Equipment monitoring visualization system based on digital twinning technology

Abstract

The application relates to the technical field of visual monitoring, in particular to an equipment monitoring visual system based on digital twin technology, which comprises a data acquisition module, a virtual model creation module, a virtual-real mapping module and an anomaly detection module. Through the cooperative work of the modules, real-time monitoring, fault prediction, performance optimization and remote operation guidance of equipment such as motors are realized, and the operation and maintenance efficiency and reliability of the equipment are significantly improved. The application further describes the analysis logic of the temperature detection unit, the vibration detection unit and the rotating speed detection unit of the protection self-checking submodule, and the working principle of the power failure detection submodule. The application of these technologies enables the system to accurately identify and warn various potential faults, thereby providing a powerful guarantee for the stable operation of the equipment.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of visual monitoring, and particularly relates to an equipment monitoring visual system based on digital twin technology. BACKGROUND

[0002] With the continuous development of industrial automation and intelligence, as an indispensable power source in industrial production, the stability and efficiency of the motor have a direct impact on the continuity and economic benefits of the entire production system. Traditional motor monitoring systems rely on scattered sensors and control units. Although these systems can provide basic monitoring data, there are obvious deficiencies in data integration, real-time performance and fault prediction. In existing technical solutions, the monitoring of the motor usually focuses on the threshold monitoring of specific parameters such as temperature, vibration or speed. When these parameters exceed the preset threshold, the system will issue an alarm. However, this reactive monitoring method cannot provide early warning of potential faults, and due to the scattered monitoring data, it is difficult to achieve comprehensive analysis and optimization. In addition, for the maintenance and fault diagnosis of the motor, field workers often need to rely on experience to make judgments and treatments, which not only requires high skills of personnel, but also cannot guarantee the timeliness and accuracy of the treatment measures. In a complex industrial environment, motor failure can cause production lines to stall, causing huge economic losses and safety hazards. The emergence of digital twin technology provides a new way to solve the above problems. By creating a virtual model of the physical device and realizing seamless connection of virtual and real data, digital twin technology can reflect the running state of the physical device in real time, realize early detection and warning of faults.

[0003] For example, the Chinese patent with the authorized announcement number CN105205295B relates to a modeling method of a traveling wave type rotary ultrasonic motor, which is a modeling method of a TRUM "electric-mechanical-force" hybrid model, including the following steps: first step: establishing a stator equivalent circuit model of the "electric energy-high frequency vibration mechanical energy" process; second step: establishing a stator-rotor discrete contact model of the "high frequency vibration mechanical energy-rotational kinetic energy" process; third step, determining the interface parameters to obtain the TRUM "electric-mechanical-force" hybrid model. This method realizes the deconstruction of the motor mechanism according to the electromechanical energy conversion relationship, models each part by different methods, and finally connects each part model through reasonable interface parameters, thereby establishing a motor hybrid model that can reflect the characteristics of the motor and is easy to analyze, overcoming the defects of the existing TRUM modeling method that cannot describe the tooth structure of the stator, the calculation difficulty and complexity are increased, and it is not practical.

[0004] The above patents all have the problems proposed in the background art: the motor has a complex internal structure and geometry, it is difficult to accurately model the ultrasonic motor internally through circuit simulation analysis only, although the digital twin technology can realize the visualization of faults, it lacks in-depth fault prediction and diagnosis capability, and cannot provide strong support for maintenance decisions, and there is a delay in the synchronous update of the virtual model and the physical device, affecting the real-time and accuracy of monitoring. SUMMARY

[0005] The technical problem to be solved by the present application is to solve the problems of the prior art, and a equipment monitoring visualization system based on digital twin technology is provided, which comprises a data acquisition module, a virtual model creation module, a virtual-real mapping module and an anomaly detection module. Through the cooperative work of these modules, real-time monitoring, fault prediction, performance optimization and remote operation guidance of the motor and other equipment are realized, which significantly improves the operation and maintenance efficiency and reliability of the equipment, and the analysis logic of the temperature detection unit, the vibration detection unit and the rotating speed detection unit of the protection self-checking sub-module, and the working principle of the power failure detection sub-module are described in detail. The application of these technologies enables the system to accurately identify and warn various potential faults, providing a strong guarantee for the stable operation of the equipment.

