An equipment data analysis system based on artificial intelligence

Abstract

The present invention relates to the field of data analysis technology, and in particular to an artificial intelligence-based equipment data analysis system, including a data acquisition module, a data processing module, a feature extraction module, an initial modeling module, and an update modeling module; the update modeling module determines the corresponding component characterization state, and determines the monitoring update strategy of the corresponding component according to the component characterization state, including: determining the motion characterization trend of the component according to the digital twin model to determine the update frequency of the component, and determining the monitoring method according to the update frequency; determining the motion characterization trend of each subcomponent in the component according to the digital twin model to determine the update frequency of each subcomponent, and determining the monitoring method according to the update frequency of the subcomponent. The present invention realizes the refined classification and targeted processing of each component of the equipment, can accurately allocate resources to components in different states, timely discover potential failure risks, and promote the intelligent and scientific management of the entire life cycle of the equipment.

Description

Technical Field

[0001] The present invention relates to the technical field of data analysis, and in particular to an artificial intelligence-based device data analysis system. Background Art

[0002] In the modern industrial field, with the increasing complexity of equipment and the continuous improvement of automation, accurate monitoring, analysis and efficient management of equipment operating status have become key requirements. Traditional equipment monitoring methods often rely on regular manual inspections, single sensor data collection and simple judgments based on fixed thresholds. This method has many limitations; manual inspections are difficult to achieve real-time and continuous monitoring, and it is easy to miss abnormal situations that occur instantly during the operation of the equipment. In addition, the accuracy and efficiency of the inspections are greatly affected by the professional level and working status of the inspectors; a single sensor can only obtain information on one aspect of the equipment and cannot fully reflect the overall operating status of the equipment. The judgment method based on fixed thresholds lacks flexibility and cannot adapt to changes in equipment under different working conditions and different operating stages. It is easy to have false alarms or missed alarms, and it is impossible to predict the occurrence of potential failures in advance. Often, a response can only be made after the failure has occurred and caused certain losses.

[0003] Chinese patent announcement number CN118606157B discloses a device data analysis system and method based on artificial intelligence, the data analysis method includes the following steps: obtaining the running time and average response time of the smart device; obtaining the theoretical response value of the smart device based on the running time and average response time; capturing several usage operations of the user in the process of contacting and leaving the smart device, generating a smart device usage event, and obtaining the actual response value of the smart device; generating performance-related events based on the obtained performance difference; analyzing the similarity between historical usage operations and performance-related events, and determining whether to generate new performance-related events; analyzing the user's habitual behavior, and generating the error rate of the impact of two consecutive performance-related events on performance; and obtaining the predicted time range of abnormality based on the frequency of occurrence of the user's habitual behavior and the performance difference corresponding to each habitual behavior. It can be seen that the invention has the following problems:

[0004] This invention fails to analyze the motion status of numerous components in complex equipment, and is unable to effectively distinguish components and sub-assemblies with different motion characteristics and formulate targeted monitoring strategies, resulting in a lack of accurate basis when formulating equipment maintenance plans, causing excessive maintenance to waste resources or insufficient maintenance to increase the risk of equipment failure. It is also not conducive to the deep optimization of equipment operating performance and scientific management of the entire life cycle. Summary of the invention

[0005] To this end, the present invention provides an artificial intelligence-based equipment data analysis system to overcome the problem that the prior art fails to analyze the motion states of numerous components in complex equipment, and fails to effectively distinguish components and sub-assemblies with different motion characteristics and formulate targeted monitoring strategies, resulting in a lack of accurate basis when formulating equipment maintenance plans.

[0006] To achieve the above object, the present invention provides an artificial intelligence-based device data analysis system, comprising:

[0007] A data acquisition module, used to collect binocular image sequences during the operation of the device through a binocular camera;

[0008] Among them, each binocular image includes the appearance shape, color, texture and movement status of each component of the device;

[0009] A data processing module, which is connected to the data acquisition module, and is used to perform denoising and enhancement processing on each of the binocular images to highlight the components of each of the binocular images, determine the disparity information of each pixel by performing stereo matching processing on the left image and the right image with the same timestamp, and determine the spatial depth information of each pixel according to the disparity information and the calibration parameters of the camera to construct a three-dimensional point cloud model of the device;

[0010] A feature extraction module, connected to the data processing module, for extracting key features of a device according to the three-dimensional point cloud model by computer vision technology, wherein the key features include shape features, motion features and surface texture features;

[0011] Wherein, the shape features include edge curves and geometric shapes, and the motion features include motion trajectories, speeds, and accelerations;

[0012] an initial modeling module, which is connected to the data processing module and the feature extraction module respectively, and is used to construct a digital twin model of the device according to the three-dimensional point cloud model and the key features, and to determine the parent-child relationship between the overall model of the device and each component model and the parent-child relationship between each component model and its internal components in the digital twin model;

[0013] An update modeling module, which is connected to the feature extraction module and the initial modeling module respectively, is used to determine the corresponding component representation state according to the parent-child relationship between the component model and its internal components, and determine the monitoring update strategy of the corresponding component according to the component representation state, including:

[0014] Determine a motion characterization trend of a component according to the digital twin model to determine an update frequency of the component, and determine a monitoring method according to the update frequency;

[0015] The motion characterization trend of each sub-component in the component is determined according to the digital twin model to determine the update frequency of each sub-component, and the monitoring method is determined according to the update frequency of the sub-component.

