Digital twin-enhanced turnaround maintenance process optimization and visualization system

Abstract

The present invention discloses a digital twin enhanced transfer maintenance process optimization and visualization system, which belongs to the field of equipment maintenance technology, and specifically includes: a digital twin model building module and a maintenance optimization module; the digital twin model building module quickly generates and annotates a full three-dimensional model of the transfer equipment through information acquisition, three-dimensional modeling and information annotation; the maintenance optimization module customizes the maintenance process through the three units of process definition, process execution and process optimization, uses the benchmark targeting strategy, adopts the improved energy valley optimization algorithm to evaluate the maintenance task scheduling plan, and uses the ESIA algorithm to evaluate and optimize the maintenance process, thereby realizing efficient, accurate and intelligent management of transfer equipment maintenance, and improving maintenance efficiency and equipment reliability.

Description

Technical Field

[0001] The present invention belongs to the technical field of equipment maintenance, and specifically to a digital twin enhanced turnaround maintenance process optimization and visualization system. Background Art

[0002] At present, my country still mainly relies on coal-fired power generation. According to the national energy policy, the key to developing thermal power units is to solve the two major problems of energy conservation and environmental protection. With the gradual deepening of the market-oriented reform of electricity, the cost of power generation has become a decisive factor in the bidding for grid access. The proportion of the maintenance costs of thermal power equipment, especially fans, coal mills, circulating pumps, motors and other easily damaged transfer equipment, in the power cost is getting higher and higher. The current old maintenance system based on preventive maintenance is increasingly unable to meet the requirements of business operators. In order to ensure that the transfer equipment can operate long-term, stably and trouble-free under the required performance conditions, power plants currently carry out careful maintenance and regular maintenance of fans.

[0003] At present, the maintenance system in my country is characterized by planning, combining daily troubleshooting and maintenance with post-maintenance processing, and mainly includes the following contents: equipment troubleshooting records and equipment data management based on paper ledgers; spare parts management and regular inspection of units; troubleshooting, daily cleaning, and lubrication grease maintenance after equipment withdrawal. The maintenance personnel of power generation enterprises can only troubleshoot according to the alarm content and maintenance manual. The maintenance work lacks pertinence and the maintenance effect is difficult to evaluate. Such a maintenance process is bound to result in untimely discovery of problems and delay the maintenance process, causing unnecessary losses to the power plant.

[0004] For example, the Chinese patent with the authorization announcement number CN114021745B discloses a virtual maintenance process optimization method for hydropower station equipment, including: establishing a three-dimensional model of the equipment; obtaining a constraint matrix model of the equipment; establishing disassembly evaluation indicators; establishing the objective function and calculating the workload matrix; using the improved discrete whale algorithm to solve the optimal disassembly sequence; using Unity3D to recognize the sequence and automatically play the disassembly animation. The whale algorithm proposed in this technical solution combines the optimization heuristic mutation and the round-trip optimization operator, which can well balance the global search ability and the local search ability, and can efficiently solve NP problems. In addition, Unity3D can complete the three-dimensional visualization of the disassembly process with feature-rich scripts, which helps to improve maintenance efficiency.

[0005] The above existing technologies all have the following problems: there are limitations in the application field; an improved discrete whale algorithm is introduced to solve the optimal disassembly sequence, but the unique advantages or innovations of the algorithm in the maintenance of hydropower station equipment are not explained in detail; Unity3D is used for three-dimensional visualization of the disassembly process, but there is no mention of how to integrate and interact these visualization information with other links of the maintenance process; when setting up disassembly evaluation indicators and establishing objective functions, more emphasis may be placed on the optimization of the disassembly process, while ignoring the integrity and flexibility of the maintenance process. Summary of the invention

[0006] In view of the shortcomings of the prior art, the present invention proposes a digital twin enhanced transfer maintenance process optimization and visualization system, including a digital twin model building module and a maintenance optimization module; the digital twin model building module quickly generates and annotates the full three-dimensional model of the transfer equipment through information acquisition, three-dimensional modeling and information labeling; the maintenance optimization module uses the benchmark targeting strategy to customize the maintenance process through the three units of process definition, process execution and process optimization, adopts the improved energy valley optimization algorithm to evaluate the maintenance task scheduling plan, and uses the ESIA algorithm to evaluate and optimize the maintenance process, thereby realizing efficient, accurate and intelligent management of transfer equipment maintenance, and improving maintenance efficiency and equipment reliability.