[0006] To achieve the above object, the present application provides the following technical scheme:

[0007] The equipment monitoring visualization system based on digital twin technology comprises a data acquisition module, a virtual model creation module, a virtual-real mapping module and an anomaly detection module.

[0008] The data acquisition module is used to acquire appearance data and running data of the motor, wherein the appearance data comprises motor external images and motor internal point cloud data.

[0009] The virtual model creation module is configured with a motor modeling strategy, which is used to pre-process the appearance data and analyze the pre-processed appearance data to output a motor three-dimensional model.

[0010] The virtual-real mapping module is used to construct a motor logic model, map the motor logic model to an information space, and construct a digital twin motor in combination with the motor three-dimensional model.

[0011] The anomaly detection module is used to investigate suspicious fault motors, locate the number and position of the fault motor, and alarm the fault condition during the motor visualization monitoring process.

[0012] The motor modeling strategy includes external feature extraction logic, internal feature extraction logic, and 3D modeling logic. The virtual model creation module includes an image processing unit, a point cloud processing unit, and a model modeling unit. The external feature extraction logic is configured within the image processing unit, the internal feature extraction logic is configured within the point cloud processing unit, and the 3D modeling logic is configured within the model modeling unit. The image processing unit processes the external image of the motor to extract external structural features. The point cloud processing unit processes the internal point cloud data of the motor to extract internal point cloud features. The model modeling unit processes the external structural features and internal point cloud features to output a 3D model of the motor.

[0013] The specific steps of the external feature extraction logic are as follows:

[0014] S3.1: Denoise the external image of the motor;

[0015] S3.2: Perform histogram statistics on the grayscale values ​​of the denoised external image of the motor, and use statistical learning methods to cut out the target region of the pixels in the external image of the motor;

[0016] S3.3: Smooth the target region image through edge detection, calculate the Gaussian difference based on the smoothed target region image, extract the key points of the target region image, calculate the medium intensity of the key points, divide the target region image into four components R, G, B and Deep for pixel decomposition, and calculate the medium intensity of the decomposed key points.

[0017] S3.4: Based on the medium intensity of the keypoints and the medium intensity of the four components of the decomposed keypoints, perform medium intensity matching on the image and calculate the external structural features. The formula for calculating the external structural features is as follows:

[0018]

[0019] Among them, O c Let j represent a single keypoint, N represent the total number of keypoints, δ represent the Fourier transform factor, and x represent the external structural features. j The x-coordinate of the j-th keypoint is represented by the pixel x-coordinate, y j Let represent the pixel ordinate of the j-th keypoint, max[·] denotes finding the maximum value of the function, M(·) represents the medium intensity matching score matrix, and m Rk The R component represents the medium strength at key points, m Gk The key point G component represents the medium strength, m Bk The medium strength of component B at key point is represented by m. Dk The deep component represents the medium strength at key points, m. kMedium intensity representing key points.

[0020] The internal feature extraction logic specifically includes the following steps:

[0021] S4.1: Preprocess the motor internal point cloud data according to the motor internal point cloud data, and perform coordinate conversion on the preprocessed motor internal point cloud data through rigid body transformation;

[0022] S4.2: Divide the grid at equal intervals, voxelize the motor internal point cloud data, project the voxel index into the corresponding grid, connect the points in the voxel, strengthen the local feature information, and obtain the point cloud voxel feature;

[0023] S4.3: Obtain motor internal structure standard point cloud data, classify the point cloud voxel feature through a clustering algorithm, merge the grid where the point cloud voxel feature under the same center cluster is located, and calculate the internal point cloud feature.