[0016] Furthermore, the feature extraction module extracts edge curves of various components in the device through an edge detection algorithm, and determines the geometric shape of the component through a shape analysis algorithm.

[0017] Furthermore, the feature extraction module determines the motion features of each component during operation by analyzing the changes in the positions of each component in a continuous binocular image sequence.

[0018] Furthermore, the update modeling module determines the number of moving subcomponents according to the motion state of each subcomponent in the component model, and determines the corresponding component representation state according to the number of moving subcomponents, including:

[0019] If the number of moving subcomponents within the component model is greater than 1, the component representation state of the component is determined to be a multi-element moving component;

[0020] If the number of moving subcomponents inside the component model is equal to 1, the component representation state of the component is determined to be a unit moving component;

[0021] If the number of moving subcomponents within the component model is less than 1, the component representation state of the component is determined to be a stationary component.

[0022] Furthermore, the update modeling module determines the monitoring update strategy of the corresponding component according to the component representation state, including:

[0023] If the component characterization state is a static component, the monitoring and updating strategy is to monitor and update the component at a minimum update frequency;

[0024] The minimum update frequency is once a month.

[0025] Furthermore, the update modeling module determines the monitoring update strategy of the corresponding component according to the component representation state, including:

[0026] If the component representation state is a multi-element moving component, the movement representation trend of each subcomponent in the component is determined according to the digital twin model to determine the update frequency of each subcomponent, and the monitoring method is determined according to the update frequency of the subcomponent;

[0027] If the component representation state is a unit motion state, the motion representation trend of the component is determined according to the digital twin model to determine the update frequency of the component, and the monitoring method is determined according to the update frequency.

[0028] Further, the updating modeling module determines whether the component or subassembly performs uniform reciprocating motion according to the digital twin model to determine the corresponding motion characterization trend, including:

[0029] If the component or subassembly performs uniform reciprocating motion, the corresponding motion representation trend is determined to be a regular operation trend;

[0030] If the component or subassembly does not perform uniform reciprocating motion, the corresponding motion characterization trend is determined to be an irregular operation trend.

[0031] Furthermore, the update modeling module determines the update frequency of the component according to the motion characterization trend of the component and determines the monitoring method according to the update frequency, including:

[0032] If the movement characterization trend of the component is a regular operation trend, the update frequency is determined according to its operation cycle and the monitoring method is determined to select a number of monitoring points on the movement path of the component for monitoring;

[0033] If the motion characterization trend of the component is an irregular operation trend, its update frequency is determined to be real-time update and real-time monitoring is performed based on the real-time updated digital twin model part corresponding to the component.

[0034] Furthermore, the update modeling module determines the update frequency of each sub-assembly according to the motion representation trend of each sub-assembly in the component, and determines the monitoring method according to the update frequency, including:

[0035] If the motion representation trends of each sub-component in the component are all regular operation trends, the update frequency of the component is determined according to the average operation cycle of each sub-component, and the detection method is determined to select at least one monitoring point on the motion path of each sub-component for monitoring;

[0036] If the motion characterization trends of all sub-components in the component are irregular operation trends, the update frequency of the component is determined to be real-time update and real-time monitoring is performed based on the real-time updated digital twin model corresponding to the component.

[0037] Furthermore, the update modeling module determines the update frequency of each sub-assembly according to the motion representation trend of each sub-assembly in the component, and determines the monitoring method according to the update frequency, and also includes:

[0038] If the motion characterization trends of each subcomponent in the component include regular operation trends and irregular operation trends, the subcomponents with irregular operation trends will be updated in real time and the subcomponents with regular operation trends will be updated at intervals according to the preset update frequency to obtain a real-time updated component digital twin model for real-time monitoring of the component.

[0039] Compared with the prior art, the beneficial effect of the present invention lies in that the artificial intelligence-based equipment data analysis system provided by the embodiment of the present invention accurately collects rich image data during the operation of the equipment through a binocular camera, and uses the data processing module to effectively process it to construct a three-dimensional point cloud model, providing an accurate geometric data basis for subsequent analysis; the feature extraction module is used to extract key features, which helps to deeply understand the material, wear and other conditions of the equipment; the initial modeling module is used to construct a digital twin model with a clear structure and clear relationships, which is convenient for local and overall control and simulation; the update modeling module can flexibly determine the update frequency and monitoring method according to the motion representation trend of the components and their sub-assemblies, so as to achieve efficient and accurate monitoring and analysis;

[0040] Furthermore, the updated modeling module can effectively distinguish the component characterization states of multi-moving components, unit moving components and stationary components by accurately determining the number of moving sub-components within the component model, so that personalized monitoring and management strategies can be formulated according to different component characteristics, and refined classification and targeted processing of various equipment components can be achieved, which helps to improve the accuracy and comprehensiveness of equipment data analysis, and plays a key role in ensuring stable equipment operation, optimizing equipment maintenance plans, and improving overall equipment management efficiency. It can accurately allocate resources to components in different states, timely discover potential failure risks, and promote intelligent and scientific management of the entire life cycle of equipment;