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

[0008] The digital twin-enhanced turnaround maintenance process optimization and visualization system includes:

[0009] The digital twin model building module includes an information acquisition unit, a three-dimensional modeling unit, and an information annotation unit; the information acquisition unit is used to acquire the information of the parts of the transfer equipment; the three-dimensional modeling unit is configured with a three-dimensional reconstruction strategy, and the three-dimensional reconstruction strategy is used to quickly generate a three-dimensional model of the transfer equipment; the information annotation unit is used to annotate the engineering information on the three-dimensional model according to the preset annotation rules to form a full three-dimensional display mode;

[0010] The maintenance optimization module includes a process definition unit, a process execution unit, and a process optimization unit; the process definition unit is configured with a benchmark targeting strategy, which is used to customize the maintenance process according to actual conditions and needs, and complete the decomposition, sorting and combination of maintenance tasks; the process execution unit is configured with an improved energy valley optimization algorithm, which is used to evaluate the advantages and disadvantages of different maintenance task scheduling schemes; the process optimization unit is configured with an ESIA algorithm, which includes elimination, simplification, integration and automation, and is used to evaluate and optimize the maintenance process;

[0011] The specific steps of the benchmarking strategy include:

[0012] B1: Clarify the target of the benchmark and generate a target specification for maintenance process optimization;

[0013] B2: According to the maintenance process optimization target specification, select benchmark objects through research and obtain the maintenance process information of benchmark objects;

[0014] B3: Compare and analyze your own maintenance process with that of the benchmark object to obtain gap information, and organize the analysis results to form a maintenance process gap analysis report;

[0015] B4: According to the maintenance process gap analysis report, formulate targeted improvement plans, and implement the optimization of maintenance processes step by step according to the requirements of the improvement plans, and form a maintenance process improvement implementation plan and execution report;

[0016] B5: Evaluate the improved maintenance process according to the maintenance process improvement implementation plan and execution report, and form a maintenance process optimization evaluation report based on the evaluation results. At the same time, provide feedback and make adjustments based on the evaluation report;

[0017] The specific steps of the improved energy valley optimization algorithm include:

[0018] C1: Define the objective function of maintenance task scheduling and initialize the particle swarm, where each particle represents a maintenance task scheduling scheme;

[0019] C2: Set the maximum number of iterations r and the number of particles n of the improved energy valley optimization algorithm, and calculate the objective function value for each maintenance task scheduling scheme;

[0020] C3: Determine the neutron enrichment level of each maintenance task scheduling plan based on the objective function value ;

[0021] The specific steps of the improved energy valley optimization algorithm also include:

[0022] C4: Setting the neutron enrichment level threshold ;

[0023] like , then decay may occur, and according to the decay type, the maintenance task scheduling plan is updated;

[0024] like , then there is a deviation in the maintenance task scheduling plan, that is, fine-tune the maintenance task scheduling plan;

[0025] C5: Repeat the above steps until the maximum number of iterations is reached, and in each iteration, update the optimal maintenance task scheduling plan according to the update of the maintenance task scheduling plan;

[0026] C6: Output the optimal maintenance task scheduling plan.

[0027] Specifically, the specific steps of the three-dimensional reconstruction strategy include:

[0028] A1: Use a laser scanner or camera to collect images, videos or laser scanned image data of the transfer equipment, and pre-process the collected image data;

[0029] A2: Extract features from the preprocessed image data and use the corner point detection method to extract feature points of the transfer equipment to obtain the transfer equipment feature point data;

[0030] A3: Based on the feature point data of the transfer equipment, the edge detection method is used to extract the contour image information of the transfer equipment and generate the transfer equipment contour image dataset;

[0031] A4: Based on the generated transfer equipment contour image dataset, use a binocular camera to capture the transfer equipment contour images from different perspectives, and pre-process the captured transfer equipment contour images to obtain the transfer equipment visual images;

[0032] A5: Calculate the disparity map between the images taken by the left and right lenses of the same scene in the visual image of the transfer device, and convert the pixels in the image into point cloud data in three-dimensional space based on the camera's intrinsic parameter matrix and disparity map;

[0033] A6: Align the point cloud data from different perspectives, and fuse the registered point cloud data to construct a three-dimensional model of the transfer equipment.

[0034] Specifically, the specific steps of using the corner point detection method to extract the feature points of the transfer equipment in A2 include:

[0035] A2.1: Perform grayscale transformation on the preprocessed image data to generate a grayscale image of the transfer equipment , where R represents the range, A and B represent the row and column values ​​of the grayscale image of the transfer device, and , M represents the number of pixels in the grayscale image of the transfer device;

[0036] A2.2: Calculate the gradient of the grayscale image of the transfer device in the x and y directions at each pixel (p, q) and ,in, Represents the gradient of the grayscale image of the transfer device in the x direction at the pixel point (p, q), represents the Sobel operator function in the x direction, Represents the gradient of the grayscale image of the transfer device in the y direction at the pixel point (p, q), Represents the Sobel operator function in the y direction;