[0024] The three-dimensional modeling logic includes a feature classification sub-logic and a motor modeling sub-logic, the feature classification sub-logic includes constructing a feature classification network, training the feature classification network through the external structure feature and the internal point cloud feature, and outputting the motor component type corresponding to the feature, and the motor modeling sub-logic includes setting rendering parameters of the motor component corresponding to the motor component type through modeling software, rendering the external structure feature and the internal point cloud feature corresponding to the motor component type, and outputting a three-dimensional model of the ultrasonic motor;

[0025] The feature classification network includes an input layer, a hidden layer and an output layer;

[0026] The input layer is used to input the external structure feature and the internal point cloud feature, perform logarithmic operation and size reconstruction on the external structure feature and the internal point cloud feature, and obtain the input parameters of the hidden layer;

[0027] The hidden layer is used to train the input parameters of the hidden layer, and calculate the motor component type parameters through an attention mechanism;

[0028] The output layer is used to output the classification result according to the motor component type parameters.

[0029] The virtual-real mapping module includes a data integration unit, a physical mapping unit and a visualization display unit, the data integration unit is used to receive input from sensors and control systems, convert the input into a unified format, and display it in the motor instrument panel, the physical mapping unit is used to convert the physical properties of the motor into a digital model, and map the digital model to the information space, and the visualization display unit is used to provide an interactive page, and construct a digital twin motor through the motor three-dimensional model.

[0030] The abnormality detection module comprises a protection self-checking submodule, a power failure detection submodule and a fault warning submodule.

[0031] The protection self-checking submodule is used for detecting fault problems in the operation of the motor.

[0032] The power failure detection submodule is used for detecting fault problems that cannot be found in the operation of the motor.

[0033] The fault warning submodule is used for warning the faults detected by the protection self-checking submodule and the power failure detection submodule, and sending a notification to the staff.

[0034] The protection self-checking submodule comprises a temperature detection unit, a vibration detection unit and a rotating speed detection unit.

[0035] The temperature detection unit comprises temperature analysis logic, which is used for monitoring temperature data of the motor according to a temperature sensor arranged on the motor, cleaning and preprocessing the collected temperature data, drawing a frequency distribution histogram of the temperature data, and outputting a temperature state analysis result of the motor according to the frequency distribution histogram, wherein the temperature state analysis result comprises motor overheating and normal motor temperature.

[0036] The vibration detection unit comprises vibration signal analysis logic, which comprises vibration signal preprocessing, vibration signal feature extraction and vibration signal feature analysis, and outputs a vibration signal analysis result, wherein the vibration signal analysis result comprises motor imbalance, motor bearing damage and shaft bending.

[0037] The rotating speed detection unit comprises rotating speed analysis logic, which is used for monitoring rotating speed data of the motor according to a rotating speed sensor arranged on the motor, cleaning and preprocessing the collected rotating speed data, drawing a rotating speed change trend chart, judging whether a long-term drift phenomenon exists in the speed fluctuation, comparing the actual rotating speed with a predetermined working rotating speed interval, judging whether the motor is in an optimal working state, and outputting a rotating speed analysis result, wherein the rotating speed analysis result comprises motor overspeed, motor underspeed and motor unstable rotating speed.

[0038] Compared with the prior art, the present application has the following beneficial effects:

[0039] 1. The three-dimensional modeling technology of the present application visualizes the motor, realizes real-time monitoring of various parameters of the motor through an instrument panel, enhances the intuitive understanding of the operation of the equipment by the operation and maintenance personnel, optimizes resource allocation, and improves the utilization rate of the equipment.

[0040] 2.The system can ensure real-time synchronization update of the virtual model and the physical device through the virtual-real mapping module, and improve the real-time performance and accuracy of monitoring, which helps to discover abnormal conditions of the device in time and reduce the risk of failure;

[0041] 3.The system can discover potential fault risks through deep analysis of the operation data and issue early warning in time, which provides strong support for maintenance decision, reduces the equipment downtime and maintenance cost, and improves the production efficiency. BRIEF DESCRIPTION OF DRAWINGS

[0042] Other features, objects and advantages of the present application will become more apparent from the following detailed description of non-limiting embodiments thereof, read in conjunction with the accompanying drawings:

[0043] Figure 1 A module diagram of the equipment monitoring visualization system based on the digital twin technology of the embodiment 1 of the present application;

[0044] Figure 2 A key point medium intensity calculation schematic diagram of the embodiment 1 of the present application;

[0045] Figure 3 A point cloud voxel feature clustering flowchart of the embodiment 1 of the present application;

[0046] Figure 4 A grid clustering result diagram of the embodiment 1 of the present application;

[0047] Figure 5 A vibration signal analysis flowchart of the embodiment 1 of the present application. DETAILED DESCRIPTION

[0048] The technical solutions in the embodiments of the present application will be described clearly and completely below in conjunction with the drawings in the embodiments of the present application, and obviously, the described embodiments are only a part of the embodiments of the present application, but not all the embodiments.

[0049] Embodiment 1:

[0050] Please refer to Figure 1 The present application provides an embodiment: an equipment monitoring visualization system based on the digital twin technology, which comprises a data acquisition module, a virtual model creation module, a virtual-real mapping module and an anomaly detection module.

[0051] The data acquisition module is used to acquire appearance data and operation data of the motor, wherein the appearance data comprises motor external images and motor internal point cloud data.

[0052] The virtual model creation module is internally configured with a motor modeling strategy, which is used to pre-process the appearance data and analyze the pre-processed appearance data to output a motor three-dimensional model;

[0053] The motor modeling strategy includes external feature extraction logic, internal feature extraction logic and three-dimensional modeling logic, the virtual model creation module includes an image processing unit, a point cloud processing unit and a model modeling unit, the external feature extraction logic is configured in the image processing unit, the internal feature extraction logic is configured in the point cloud processing unit, and the three-dimensional modeling logic is configured in the model modeling unit, the image processing unit is used to process the motor external image according to the motor external image, extract the external structure feature, the point cloud processing unit is used to process the motor internal point cloud data according to the motor internal point cloud data, and extract the internal point cloud feature, and the model modeling unit is used to process the external structure feature and the internal point cloud feature to output the three-dimensional model of the motor.

[0054] The external feature extraction logic specifically includes the following steps:

[0055] S3.1: denoising processing is performed on the motor external image, wherein the motor external image noise includes AD quantization noise, dark current noise and part reflection noise;

[0056] S3.2: histogram statistics is performed on the gray value of the denoised motor external image, the peak-valley gray value between the two peak values is taken as a segmentation threshold one, a segmentation threshold two is calculated according to the probability density function of the motor external image, the optimal segmentation threshold is obtained by weighted summation of the segmentation threshold one and the segmentation threshold two, the motor external image pixel is compared with the optimal segmentation threshold, if greater than or equal to the optimal segmentation threshold, the pixel is marked as a target region, and if less than the optimal segmentation threshold, the pixel is marked as a background region;

[0057] S3.3: the pixel points marked as the target region are connected, a target region image is obtained, and the target region image is smoothed through edge detection, the edge detection includes image Gaussian filtering processing, difference method gradient amplitude and direction calculation, non-maximum suppression of gradient amplitude and double-threshold edge connection;

[0058] Please refer to Figure 2 , the key point medium intensity calculation schematic diagram of the embodiment of the application, S3.4: the Gaussian difference is calculated according to the smoothed target region image, the key points of the target region image are extracted, the medium intensity of the key points is calculated, the target region image is divided into R, G, B and Deep four components for pixel element decomposition, the key point medium intensity after decomposition is calculated, and the medium intensity calculation formula of the key points is:

[0059]

[0060] wherein m k denotes the medium intensity of the key point at the pixel position k, m denotes the total number of pixel positions, n denotes the number of spectral bands, arg min (·) denotes the minimum value of a function, μ k denotes the average intensity value of the pixel position k, x ik denotes the pixel intensity value of the pixel position k on the spectral band;