[0041] Furthermore, the update modeling module accurately formulates corresponding monitoring and updating strategies according to the component characterization status, and adopts the minimum update frequency for monitoring and updating of static components, effectively reducing unnecessary computing resources and time waste, and improving system operation efficiency while ensuring the integrity of the overall equipment model; for multiple moving parts, the update frequency and monitoring method are determined by in-depth analysis of the motion characterization trend of each sub-component, so as to achieve refined management of complex moving parts, timely capture subtle changes and potential fault hazards of each sub-component, and improve equipment operation reliability; and for unit moving parts, the update frequency and monitoring method are determined according to their specific motion characterization trend, ensuring effective monitoring of a single motion path on the basis of simplifying the monitoring process, which helps to improve the pertinence and accuracy of equipment monitoring, thereby comprehensively optimizing the equipment monitoring and management system, reducing equipment maintenance costs and extending equipment service life;

[0042] Furthermore, the update modeling module determines the motion characterization trend by judging whether the component or sub-assembly is performing uniform reciprocating motion. When a regular operating trend of uniform reciprocating motion is identified, relatively fixed and efficient monitoring parameters and update cycles can be set based on this stable motion mode, which helps to reduce the complexity and resource consumption of data processing, and can also make accurate performance predictions and fault warnings based on regular trends. When it is determined to be an irregular operating trend, the monitoring intensity and frequency can be strengthened in a targeted manner, and the update strategy can be flexibly adjusted to capture abnormal changes and potential risks in a timely manner. This plays an extremely critical role in ensuring the stable operation of the equipment, optimizing the equipment maintenance plan, and improving the intelligent level of equipment management. It can fully adapt to the monitoring needs of components or sub-assemblies with different motion characteristics, and effectively improve the reliability and adaptability of the entire equipment data analysis system.

[0043] Furthermore, when the component is in a unit motion state, the update modeling module determines the update frequency and monitoring method according to the motion representation state of the component: when the component motion representation trend is a regular operation trend, relying on the accurate judgment of the determinism of the motion trajectory, a small number of monitoring points (1 to 3) are selected on the motion path for real-time monitoring. The efficient method is used to reasonably determine the update frequency according to the operation cycle (update once every 50 to 100 times the operation cycle when there is no abnormality), which greatly reduces the amount of monitoring data and the occupation of computing resources. At the same time, it can respond to updates immediately when an abnormality occurs, and realize the optimal allocation of resources under the premise of ensuring the normal operation monitoring of the equipment; and when the motion representation trend is an irregular operation trend, by determining the real-time update frequency and real-time monitoring based on the real-time updated digital twin model part, it can comprehensively and timely capture every change in the component operation state, and not miss any potential abnormal situation. Regardless of whether it is a unit motion component with regular or irregular motion, the update modeling module can flexibly and accurately formulate monitoring and update strategies, effectively improve the accuracy and efficiency of equipment monitoring and adaptability to different operating conditions, escort the stable and reliable operation of the equipment, and help the intelligent and refined development of equipment management and maintenance work;

[0044] Furthermore, the update modeling module determines the update frequency and monitoring method according to the motion representation state of the sub-components in the component when the component is in a multivariate motion state: when the motion representation trends of each sub-component are all regular operation trends, a small number of monitoring points (1 to 2) can be set on a single sub-component to determine the certainty of the motion trajectory of each sub-component. Based on the average operation cycle of the sub-component, a reasonable component update frequency can be determined (updated once 50 to 100 times the average operation cycle when there is no abnormality), which effectively reduces the amount of monitoring data and computing resource consumption, and can respond to updates quickly in abnormal situations; when each sub-component has an irregular operation trend, comprehensive real-time monitoring is carried out through real-time update frequency and the digital twin model part based on real-time update to ensure that there is no abnormality. Any potential anomalies are ignored; and when the movement characterization trend of sub-components is both regular and irregular, an innovative strategy is adopted to combine the real-time update of irregular sub-components with the preset interval update of regular sub-components. This not only takes into account the rational use of resources and efficient monitoring, but also can flexibly build real-time digital twin models of components at different update nodes for precise monitoring. At the same time, when the regular sub-component update node has not been reached, its monitoring points are still monitored in real time, so as to fully and accurately adapt to the complex and diverse operating conditions of multiple moving components, greatly improving the accuracy and efficiency of equipment monitoring and the adaptability to the operating characteristics of multiple moving components, effectively ensuring the overall stable and reliable operation of the equipment, and promoting equipment management and maintenance to move forward in the direction of intelligence and refinement. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] Figure 1 A connection diagram of an artificial intelligence-based device data analysis system according to an embodiment of the present invention;

[0046] Figure 2 This is an analysis process diagram of the device data analysis system based on artificial intelligence according to an embodiment of the present invention;

[0047] Figure 3 A flowchart of updating a modeling module to determine a monitoring update strategy according to an embodiment of the present invention. DETAILED DESCRIPTION

[0048] In order to make the objects and advantages of the present invention more clearly understood, the present invention is further described below in conjunction with embodiments; it should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0049] The preferred embodiments of the present invention are described below with reference to the accompanying drawings. It should be understood by those skilled in the art that these embodiments are only used to explain the technical principles of the present invention and are not intended to limit the protection scope of the present invention.

[0050] It should be noted that, in the description of the present invention, terms such as "up", "down", "left", "right", "inside" and "outside" indicating directions or positional relationships are based on the directions or positional relationships shown in the drawings. This is merely for the convenience of description and does not indicate or imply that the device or element must have a specific orientation, be constructed and operated in a specific orientation. Therefore, it cannot be understood as a limitation on the present invention.