[0037] A2.3: For each pixel (p, q), calculate an autocorrelation matrix M cor = [ ∇ 2 x ∇ x ∇ y ∇ x ∇ y ∇ 2 y ] ;

[0038] A2.4: Based on the autocorrelation matrix , using the singular value decomposition algorithm Calculate eigenvalues and , where U and V represent orthogonal matrices, represents a diagonal matrix, and the diagonal elements are eigenvalues;

[0039] A2.5: Based on eigenvalues and , calculate the corner point responsivity ,in, represents the minimum function;

[0040] A2.6: Corner responsiveness for all pixels Sort in descending order and set a responsiveness threshold ;

[0041] like , then the pixel point is selected as the corner point, and the coordinates and responsiveness of the selected corner point are output as feature point data.

[0042] Specifically, the specific steps of A5 include:

[0043] A5.1: Receive visual images from the transfer device, perform camera calibration, and perform epipolar correction on the images taken by the left and right lenses so that the corresponding epipolar lines are parallel to each other;

[0044] A5.2: For each pixel in the left eye view, calculate the matching cost between it and the matching pixel in the right eye view, that is, the matching degree, and extend the matching cost of a single pixel to its neighborhood through a 5×5 sliding window to obtain the aggregated matching cost;

[0045] A5.3: Use the aggregated matching cost to calculate the optimal disparity value for each pixel, and refine and smooth the disparity value to obtain the final disparity map;

[0046] A5.4: Calculate the depth value of each pixel using the triangulation principle based on the camera's intrinsic parameter matrix and disparity map;

[0047] A5.5: Combine the depth value, camera parameters, and pixel coordinates to convert them into point cloud data in three-dimensional space.

[0048] Specifically, in the information acquisition unit, for regular transfer equipment parts, the transfer equipment parts information is acquired through drawings or actual surveying dimensions; for special-shaped transfer equipment parts, a three-dimensional laser scanning method is used to collect three-dimensional point cloud data of special-shaped or non-standard transfer equipment parts.

[0049] Specifically, the labeling rules preset in the information labeling unit include: dimension labeling rules, material labeling rules, and precision labeling rules for parts of transfer equipment.

[0050] Compared with the prior art, the present invention has the following beneficial effects:

[0051] 1. The present invention proposes a digital twin-enhanced turnaround maintenance process optimization and visualization system, and optimizes and improves the architecture, operating steps and processes. The system has the advantages of simple process, low investment and operating costs, and low production work costs.

[0052] 2. The present invention proposes a digital twin enhanced transfer maintenance process optimization and visualization system, wherein the digital twin model building module realizes the full three-dimensional visualization of the transfer equipment through information acquisition, three-dimensional modeling and information labeling, providing intuitive and accurate equipment information for maintenance work, which helps to improve maintenance efficiency and accuracy.

[0053] 3. The present invention proposes a digital twin-enhanced turnaround maintenance process optimization and visualization system, which customizes the maintenance process through benchmark targeting strategy, uses the improved energy valley optimization algorithm to evaluate the task scheduling plan, and uses the ESIA algorithm to evaluate and optimize the process, thereby achieving efficient, flexible and intelligent maintenance processes, and further improving the efficiency and reliability of equipment maintenance. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] Figure 1 This is the architecture diagram of the turnaround maintenance process optimization and visualization system enhanced by the digital twin of the present invention;

[0055] Figure 2 The digital structure diagram of the turnaround maintenance process optimization and visualization system maintenance enhanced by the digital twin of the present invention;

[0056] Figure 3 It is a flowchart of the overall implementation of the turnaround maintenance process optimization and visualization system enhanced by the digital twin of the present invention;

[0057] Figure 4 This is a flow chart of the disassembly and assembly visualization technology of the turnaround maintenance process optimization and visualization system enhanced by the digital twin of the present invention;

[0058] Figure 5 This is a flow chart of the three-dimensional reconstruction strategy of the digital twin-enhanced turnaround maintenance process optimization and visualization system of the present invention. DETAILED DESCRIPTION

[0059] Example 1

[0060] See also Figure 1-Figure 4 , an embodiment provided by the present invention: a digital twin enhanced turnaround maintenance process optimization and visualization system, comprising:

[0061] Digital twin model building module, maintenance optimization module, disassembly and assembly visualization module, and system architecture module;

[0062] The digital twin model building module creates a digital twin model of the transfer equipment through 3D modeling and information annotation technology, realizing the accurate mapping of physical entities to virtual space;

[0063] Maintenance optimization module, which is used to define, execute, monitor and optimize the maintenance process of transfer equipment to improve maintenance efficiency and quality;

[0064] The disassembly and assembly visualization module is used to realize the visualization management of the disassembly and assembly process of the transfer equipment through 3D models and animations, which is convenient for training and self-inspection;

[0065] The system architecture module is used to provide the basic architecture and infrastructure of the entire system to ensure the stable operation of the system.