[0061] S3.5: Obtain the fused medium intensity according to the medium intensity of the key point and the medium intensity of the four components of the decomposed key point, perform medium intensity matching on the image to obtain a medium intensity matching score matrix, and according to the Sinkhorn algorithm, first perform normalization row by row, that is, divide each element in the first row by the sum of each element in the first row to obtain a new row, perform the same operation on each row, then perform normalization column by column, finally obtain the maximum value of the distribution matrix, compare the maximum value, obtain the optimal distribution result, and calculate the external structure feature according to the score matrix, and the calculation formula of the external structure feature is:

[0062]

[0063] wherein O c denotes the external structure feature, j denotes a single key point, N denotes the total number of key points, δ denotes a Fourier transform factor, x j denotes the horizontal pixel coordinate of the jth key point, y j denotes the vertical pixel coordinate of the jth key point, max [·] denotes the maximum value of a function, M (·) denotes the medium intensity matching score matrix, m Rk denotes the medium intensity of the R component of the key point, m Gk denotes the medium intensity of the G component of the key point, m Bk denotes the medium intensity of the B component of the key point, m Dk denotes the medium intensity of the Deep component of the key point, m k denotes the medium intensity of the key point.

[0064] The internal feature extraction logic specifically comprises the following steps:

[0065] S4.1: Preprocess the motor internal point cloud data according to the motor internal point cloud data;

[0066] S4.2: Find the most suitable three-dimensional plane through plane fitting, and perform coordinate conversion on the preprocessed motor internal point cloud data through rigid body transformation to convert the motor internal point cloud data from the radar coordinate system to the world coordinate system;

[0067] S4.3: equally dividing the world coordinate system into grids, voxelizing the point cloud data inside the motor, projecting the voxel index into the corresponding grid, associating the points in the voxel, strengthening the local feature information, and obtaining the point cloud voxel feature;

[0068] S4.4: obtaining the standard point cloud data of the motor internal structure, preliminarily clustering the standard point cloud data of the motor internal structure according to the DBSCAN clustering algorithm, obtaining the core point, and determining the number of clustering centers by the number of the core points;

[0069] S4.5: taking the core points after DBSCAN clustering as the initial center cluster of the K-means clustering algorithm, and calculating the Euclidean distance from the laser scanning point cloud voxel feature to each initial center cluster;

[0070] S4.6: clustering and dividing according to the Euclidean distance, calculating the clustering average error of the initial center cluster, updating the center cluster according to the clustering average error, judging whether the updated center cluster satisfies the convergence condition, if the convergence condition is not satisfied, continuing to update the center cluster, and if the convergence condition is satisfied, outputting the clustering result;

[0071] S4.7: extracting the point cloud voxel feature after clustering, merging the grids where the point cloud voxel features under the same center cluster are located, and extracting the internal point cloud feature, the calculation formula of the internal point cloud feature is:

[0072]

[0073] wherein, T a represents the internal point cloud feature corresponding to the laser scanning point cloud voxel feature in the same center cluster after clustering, i represents a single laser scanning point cloud voxel feature in the center cluster, M represents the total number of laser scanning point cloud voxel features in the center cluster, η i represents the feature value of the i-th laser scanning point cloud voxel feature, d represents the clustering threshold value corresponding to the center cluster, r2 represents the farthest distance scanned by the laser scanner, r1 represents the nearest distance scanned by the laser scanner, p i_x represents the voxel horizontal coordinate of the i-th laser scanning point cloud voxel feature, p i_y represents the voxel vertical coordinate of the i-th laser scanning point cloud voxel feature, p i_z represents the voxel vertical coordinate of the i-th laser scanning point cloud voxel feature;

[0074] Please refer to Figure 3The embodiment of the application is a point cloud voxel feature clustering flowchart, and specifically, by combining two clustering algorithms of DBSCAN and K-means, the standard point cloud data of the internal structure of the motor is subjected to clustering analysis, the center cluster is iteratively optimized, and a more accurate clustering result is obtained, so that the classification state of the point cloud data is better understood, and the operation efficiency of classification is improved and energy loss is reduced. DBSCAN is a density clustering algorithm, which is used to identify data points closely connected in space and divide them into clusters. DBSCAN can be used for preliminary clustering to obtain core points. The core point refers to a core area containing at least a specified number of data points within a given radius. Since each core point may represent a cluster, the number of core points can be used to preliminarily determine the number of center clusters of clustering.