[0051] In addition, it should be noted that in the description of the present invention, unless otherwise clearly specified and limited, the terms "installed", "connected", and "connected" should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection, or it can be indirectly connected through an intermediate medium, or it can be the internal communication of two components. For those skilled in the art, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0052] See also Figure 1 and Figure 2 As shown, they are respectively a connection diagram of the device data analysis system based on artificial intelligence according to an embodiment of the present invention and an analysis process diagram of the device data analysis system based on artificial intelligence according to an embodiment of the present invention. An embodiment of the present invention provides a device data analysis system based on artificial intelligence, comprising:

[0053] The data acquisition module is used to collect binocular image sequences (binocular image sequences refer to sequences of binocular images) during the operation of the equipment through a binocular camera. It can be understood that the binocular camera synchronously captures two images (referred to as the left image and the right image) on the left and right sides. These images contain rich information such as the appearance shape, color, texture and movement state of the equipment's components. For example, for a CNC machine tool, the camera can capture details such as the position change of the tool, the movement of the worktable and the processing process of the workpiece.

[0054] Among them, each binocular image includes the appearance shape, color, texture and movement status of each component of the device;

[0055] In the implementation, in order to facilitate subsequent processing and analysis, the collected image data is timestamped and the collection time corresponding to each image is recorded, so as to establish the corresponding relationship between the image data and the equipment operation timeline;

[0056] A data processing module, which is connected to the data acquisition module, and is used to perform denoising and enhancement processing on each of the binocular images to highlight the components of each of the binocular images, determine the disparity information of each pixel by performing stereo matching processing on the left image and the right image with the same timestamp, and determine the spatial depth information of each pixel according to the disparity information and the calibration parameters of the camera to construct a three-dimensional point cloud model of the device;

[0057] It can be understood that the collected binocular images are denoised by using filtering algorithms (such as Gaussian filtering, median filtering, etc.) to remove noise interference in the image and improve the clarity and quality of the image; among them, Gaussian filtering can smooth the noise in the image while retaining the edge and detail information of the image; median filtering is more effective in removing impulse noise;

[0058] It is understood that image enhancement operations include contrast enhancement, brightness adjustment, etc., to highlight the key features and components of the device. By adjusting the gray value distribution or color balance of the image, the device's contour, texture and other features are made more obvious, which is convenient for subsequent feature extraction and analysis. For some device components with darker surface colors or uneven lighting, appropriately increasing the contrast can make them easier to identify and distinguish in the image;

[0059] In implementation, when performing image enhancement operation, if it is determined that the light of the acquired binocular image is uneven, the aperture of the binocular camera is controlled to increase to ensure image quality;

[0060] In the implementation, the disparity information of each pixel is calculated by finding the matching relationship between the corresponding pixels in the left image and the right image. Common stereo matching algorithms include feature-based matching algorithms (such as SIFT, SURF and other feature point matching) and region-based matching algorithms (such as SAD, SSD, etc.); according to the disparity information and the calibration parameters of the camera, the spatial depth information corresponding to each pixel in the image is calculated, so as to construct a three-dimensional point cloud model of the equipment. The depth information can reflect the position relationship and distance of each component of the equipment in space, and provides an important geometric data foundation for the construction of the digital twin model; in one implementation, for a complex mechanical assembly equipment, the relative position and assembly gap between each component can be accurately determined through the depth information;

[0061] A feature extraction module, connected to the data processing module, for extracting key features of a device according to the three-dimensional point cloud model by computer vision technology, wherein the key features include shape features, motion features and surface texture features;

[0062] In implementation, the surface texture features of the equipment are analyzed, and texture feature extraction algorithms (such as grayscale co-occurrence matrix, local binary pattern, etc.) can be used to extract texture information on the equipment surface. These texture features can be used for equipment material identification, surface wear detection, etc.; in one implementation, by analyzing the texture changes on the surface of a workpiece of a metal processing equipment, the degree of tool wear and changes in processing quality can be determined.

[0063] Wherein, the shape features include edge curves and geometric shapes, and the motion features include motion trajectories, speeds, and accelerations;

[0064] an initial modeling module, which is connected to the data processing module and the feature extraction module respectively, and is used to construct a digital twin model of the device according to the three-dimensional point cloud model and the key features, and to determine the parent-child relationship between the overall model of the device and each component model and the parent-child relationship between each component model and its internal components in the digital twin model;

[0065] In implementation, according to the 3D point cloud model and the extracted feature information, a 3D geometric model of the factory equipment is created using 3D modeling software (such as Blender, Maya, 3ds Max, etc.) or professional industrial modeling tools (such as Siemens NX, CATIA, etc.). In the modeling process, the extracted component shape features are accurately converted into geometric elements of the 3D model based on the actual size, shape and structure of the equipment, and the 3D models of the equipment's shell, internal structure, transmission parts and other components are constructed; then the basic architecture of the digital twin model is built. The construction of the digital twin model is a prior art and will not be elaborated on here; a parent-child relationship is established between the overall model of the equipment and the models of each component, and a parent-child relationship is established between each component model and its internal components, so that local and overall control can be conveniently performed during model updates and interactive operations.

[0066] An update modeling module, which is connected to the feature extraction module and the initial modeling module respectively, is used to determine the corresponding component representation state according to the parent-child relationship between the component model and its internal components, and determine the monitoring update strategy of the corresponding component according to the component representation state, including:

[0067] Determine a motion characterization trend of a component according to the digital twin model to determine an update frequency of the component, and determine a monitoring method according to the update frequency;

[0068] The motion characterization trend of each sub-component in the component is determined according to the digital twin model to determine the update frequency of each sub-component, and the monitoring method is determined according to the update frequency of the sub-component.