[0066] The digital twin model building module includes: information acquisition unit, 3D modeling unit, and information annotation unit;

[0067] An information acquisition unit is used to acquire information on transfer equipment parts. For regular transfer equipment parts, the information on transfer equipment parts is acquired through drawings or actual surveying and mapping dimensions. For special-shaped transfer equipment parts, three-dimensional point cloud data of special-shaped or non-standard transfer equipment parts is collected using three-dimensional laser scanning technology.

[0068] A three-dimensional modeling unit, used to quickly generate a three-dimensional model of the transfer equipment using a modeling tool;

[0069] The information marking unit is used to mark the engineering information of the transfer equipment parts on the three-dimensional model to form a full three-dimensional display mode.

[0070] The preset marking rules in the information marking unit include: dimension marking rules, material marking rules, and precision marking rules for transfer equipment parts.

[0071] Furthermore, the marking rules specifically include:

[0072] (1) Dimensioning rules:

[0073] Basic dimensioning: marking the basic dimensions of parts, such as length, width, and height;

[0074] Tolerance marking: The dimensional tolerance of a component indicates the allowable dimensional deviation range. The "±" symbol can be used to indicate the tolerance. For example, "50±0.5mm" means the dimension is 50mm and the allowable deviation is ±0.5mm.

[0075] (2) Material marking rules:

[0076] Material name and code: mark the material name and corresponding material code of the component, such as the code in the standard material library;

[0077] Material property annotation: Annotate the material properties of parts, such as density, hardness, and tensile strength. These properties are usually obtained through experimental measurements and can be expressed with specific values. For example, "density = 7.85g / cm³";

[0078] (3) Accuracy marking rules:

[0079] Precision grade marking: marking the machining precision grade of parts, such as IT6, IT7;

[0080] Surface roughness marking: Mark the surface roughness level of parts, such as Ra0.8, Rz3.2, where the surface roughness can be obtained through experimental measurement and can be expressed by a specific value, such as "Ra=0.8μm".

[0081] The maintenance optimization module includes: process definition unit, process execution unit, and process optimization unit;

[0082] The process definition unit is used to customize the maintenance process according to the actual situation and needs. The process definition unit involves the decomposition, sorting and combination of maintenance tasks, thereby indirectly realizing the serialization of maintenance tasks;

[0083] The process execution unit is used to operate according to the steps and time nodes of the process, monitor key nodes and abnormal situations, and use the improved energy valley optimization algorithm to evaluate the advantages and disadvantages of different maintenance task scheduling schemes. At the same time, when executing the maintenance process, the maintenance tasks will be executed according to the order and rules defined in the process definition unit, thereby ensuring the serialized execution of the maintenance tasks;

[0084] The process optimization unit is used to adjust and optimize the maintenance task sequence using the ESIA algorithm according to the execution status and feedback of the maintenance tasks to improve the maintenance efficiency and quality.

[0085] Specifically, the ESIA algorithm process includes:

[0086] Eliminate: Identify and remove redundant steps in the process;

[0087] Simplify: simplify the remaining steps and improve execution efficiency;

[0088] Integration: Integrate scattered steps to ensure process consistency and coordination;

[0089] Automation: Introduce automated tools or systems to reduce human intervention and improve the level of process automation.

[0090] The disassembly and assembly visualization module includes: a three-dimensional visualization unit, a data display unit, and an interactive control unit;

[0091] 3D visualization unit, used to present the digital twin model and maintenance process in 3D, providing an intuitive operation interface;

[0092] The data display unit is used to display various parameters and status information of the transfer equipment in real time, so that users can understand the operation status of the equipment;

[0093] The interactive control unit is used to provide interactive control functions between users and digital twin models, such as zooming, rotating, and panning, so that users can observe and analyze the equipment from different angles.

[0094] The system architecture modules include: network and communication unit, storage and backup unit, and security system unit;

[0095] Network and communication unit, used to provide network communication support for the system;

[0096] Storage and backup unit, used for secure storage and backup of data;

[0097] The security system unit is used to ensure the security of the system and prevent data leakage and illegal access.

[0098] The digital twin model building module, maintenance optimization module, disassembly and assembly visualization module, system architecture module and its corresponding units work together to form a complete digital twin-enhanced turnaround maintenance process optimization and visualization system, which can significantly improve maintenance efficiency, reduce maintenance costs, and enhance equipment reliability.