[0075] The core points obtained by preliminary clustering by DBSCAN are used as the initial center clusters of the K-means clustering algorithm, providing an initial cluster center position for the K-means algorithm. The Euclidean distance is used for clustering division. Each working operation information is assigned to the nearest center cluster by calculating the Euclidean distance of the laser scanning point cloud voxel feature to each initial center cluster. The clustering average error refers to the average value of the Euclidean distance of all data points contained in the center. The clustering average error can be used to update the initial center cluster to improve the clustering effect. The position of the center cluster can be adjusted according to the calculated clustering average error, so as to update the center cluster. The convergence condition is defined to determine whether the updated center cluster meets the convergence condition to determine whether to end the iteration. If the convergence condition is met, the clustering result is output.

[0076] Please refer to Figure 4 The grid clustering result graph of the embodiment of the application first traverses the grid map, finds the first unmarked grid containing laser scanning point cloud voxel features, marks it as 1, then finds other grids containing laser scanning point cloud voxel features in the neighborhood of the grid, judges whether the center cluster of the laser scanning point cloud voxel features is the same as that of the previous laser scanning point cloud voxel features, if the same, also marks it as 1, if different, continues to find other grids, repeatedly marks the neighborhood of the grid, until all grids in the connected domain are successfully marked, records the positions of all grids in the connected domain, then repeats the above operation to find the remaining connected domains in the grid map, and marks them according to the digital serial number, finally obtains several connected regions, i.e. several grids of the same center cluster in the grid map. At this time, the laser scanning point cloud voxel features stored in each grid in the same connected domain are merged to obtain the internal point cloud features.

[0077] The specific steps of preprocessing the internal point cloud data of the motor are as follows:

[0078] S4.1.1: Calculate the y-axis coordinates of all motor internal point cloud data, and delete the point from the motor internal point cloud data if the y-axis coordinate of the point is less than 0;

[0079] S4.1.2: Traverse the query motor internal point cloud data, and delete the motor internal point cloud data with the same radar rotation angle and echo distance;

[0080] S4.1.3: Set the fitting plane equation, calculate the coefficients of the fitting plane equation according to the motor internal point cloud data, and calculate the distance from all motor internal point cloud data to the fitting plane. Compare the distance from the motor internal point cloud data to the fitting plane with the plane distance threshold value. If it is less than or equal to the plane distance threshold value, save the motor internal point cloud data. If it is greater than the plane distance threshold value, delete the motor internal point cloud data;

[0081] S4.1.4: The motor internal point cloud data is gathered in the normal direction, the position and coordinates of the sampling points are adjusted, and the motor internal point cloud data is denoised.

[0082] The three-dimensional modeling logic includes a feature classification sub-logic and a motor modeling sub-logic. The feature classification sub-logic includes constructing a feature classification network, training the feature classification network through the external structure features and internal point cloud features, and outputting the motor component type corresponding to the features. The motor modeling sub-logic includes setting rendering parameters of the motor component corresponding to the motor component type through modeling software, rendering the external structure features and internal point cloud features corresponding to the motor component type, and outputting a three-dimensional model of the ultrasonic motor. The motor component type includes an external shell, an external base, an internal stator, an internal rotor, an internal bearing, an internal friction substrate, and a piezoelectric ceramic sheet. The rendering includes scene rendering and texture rendering.

[0083] The feature classification network includes an input layer, a hidden layer, and an output layer;

[0084] The input layer is used to input the external structure features and the internal point cloud features, perform logarithmic operation and size reconstruction on the external structure features and the internal point cloud features, and obtain input parameters of the hidden layer;

[0085] The hidden layer is used to train the input parameters of the hidden layer, calculate the motor component type parameters through the attention mechanism, and the calculation formula of the motor component type parameters is:

[0086] Sesc p =σ{f 3×3 [AvgPool(γ·F+β)],MaxPool(F')},

[0087] Wherein, Sesc prepresents a motor component type parameter, represents an activation function, AvgPool(·) represents an average pooling function, represents a scale factor, represents a translation factor, MaxPool(·) represents a maximum pooling function, and F' represents an input feature through an attention mechanism;

[0088] The output layer is configured to output a classification result according to the motor component type parameter.