[0069] It can be understood that the artificial intelligence-based equipment data analysis system provided by the embodiment of the present invention uses a binocular camera to accurately collect rich image data during the operation of the equipment, and uses the data processing module to effectively process it to construct a three-dimensional point cloud model, thereby providing an accurate geometric data basis for subsequent analysis; the feature extraction module is used to extract key features, which helps to gain a deep understanding of the material, wear and other conditions of the equipment; the initial modeling module is used to construct a digital twin model with a clear structure and clear relationships, which is convenient for local and overall control and simulation; the update modeling module can flexibly determine the update frequency and monitoring method based on the motion representation trend of the components and their sub-assemblies, thereby achieving efficient and accurate monitoring and analysis: (1) The initial modeling module creates a three-dimensional geometric model of the factory equipment based on the three-dimensional point cloud model and key features , and determine the parent-child relationship between the overall equipment model and each component model, as well as between each component model and its internal components in the digital twin model. This clear hierarchical structure facilitates flexible local and overall control during model updates and interactive operations, so that the digital twin model can more realistically simulate the actual operation of the equipment and provide effective support for equipment management and optimization; (2) The update modeling module determines the corresponding update frequency and monitoring method based on the motion representation trend of the component and its sub-components. It can adjust the monitoring strategy in a targeted manner according to the actual operation dynamics of different parts and components of the equipment, avoid excessive or insufficient monitoring, improve monitoring efficiency while reducing resource consumption, realize intelligent and refined monitoring and management of equipment, timely discover potential problems and take corresponding measures to ensure stable and efficient operation of the equipment.

[0070] Specifically, the feature extraction module extracts edge curves of components in the device through edge detection algorithms (such as Canny edge detection), and determines the geometric shapes of the components through shape analysis algorithms (such as Hough transform detection of circles, rectangles, etc.).

[0071] Specifically, the feature extraction module determines the motion features of each component during operation by analyzing the position changes of each component in a continuous binocular image sequence.

[0072] It can be understood that by analyzing and calculating the changes in the positions of components in a continuous binocular image sequence, the motion parameters of the device during operation can be obtained; in one implementation, for a rotating device, its rotation speed can be calculated based on the angular changes of the rotating components in the image; for a linear motion device, its linear motion speed and acceleration can be calculated by tracking the coordinate changes of the components in the image; in one implementation, for a CNC machine tool, the binocular camera can capture details such as the position changes of the tool, the movement of the worktable, and the processing process of the workpiece.

[0073] See also Figure 3 As shown, it is a flow chart of updating the modeling module to determine the monitoring update strategy according to an embodiment of the present invention.

[0074] Specifically, the update modeling module determines the number of moving subcomponents according to the motion state of each subcomponent in the component model, and determines the corresponding component representation state according to the number of moving subcomponents, including:

[0075] If the number of moving subcomponents within the component model is greater than 1, the component representation state of the component is determined to be a multi-element moving component;

[0076] If the number of moving subcomponents inside the component model is equal to 1, the component representation state of the component is determined to be a unit moving component;

[0077] If the number of moving subcomponents within the component model is less than 1, the component representation state of the component is determined to be a stationary component.

[0078] It can be understood that a single component must be composed of several parts, and a single component is also composed of several components; in practice, if a single subcomponent inside a component is moving, it is a moving subcomponent, and if a single subcomponent inside a component is not moving, it is a stationary subcomponent.

[0079] It can be understood that the updated modeling module can effectively distinguish the component characterization states of multi-moving components, unit moving components and stationary components by accurately determining the number of moving sub-components within the component model, so that personalized monitoring and management strategies can be formulated according to different component characteristics, and refined classification and targeted processing of various equipment components can be achieved. This helps to improve the accuracy and comprehensiveness of equipment data analysis, and plays a key role in ensuring stable equipment operation, optimizing equipment maintenance plans, and improving overall equipment management efficiency. It can accurately allocate resources to components in different states, promptly discover potential failure risks, and promote the intelligent and scientific management of the entire life cycle of equipment.

[0080] Specifically, the update modeling module determines the monitoring update strategy of the corresponding component according to the component representation state, including:

[0081] If the component characterization state is a static component, the monitoring and updating strategy is to monitor and update the component at a minimum update frequency;

[0082] The minimum update frequency is once a month.

[0083] It is understandable that for static components, each time the digital twin model is updated, the component is replaced by the last updated model; since static components do not move, they should be fault-free / error-free during the entire device operation, so there is no need to monitor and update them in real time, and they only need to be updated at the minimum update frequency (i.e. once a month);

[0084] In implementation, the update frequency is ≥ 1 time / month.

[0085] Specifically, the update modeling module determines the monitoring update strategy of the corresponding component according to the component representation state, including:

[0086] If the component representation state is a multi-motion component, the motion representation trend of each subcomponent in the component is determined according to the digital twin model to determine the update frequency of each subcomponent, and the monitoring method is determined according to the update frequency of the subcomponent; it can be understood that for a component with a multi-motion state, it contains several motion modes / paths, that is, the motion of the component is multi-motion, then it is necessary to split the component into several components, and then determine how to update the model part of the component in the digital twin model according to the motion representation trend of each component;

[0087] If the component representation state is a unit motion state, the motion representation trend of the component is determined according to the digital twin model to determine the update frequency of the component, and the monitoring method is determined according to the update frequency; it can be understood that for components in a unit motion state, there is only one motion mode / path inside it, and it is only necessary to judge the motion mode / path to determine the update frequency and monitoring method of the component.

