behavior that can lead to multiple dangerous events and an event that can be caused by multiple behaviors, the frequency of different behaviors leading to events is different. Therefore, a weight is assigned to each associated edge, which indicates the closeness of the association between the behavior and the result. For a behavior that can only lead to one event and the event can only be caused by it, the weight of their associated edge is the actual frequency of the dangerous event when the corresponding behavior condition exists; The method for inferring the probability of danger and risk is to calculate the probability of occurrence of the associated dangerous events one by one according to the identified behaviors, and the calculation method is: (a) For an event that can be caused by a single behavior condition or by the existence of multiple behaviors, the risk probability is: Where N is the total number of independent behavioral conditions identified, and p n Indicates the degree of connection between the corresponding behavior and the event, that is, the weight of the edge connecting the two nodes; (b) For an event caused by multiple behavioral conditions, the risk probability is: Where N represents the total number of associated behavior conditions, w n It indicates the degree of connection between the corresponding behavior and the event, that is, the weight of the edge connecting the two nodes.

2. A drilling production safety management and control method based on knowledge graph according to claim 1, characterized in that: The YOLOv5 target recognition algorithm is used to achieve the following effects: after effective training in an image or video data set collected in a drilling production environment, the model can recognize the behavioral actions involved in the training, determine whether there is a target behavior in the image, and can give the location and range of the target, where the range is represented by a rectangular edge position or a rectangular vertex position.

3. According to a method for drilling production safety management and control based on knowledge graph according to claim 1, it is characterized in that: The weight coefficient is obtained as follows: (a) According to the behavior in the drilling production operation specification, the weights are manually assigned, and the connections with higher correlation are given higher weights; (b) Based on the training set data taken at the drilling site and the actual frequency of occurrence of the corresponding dangerous events, the weight of the "independent cause" relationship edge is calculated by the ratio of the number of times the danger occurs when the behavior condition occurs to the number of times the behavior condition occurs; the weight of the "common cause" relationship edge is N j / N a , where N j N is the total number of times the corresponding behavior condition occurs when the danger occurs. a It is the total number of times all the behavior conditions associated with the hazard occur when the hazard occurs.

4. A drilling production safety management and control method based on knowledge graph according to claim 1, characterized in that: The hazard and risk probability is controlled by setting a threshold value θ to determine whether to issue an alarm or take countermeasures: When R is 1, when the danger probability exceeds the threshold, action is taken, otherwise no action is taken.

Images