[0099] In summary, the overall implementation process of the digital twin-enhanced turnaround maintenance process optimization and visualization system includes: (1) obtaining the size information and 3D point cloud data of the turnaround equipment parts, using modeling tools to quickly generate a 3D model of the turnaround equipment based on the input information, and marking the size, material, precision and other engineering information of the turnaround equipment parts on the 3D model to obtain a fully labeled digital twin model; (2) according to the actual situation and needs, formulate a maintenance process based on the fully labeled digital twin model, and perform actual operations according to the steps and time nodes of the customized maintenance process, while monitoring key nodes and abnormal conditions, and generating maintenance records and abnormality reports; (3) according to the production (4) The complete maintenance record form and abnormal report are used to summarize and evaluate the maintenance process, put forward optimization suggestions, and obtain the optimized maintenance process; (5) The digital twin model with complete annotations and the optimized maintenance process are presented in three dimensions, and an intuitive operation interface is provided; (6) The real-time operation data of the transfer equipment is obtained, and various parameters and status information of the transfer equipment are displayed in real time. The user is provided with interactive control functions such as zooming, rotating, and translating between the user and the digital twin model to obtain a three-dimensional model view after user interaction; (7) At the same time, all data generated by the system, including maintenance records, real-time data, and user interaction data, are securely stored and backed up to ensure data integrity and recoverability.

[0100] The advantages of the digital maintenance method in the present invention include: improving maintenance efficiency and accuracy during the maintenance of the fan; providing detailed equipment performance data to help managers better plan maintenance work and reasonably arrange human resources; providing more comprehensive and accurate equipment status information to help managers promptly discover safety hazards, take preventive measures, and reduce the risk of accidents; collecting and analyzing a large amount of equipment operation data to provide decision support for management and help them formulate more scientific and reasonable maintenance strategies and plans; reducing unnecessary repairs and replacement of parts, and reducing maintenance costs through accurate fault diagnosis and predictive maintenance; promoting the cultivation of talents and improving the core competitiveness of the enterprise.

[0101] Example 2

[0102] See also Figure 5 , in this embodiment:

[0103] The specific steps of the 3D reconstruction strategy include:

[0104] A1: Use a laser scanner or camera to collect images, videos or laser scanned image data of the transfer equipment, and pre-process the collected image data;

[0105] Among them, preprocessing includes denoising, grayscale, image enhancement and other operations, which help to improve the accuracy of subsequent feature extraction and contour extraction.

[0106] A2: Extract features from the preprocessed image data, and use a corner point detection method to extract feature points of the transfer equipment to obtain feature point data of the transfer equipment, wherein a contour tracking algorithm is used to extract features from the preprocessed image data, and the contour tracking algorithm is a prior art in the art, and is not an inventive solution of the present application, and is not described in detail herein;

[0107] A3: Based on the feature point data of the transfer equipment, an edge detection method is used to extract the contour image information of the transfer equipment, and a transfer equipment contour image data set is generated. The edge detection method is a prior art in the art and is not an inventive solution of the present application, and is not described in detail here.

[0108] A4: Based on the generated transfer equipment contour image dataset, use a binocular camera to capture the transfer equipment contour images from different perspectives, and pre-process the captured transfer equipment contour images to obtain the transfer equipment visual images;

[0109] A5: Calculate the disparity map between the images taken by the left and right lenses of the same scene in the visual image of the transfer device, and convert the pixels in the image into point cloud data in three-dimensional space based on the camera's intrinsic parameter matrix and disparity map;

[0110] A6: Use a registration algorithm (such as the ICP algorithm) to align the point cloud data from different perspectives, and fuse the registered point cloud data to construct a three-dimensional model of the transfer equipment.

[0111] Furthermore, the specific steps of A6 include:

[0112] A6.1: Select a reference point cloud and a target point cloud to be registered, and initialize an initial transformation matrix, which can be an identity matrix;

[0113] A6.2: In the matching step, the nearest neighbor point of each point in the target point cloud in the reference point cloud is found by calculating the Euclidean distance, wherein the Euclidean distance calculation formula is the prior art content in this field, is not the inventive solution of this application, and is not described in detail here;

[0114] A6.3: Based on the matched point pairs, use the least squares method to calculate an optimal rotation matrix and translation vector so that the target point cloud is as close as possible to the reference point cloud. The least squares method is a prior art in this field and is not an inventive solution of the present application, so it will not be described in detail here.

[0115] A6.4: Apply the calculated transformation matrix to the target point cloud to obtain a new transformed target point cloud;

[0116] A6.5: Repeat the steps of matching, calculating the transformation matrix and updating until the maximum number of iterations is reached;

[0117] A6.6: Fuse the registered point cloud data to construct a complete 3D model of the transfer equipment.