[0088] It can be understood that the update modeling module accurately formulates corresponding monitoring and updating strategies according to the component characterization status, and adopts the minimum update frequency for monitoring and updating of static components, which effectively reduces unnecessary computing resources and time waste, and improves the system operation efficiency while ensuring the integrity of the overall equipment model; for multiple moving parts, the update frequency and monitoring method are determined by in-depth analysis of the motion characterization trend of each sub-component, so as to achieve refined management of complex moving parts, timely capture subtle changes and potential fault hazards of each sub-component, and improve the reliability of equipment operation; and for unit moving parts, the update frequency and monitoring method are determined according to their specific motion characterization trends, and effective monitoring of a single motion path is ensured on the basis of simplifying the monitoring process, which helps to improve the pertinence and accuracy of equipment monitoring, thereby comprehensively optimizing the equipment monitoring and management system, reducing equipment maintenance costs and extending equipment service life.

[0089] Specifically, the update modeling module determines whether the component or subassembly performs uniform reciprocating motion according to the digital twin model to determine the corresponding motion characterization trend, including:

[0090] If a component or subassembly performs uniform reciprocating motion, the corresponding motion characterization trend is determined to be a regular operation trend; that is, if a component performs uniform reciprocating motion, the motion characterization trend of the component is a regular operation trend, and if a subassembly performs uniform reciprocating motion, the motion characterization trend of the component is a regular operation trend;

[0091] If the component or sub-assembly does not perform uniform reciprocating motion, the corresponding motion characterization trend is determined to be an irregular operation trend; that is, if the component does not perform uniform reciprocating motion, the motion characterization trend of the component is an irregular operation trend, and if the sub-assembly does not perform uniform reciprocating motion, the motion characterization trend of the component is an irregular operation trend.

[0092] It can be understood that whether each component / subassembly performs uniform reciprocating motion can be determined in the constructed digital twin model, and it can also be known through the structure and operation principle of the equipment before constructing the digital twin model.

[0093] It can be understood that the update modeling module determines the motion characterization trend by judging whether the components or sub-assemblies are performing uniform reciprocating motion. When a regular operating trend of uniform reciprocating motion is identified, relatively fixed and efficient monitoring parameters and update cycles can be set based on this stable motion pattern, which helps to reduce the complexity and resource consumption of data processing, and at the same time can make accurate performance predictions and fault warnings based on regular trends; and when it is determined to be an irregular operating trend, it can strengthen the monitoring intensity and frequency in a targeted manner, and flexibly adjust the update strategy to capture abnormal changes and potential risks in a timely manner. It plays an extremely critical role in ensuring the stable operation of the equipment, optimizing the equipment maintenance plan, and improving the intelligent level of equipment management. It can fully adapt to the monitoring needs of components or sub-assemblies with different motion characteristics, and effectively improve the reliability and adaptability of the entire equipment data analysis system.

[0094] Specifically, when the component representation state is a unit motion state, the update modeling module determines the update frequency of the component according to the motion representation trend of the component and determines the monitoring method according to the update frequency, including:

[0095] If the motion characterization trend of the component is a regular operation trend, it means that the operation trajectory and operation mode of the component are determined, and there is no need to monitor the entire operation state. It is only necessary to monitor a few points on the motion trajectory to know whether the operation of the component is normal. According to its operation cycle, the update frequency of the component is determined and the monitoring method is determined to select a number of monitoring points on the motion path of the component (usually about 3 monitoring points are selected) for real-time monitoring; it can be understood that in the process of real-time monitoring of each monitoring point, if no abnormal situation occurs, the update frequency is usually set to update once every 100 times the operation cycle to once every 50 times the operation cycle. If an abnormal situation occurs, the part of the component in the digital twin model is updated immediately;

[0096] If the motion characterization trend of a component is an irregular operation trend, it means that the operation trajectory and operation mode of the component are irregular, and the entire operation status needs to be monitored and updated. Therefore, the update frequency is determined to be real-time update and real-time monitoring is performed based on the real-time updated digital twin model corresponding to the component.

[0097] It can be understood that the update modeling module determines the update frequency and monitoring method according to the motion representation state of the component when the component is in the unit motion state: when the component motion representation trend is a regular operation trend, relying on the accurate judgment of the determinism of the motion trajectory, a small number of monitoring points (1 to 3) are selected on the motion path for real-time monitoring. The efficient method is adopted, and the update frequency is reasonably determined according to the operation cycle (updated once every 50 to 100 times the operation cycle when there is no abnormality), which greatly reduces the amount of monitoring data and the occupation of computing resources. At the same time, it can respond to updates immediately when an abnormality occurs, and realize the optimal allocation of resources under the premise of ensuring the normal operation monitoring of the equipment; and when the motion representation trend is an irregular operation trend, by determining the real-time update frequency and real-time monitoring based on the real-time updated digital twin model part, it can comprehensively and timely capture every change in the component operation state, and not miss any potential abnormal situation. Regardless of whether it is a unit motion component with regular or irregular motion, the update modeling module can flexibly and accurately formulate monitoring and update strategies, effectively improve the accuracy and efficiency of equipment monitoring and adaptability to different operating conditions, safeguard the stable and reliable operation of the equipment, and help the intelligent and refined development of equipment management and maintenance.

[0098] Specifically, when the component representation state is a multivariate motion state, the update modeling module determines the update frequency of each subcomponent according to the motion representation trend of each subcomponent in the component, and determines the monitoring method according to the update frequency, including:

[0099] If the motion characterization trends of each subcomponent in the component are all regular operation trends, it means that the operation trajectory and operation mode of each component are determined, and there is no need to monitor the operation status of the entire component in real time. It is only necessary to monitor a few points on each motion trajectory to know whether the operation of this component is normal (usually 1 to 2 monitoring points are determined on a single component). The update frequency of the component is determined according to the average operation cycle of each subcomponent, and the detection method is determined to select at least one monitoring point on the motion path of each subcomponent for monitoring; it can be understood that in the process of real-time monitoring of each monitoring point, if no abnormal situation occurs, the update frequency is usually set to update once every 100 times the average operation cycle to once every 50 times the average operation cycle. If an abnormal situation occurs, the part of the component corresponding to each component in the digital twin model is updated immediately;

[0100] If the motion characterization trends of each subcomponent in the component are irregular operation trends, it means that the operation trajectory and operation mode of each subcomponent in the component are irregular, and the entire operation status of the component needs to be monitored and updated. Therefore, the update frequency of the component is determined to be real-time update and real-time monitoring is performed based on the real-time updated digital twin model part corresponding to the component.

[0101] Specifically, the update modeling module determines the update frequency of each sub-assembly according to the motion representation trend of each sub-assembly in the component, and determines the monitoring method according to the update frequency, and also includes:

[0102] If the motion characterization trends of each subcomponent in the component include regular operation trends and irregular operation trends (that is, some subcomponents in a single component have regular operation trends, and some subcomponents have irregular operation trends), the subcomponents with irregular operation trends are updated in real time and the subcomponents with regular operation trends are updated at intervals according to a preset update frequency to obtain a real-time updated component digital twin model for real-time monitoring of the component; it can be understood that before the update node of the subcomponent with a regular operation trend is reached, the real-time digital twin model of the component is composed of the real-time updated subcomponent model of the irregular operation trend and the subcomponent model of the regular operation trend updated at the previous update node; when the update node of the subcomponent with a regular operation trend is reached, the real-time digital twin model of the component is composed of the real-time updated subcomponent model of the irregular operation trend and the subcomponent model of the regular operation trend updated at the current update node;

[0103] In implementation, when the update node of the regular operation trend has not been reached, it is also necessary to perform real-time monitoring on the monitoring points of the subcomponents with the regular operation trend (select about 2 monitoring points on the operation path of the subcomponents).

[0104] It can be understood that when the component is in a multivariate motion state, the update modeling module determines the update frequency and monitoring method according to the motion representation state of the subcomponents in the component: when the motion representation trends of each subcomponent are regular operation trends, a small number of monitoring points (1 to 2) can be set on a single subcomponent to determine the certainty of the motion trajectory of each subcomponent. Based on the average operation cycle of the subcomponent, a reasonable component update frequency is determined (updated once 50 to 100 times the average operation cycle when there is no abnormality), which effectively reduces the amount of monitoring data and computing resource consumption, and can respond to updates quickly in abnormal situations; when each subcomponent has an irregular operation trend, comprehensive real-time monitoring is carried out through real-time update frequency and the digital twin model part based on real-time update to ensure Do not miss any potential anomalies; and when the movement characterization trends of sub-components are both regular and irregular, an innovative strategy is adopted to combine the real-time update of irregular sub-components with the preset interval update of regular sub-components. This not only takes into account the rational use of resources and efficient monitoring, but also can flexibly build real-time digital twin models of components at different update nodes for precise monitoring. At the same time, when the regular sub-component update node has not been reached, its monitoring points are still monitored in real time, thereby fully and accurately adapting to the complex and diverse operating conditions of multiple moving components, greatly improving the accuracy and efficiency of equipment monitoring and the adaptability to the operating characteristics of multiple moving components, effectively ensuring the overall stable and reliable operation of the equipment, and promoting equipment management and maintenance to move forward in the direction of intelligence and refinement.

[0105] So far, the technical solutions of the present invention have been described in conjunction with the preferred embodiments shown in the accompanying drawings. However, it is easy for those skilled in the art to understand that the protection scope of the present invention is obviously not limited to these specific embodiments. Without departing from the principle of the present invention, those skilled in the art can make equivalent changes or substitutions to the relevant technical features, and the technical solutions after these changes or substitutions will fall within the protection scope of the present invention.