[0118] The specific steps of using the corner point detection method to extract the feature points of the transfer equipment in A2 include:

[0119] A2.1: Perform grayscale transformation on the preprocessed image data to generate a grayscale image of the transfer equipment , where R represents the range, A and B represent the row and column values ​​of the grayscale image of the transfer device, and , M represents the number of pixels in the grayscale image of the transfer device;

[0120] A2.2: Calculate the gradient of the grayscale image of the transfer device in the x and y directions at each pixel (p, q) and ,in, Represents the gradient of the grayscale image of the transfer device in the x direction at the pixel point (p, q), represents the Sobel operator function in the x direction, Represents the gradient of the grayscale image of the transfer device in the y direction at the pixel point (p, q), Represents the Sobel operator function in the y direction;

[0121] A2.3: For each pixel (p, q), calculate an autocorrelation matrix M cor = [ ∇ 2 x ∇ x ∇ y ∇ x ∇ y ∇ 2 y ] ;

[0122] A2.4: Based on the autocorrelation matrix , using the singular value decomposition algorithm Calculate eigenvalues and , where U and V represent orthogonal matrices, represents a diagonal matrix, and the diagonal elements are eigenvalues;

[0123] A2.5: Based on eigenvalues and , calculate the corner point responsivity ,in, represents the minimum function;

[0124] A2.6: Corner responsiveness for all pixels Sort in descending order and set a responsiveness threshold ;

[0125] like , then the pixel point is selected as the corner point, and the coordinates and responsiveness of the selected corner point are output as feature point data.

[0126] The specific steps of A5 include:

[0127] A5.1: Receive visual images from the transfer device, perform camera calibration, and perform epipolar correction on the images taken by the left and right lenses so that the corresponding epipolar lines are parallel to each other;

[0128] Furthermore, in the camera calibration process, firstly, single-target calibration is performed to determine the camera's intrinsic parameter matrix, including focal length, principal point coordinates, and distortion coefficients, and then dual-target calibration is performed to determine the extrinsic parameters between the two cameras, including the rotation matrix and translation vector.

[0129] A5.2: For each pixel in the left eye view, calculate the matching cost between it and the matching pixel in the right eye view, that is, the matching degree, and extend the matching cost of a single pixel to its neighborhood through a 5×5 sliding window to obtain the aggregated matching cost;

[0130] Among them, on the one hand, when calculating the matching cost or the degree of matching, the judgment is made by calculating the similarity between the two pixels, and the similarity calculation is the prior art content in this field, which is not the creative solution of the present application and will not be elaborated here; on the other hand, when expanding the matching cost, the concept of tensor is adopted, and the pixel point is expanded in the neighborhood to replace the original single pixel.

[0131] A5.3: Use the aggregated matching cost to calculate the optimal disparity value for each pixel, and refine and smooth the disparity value to obtain the final disparity map;

[0132] A5.4: Calculate the depth value of each pixel using the triangulation principle based on the camera's intrinsic parameter matrix and disparity map. The triangulation principle is a prior art in the art and is not an inventive solution of the present application, so it will not be described in detail here.

[0133] A5.5: Combine the depth value, camera parameters, and pixel coordinates to convert them into point cloud data in three-dimensional space.

[0134] For example, assume that the pixel coordinates in the image are (p, q), the depth is z, and the camera intrinsic matrix is ,in, and Represent the components of the focal length in the x and y directions, respectively. and Represents the coordinates of the principal point, then the coordinates of the pixel in three-dimensional space It can be expressed as: [ Y 1 Y 2 Y 3 ] = z × M nei − 1 × [ p q 1 ] , after expansion, we get: , , .

[0135] The specific steps of the benchmarking strategy include:

[0136] B1: Clarify the target of benchmarking, that is, what kind of maintenance process optimization effect is expected to be achieved through benchmarking, and generate a maintenance process optimization target specification;

[0137] B2: According to the maintenance process optimization target specification, select benchmark objects through research, that is, select enterprises or departments with excellent maintenance processes in the industry or field as benchmark objects, and obtain the maintenance process information of the benchmark objects;

[0138] B3: Compare and analyze your own maintenance process with that of the benchmark object to obtain gap information, and organize the analysis results to form a maintenance process gap analysis report;

[0139] B4: According to the maintenance process gap analysis report, formulate targeted improvement plans, and implement the optimization of maintenance processes step by step according to the requirements of the improvement plans, and form a maintenance process improvement implementation plan and execution report;

[0140] Among them, the improvement plans include process optimization, personnel training, equipment improvement, enhanced management, and resource guarantee.

[0141] B5: According to the maintenance process improvement implementation plan and execution report, evaluate the improved maintenance process, and form a maintenance process optimization evaluation report based on the evaluation results. At the same time, provide feedback and adjustments based on the evaluation report.