[0106] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. An artificial intelligence-based equipment data analysis system, characterized in that: include: A data acquisition module, used to collect binocular image sequences during the operation of the device through a binocular camera; Among them, each binocular image includes the appearance shape, color, texture and movement status of each component of the device; A data processing module, which is connected to the data acquisition module, and is used to perform denoising and enhancement processing on each of the binocular images to highlight the components of each of the binocular images, determine the disparity information of each pixel by performing stereo matching processing on the left image and the right image with the same timestamp, and determine the spatial depth information of each pixel according to the disparity information and the calibration parameters of the camera to construct a three-dimensional point cloud model of the device; A feature extraction module, connected to the data processing module, for extracting key features of a device according to the three-dimensional point cloud model by computer vision technology, wherein the key features include shape features, motion features and surface texture features; Wherein, the shape features include edge curves and geometric shapes, and the motion features include motion trajectories, speeds, and accelerations; an initial modeling module, which is connected to the data processing module and the feature extraction module respectively, and is used to construct a digital twin model of the device according to the three-dimensional point cloud model and the key features, and to determine the parent-child relationship between the overall model of the device and each component model and the parent-child relationship between each component model and its internal components in the digital twin model; An update modeling module, which is connected to the feature extraction module and the initial modeling module respectively, is used to determine the corresponding component representation state according to the parent-child relationship between the component model and its internal components, and determine the monitoring update strategy of the corresponding component according to the component representation state, including: Determine a motion characterization trend of a component according to the digital twin model to determine an update frequency of the component, and determine a monitoring method according to the update frequency; Determine the motion characterization trend of each subassembly in the component according to the digital twin model to determine the update frequency of each subassembly, and determine the monitoring method according to the update frequency of the subassembly; The update modeling module determines the number of moving sub-components according to the motion state of each sub-component inside the component model, and determines the corresponding component representation state according to the number of moving sub-components; Wherein, the component representation states include multi-element moving components, unit moving components and stationary components; The update modeling module determines whether the component or subassembly performs uniform reciprocating motion according to the digital twin model to determine the corresponding motion characterization trend; The movement characterization trend includes a regular movement trend and an irregular movement trend.

2. The artificial intelligence-based equipment data analysis system according to claim 1, characterized in that: The feature extraction module extracts edge curves of various components in the device through an edge detection algorithm, and determines the geometric shape of the components through a shape analysis algorithm.

3. The artificial intelligence-based equipment data analysis system according to claim 1, characterized in that: The feature extraction module determines the motion features of each component during operation by analyzing the position changes of each component in a continuous binocular image sequence.

4. The artificial intelligence-based equipment data analysis system according to claim 1, characterized in that: The update modeling module determines the number of moving sub-components according to the motion state of each sub-component in the component model, and determines the corresponding component representation state according to the number of moving sub-components, including: If the number of moving subcomponents within the component model is greater than 1, the component representation state of the component is determined to be a multi-element moving component; If the number of moving subcomponents inside the component model is equal to 1, the component representation state of the component is determined to be a unit moving component; If the number of moving subcomponents within the component model is less than 1, the component representation state of the component is determined to be a stationary component.

5. The artificial intelligence-based equipment data analysis system according to claim 4, characterized in that: The update modeling module determines the monitoring update strategy of the corresponding component according to the component representation state, including: If the component characterization state is a static component, the monitoring and updating strategy is to monitor and update the component at a minimum update frequency; The minimum update frequency is once a month.

6. The artificial intelligence-based equipment data analysis system according to claim 5, characterized in that: The update modeling module determines the monitoring update strategy of the corresponding component according to the component representation state, including: If the component representation state is a multi-element moving component, the movement representation trend of each subcomponent in the component is determined according to the digital twin model to determine the update frequency of each subcomponent, and the monitoring method is determined according to the update frequency of the subcomponent; If the component representation state is a unit motion state, the motion representation trend of the component is determined according to the digital twin model to determine the update frequency of the component, and the monitoring method is determined according to the update frequency.

7. The artificial intelligence-based equipment data analysis system according to claim 6, characterized in that: The updating modeling module determines whether the component or subassembly performs uniform reciprocating motion according to the digital twin model to determine the corresponding motion characterization trend, including: If the component or subassembly performs uniform reciprocating motion, the corresponding motion representation trend is determined to be a regular operation trend; If the component or subassembly does not perform uniform reciprocating motion, the corresponding motion characterization trend is determined to be an irregular operation trend.

8. The artificial intelligence-based equipment data analysis system according to claim 6, characterized in that: The update modeling module determines the update frequency of the component according to the motion characterization trend of the component and determines the monitoring method according to the update frequency, including: If the movement characterization trend of the component is a regular operation trend, the update frequency is determined according to its operation cycle and the monitoring method is determined to select a number of monitoring points on the movement path of the component for monitoring; If the motion characterization trend of the component is an irregular operation trend, its update frequency is determined to be real-time update and real-time monitoring is performed based on the real-time updated digital twin model part corresponding to the component.

9. The artificial intelligence-based equipment data analysis system according to claim 6, characterized in that: The update modeling module determines the update frequency of each sub-assembly according to the motion representation trend of each sub-assembly in the component, and determines the monitoring method according to the update frequency, including: If the motion representation trends of each sub-component in the component are all regular operation trends, the update frequency of the component is determined according to the average operation cycle of each sub-component, and the detection method is determined to select at least one monitoring point on the motion path of each sub-component for monitoring; If the motion characterization trends of all sub-components in the component are irregular operation trends, the update frequency of the component is determined to be real-time update and real-time monitoring is performed based on the real-time updated digital twin model corresponding to the component.

10. The artificial intelligence-based equipment data analysis system according to claim 9, characterized in that: The update modeling module determines the update frequency of each sub-assembly according to the motion representation trend of each sub-assembly in the component, and determines the monitoring method according to the update frequency, and also includes: If the motion characterization trends of each subcomponent in the component include regular operation trends and irregular operation trends, the subcomponents with irregular operation trends will be updated in real time and the subcomponents with regular operation trends will be updated at intervals according to the preset update frequency to obtain a real-time updated component digital twin model for real-time monitoring of the component.

Images