[0142] The specific steps of the improved energy valley optimization algorithm include:

[0143] C1: Define the objective function of maintenance task scheduling and initialize the particle swarm, where each particle represents a maintenance task scheduling scheme;

[0144] Among them, when defining the objective function of maintenance task scheduling, it is necessary to clarify the optimization goal of maintenance task scheduling, such as minimizing maintenance time, maximizing equipment availability, etc., and define the objective function accordingly.

[0145] C2: Set the maximum number of iterations r and the number of particles n of the improved energy valley optimization algorithm, and calculate the objective function value for each maintenance task scheduling scheme;

[0146] C3: Determine the neutron enrichment level of each maintenance task scheduling plan based on the objective function value ,This neutron enrichment level reflects the stability or quality of the maintenance task scheduling plan;

[0147] in, N E L i = F ( g i ) , i ∈ [ 1 , n ] , represents the neutron enrichment level of the i-th particle, F represents the particle function, Represents the characteristics of the i-th particle.

[0148] C4: Setting the neutron enrichment level threshold ,and , where BS represents the NEL value of the maintenance task scheduling scheme with the best stability level, and WS represents the NEL value of the maintenance task scheduling scheme with the worst stability level;

[0149] like , then decay may occur, and according to the decay type, the maintenance task scheduling plan is updated;

[0150] like , then the maintenance task scheduling plan may have deviations, that is, fine-tune the maintenance task scheduling plan;

[0151] The decay types include , or decay;

[0152] If If it decays, the N and Z values ​​are reduced, simulating the reduction of the allocation of certain tasks or resources, where N represents the number of neutrons and Z represents the number of protons;

[0153] If Decay, then increase the Z value and decrease the N value, simulating increasing the priority of certain tasks or resources;

[0154] If If the particle decays, the N or Z value will not be changed, but the position of the particle in the solution space will be adjusted to simulate the fine-tuning of the task scheduling scheme.

[0155] C5: Repeat the above steps until the maximum number of iterations is reached, and in each iteration, update the optimal maintenance task scheduling plan according to the update of the maintenance task scheduling plan;

[0156] C6: Output the optimal maintenance task scheduling plan.

[0157] The embodiments of the present invention are described above in conjunction with the accompanying drawings, but the present invention is not limited to the above-mentioned specific implementation modes, which are merely illustrative rather than restrictive. Under the guidance of the present invention, ordinary technicians in the field may also change, modify, replace and modify the above-mentioned embodiments without departing from the purpose and scope of protection of the present invention, and all of these are within the protection of the present invention.

Claims

1. The digital twin-enhanced turnaround maintenance process optimization and visualization system is characterized by: include: Digital twin model building module and maintenance optimization module; The digital twin model building module includes an information acquisition unit, a three-dimensional modeling unit, and an information annotation unit; the information acquisition unit is used to acquire the information of the parts of the transfer equipment; the three-dimensional modeling unit is configured with a three-dimensional reconstruction strategy, and the three-dimensional reconstruction strategy is used to quickly generate a three-dimensional model of the transfer equipment; the information annotation unit is used to annotate the engineering information on the three-dimensional model according to the preset annotation rules to form a full three-dimensional display mode; The maintenance optimization module includes a process definition unit, a process execution unit, and a process optimization unit; the process definition unit is configured with a benchmark targeting strategy, which is used to customize the maintenance process according to actual conditions and needs, and complete the decomposition, sorting and combination of maintenance tasks; the process execution unit is configured with an improved energy valley optimization algorithm, which is used to evaluate the advantages and disadvantages of different maintenance task scheduling schemes; the process optimization unit is configured with an ESIA algorithm, which includes elimination, simplification, integration and automation, and is used to evaluate and optimize the maintenance process; The specific steps of the benchmarking strategy include: B1: Clarify the target of the benchmark and generate a target specification for maintenance process optimization; B2: According to the maintenance process optimization target specification, select benchmark objects through research and obtain the maintenance process information of benchmark objects; B3: Compare and analyze your own maintenance process with that of the benchmark object to obtain gap information, and organize the analysis results to form a maintenance process gap analysis report; B4: According to the maintenance process gap analysis report, formulate targeted improvement plans, and implement the optimization of maintenance processes step by step according to the requirements of the improvement plans, and form a maintenance process improvement implementation plan and execution report; B5: Evaluate the improved maintenance process according to the maintenance process improvement implementation plan and execution report, and form a maintenance process optimization evaluation report based on the evaluation results. At the same time, provide feedback and make adjustments based on the evaluation report; The specific steps of the improved energy valley optimization algorithm include: C1: Define the objective function of maintenance task scheduling and initialize the particle swarm, where each particle represents a maintenance task scheduling scheme; C2: Set the maximum number of iterations r and the number of particles n of the improved energy valley optimization algorithm, and calculate the objective function value for each maintenance task scheduling scheme; C3: Determine the neutron enrichment level of each maintenance task scheduling plan based on the objective function value ; The specific steps of the improved energy valley optimization algorithm also include: C4: Setting the neutron enrichment level threshold ; like , then decay may occur, and according to the decay type, the maintenance task scheduling plan is updated; like , then there is a deviation in the maintenance task scheduling plan, that is, fine-tune the maintenance task scheduling plan; C5: Repeat the above steps until the maximum number of iterations is reached, and in each iteration, update the optimal maintenance task scheduling plan according to the update of the maintenance task scheduling plan; C6: Output the optimal maintenance task scheduling plan.

2. The digital twin enhanced turnaround maintenance process optimization and visualization system according to claim 1, characterized in that: The specific steps of the three-dimensional reconstruction strategy include: A1: Use a laser scanner or camera to collect images, videos or laser scanned image data of the transfer equipment, and pre-process the collected image data; A2: Extract features from the preprocessed image data and use the corner point detection method to extract feature points of the transfer equipment to obtain the transfer equipment feature point data; A3: Based on the feature point data of the transfer equipment, the edge detection method is used to extract the contour image information of the transfer equipment and generate the transfer equipment contour image dataset; A4: Based on the generated transfer equipment contour image dataset, use a binocular camera to capture the transfer equipment contour images from different perspectives, and pre-process the captured transfer equipment contour images to obtain the transfer equipment visual images; A5: Calculate the disparity map between the images taken by the left and right lenses of the same scene in the visual image of the transfer device, and convert the pixels in the image into point cloud data in three-dimensional space based on the camera's intrinsic parameter matrix and disparity map; A6: Align the point cloud data from different perspectives, and fuse the registered point cloud data to construct a three-dimensional model of the transfer equipment.

3. The digital twin enhanced turnaround maintenance process optimization and visualization system according to claim 2, characterized in that: The specific steps of using the corner point detection method to extract the feature points of the transfer equipment in A2 include: A2.1: Perform grayscale transformation on the preprocessed image data to generate a grayscale image of the transfer equipment , where R represents the range, A and B represent the row and column values ​​of the grayscale image of the transfer device, and , M represents the number of pixels in the grayscale image of the transfer device; A2.2: Calculate the gradient of the grayscale image of the transfer device in the x and y directions at each pixel (p, q) and ,in, Represents the gradient of the grayscale image of the transfer device in the x direction at the pixel point (p, q), represents the Sobel operator function in the x direction, Represents the gradient of the grayscale image of the transfer device in the y direction at the pixel point (p, q), Represents the Sobel operator function in the y direction; A2.3: For each pixel (p, q), calculate an autocorrelation matrix ; A2.4: Based on the autocorrelation matrix , using the singular value decomposition algorithm Calculate eigenvalues and , where U and V represent orthogonal matrices, represents a diagonal matrix, and the diagonal elements are eigenvalues; A2.5: Based on eigenvalues and , calculate the corner point responsivity ,in, represents the minimum function; A2.6: Corner responsiveness for all pixels Sort in descending order and set a responsiveness threshold ; like , then the pixel point is selected as the corner point, and the coordinates and responsiveness of the selected corner point are output as feature point data.

4. The digital twin enhanced turnaround maintenance process optimization and visualization system according to claim 3, characterized in that: The specific steps of A5 include: A5.1: Receive visual images from the transfer device, perform camera calibration, and perform epipolar correction on the images taken by the left and right lenses so that the corresponding epipolar lines are parallel to each other; A5.2: For each pixel in the left eye view, calculate the matching cost between it and the matching pixel in the right eye view, that is, the matching degree, and extend the matching cost of a single pixel to its neighborhood through a 5×5 sliding window to obtain the aggregated matching cost; A5.3: Use the aggregated matching cost to calculate the optimal disparity value for each pixel, and refine and smooth the disparity value to obtain the final disparity map; A5.4: Calculate the depth value of each pixel using the triangulation principle based on the camera's intrinsic parameter matrix and disparity map; A5.5: Combine the depth value, camera parameters, and pixel coordinates to convert them into point cloud data in three-dimensional space.

5. The digital twin enhanced turnaround maintenance process optimization and visualization system according to claim 4, characterized in that: In the information acquisition unit, for regular transfer equipment parts, the transfer equipment parts information is acquired through drawings or actual surveying and mapping dimensions, and for special-shaped transfer equipment parts, a three-dimensional laser scanning method is used to collect three-dimensional point cloud data of special-shaped or non-standard transfer equipment parts.

6. The digital twin enhanced turnaround maintenance process optimization and visualization system according to claim 5, characterized in that: The preset marking rules in the information marking unit include: dimension marking rules, material marking rules, and precision marking rules for parts of transfer equipment.

Images