[0089] The input layer includes logarithmic operation, size reconstruction, and convolution;

[0090] The hidden layer includes a processing module A and a processing module B, the processing module A is configured to process external structure features, and includes a block A1 and a block A2, the block A1 and the block A2 include two convolution layers, one normalization layer, and one average pooling layer, the processing module B is configured to process internal point cloud features, and includes a block B1 and a block B2, the block B1 includes a channel attention module and a spatial attention module, and the block B2 includes three convolution layers and one maximum pooling layer.

[0091] The output layer includes error reconstruction and an activation function, the error reconstruction is configured to perform reverse error analysis on the motor component type parameter to determine whether an error threshold is met, if the error threshold is not met, retraining is performed, and if the error threshold is met, a classification category is output according to the activation function.

[0092] The virtual-real mapping module includes a data integration unit, a physical mapping unit, and a visualization display unit, the data integration unit is configured to receive inputs from sensors and a control system, convert the inputs into a unified format, and display the inputs in a motor instrument panel, the physical mapping unit is configured to convert physical properties of the motor into a digital model and map the digital model to an information space, and the visualization display unit is configured to provide an interactive page and construct a digital twin motor through a motor three-dimensional model.

[0093] The sensors include vibration sensors, temperature sensors, and speed sensors, the control system includes actuators, and the actuators are configured to operate the motor through control instructions, and the physical mapping unit specifically includes: 1) according to functions of the motor, decomposing the motor functions into corresponding logical nodes, and encapsulating the logical nodes according to data objects and attributes; 2) establishing an abstract communication service interface for the encapsulated logical nodes, and realizing information interaction services of the logical nodes according to the abstract communication service interface; 3) mapping the information interaction services of the logical nodes to an intelligent list of a virtual motor, and establishing an intelligent list information model; and 4) extracting signal connection point description information of the motor according to the intelligent list information model, combining the signal connection point description information with the motor physically, obtaining a digital model, and mapping the digital model to the information space.

[0094] The visualization display unit visually displays the structure and running state of the motor through three-dimensional graphics and animations, and enhances the user's understanding of the motor structure and running conditions.

[0095] The abnormality detection module includes a protection self-checking submodule, a power failure detection submodule and a fault warning submodule.

[0096] The protection self-checking submodule is used for detecting fault problems in the running process of the motor.

[0097] The power failure detection submodule is used for detecting fault problems that cannot be found in the running process of the motor.

[0098] The fault warning submodule is used for warning the faults detected by the protection self-checking submodule and the power failure detection submodule, and sending a notification to the staff.

[0099] The protection self-checking submodule includes a temperature detection unit, a vibration detection unit and a rotating speed detection unit, the temperature detection unit includes temperature analysis logic, the temperature analysis logic is used for monitoring the temperature data of the motor according to the temperature sensor arranged in the motor, cleaning and preprocessing the collected temperature data, drawing a frequency distribution histogram of the temperature data, and outputting the temperature state analysis result of the motor according to the frequency distribution histogram, the temperature state analysis result includes motor overheating and normal motor temperature.

[0100] The vibration detection unit includes vibration signal analysis logic, the vibration signal analysis logic includes vibration signal preprocessing, vibration signal feature extraction and vibration signal feature analysis, and outputs the vibration signal analysis result, the vibration signal analysis result includes motor imbalance, motor bearing damage and shaft bending.

[0101] The rotating speed detection unit includes rotating speed analysis logic, the rotating speed analysis logic is used for monitoring the rotating speed data of the motor according to the rotating speed sensor arranged in the motor, cleaning and preprocessing the collected rotating speed data, drawing a rotating speed change trend chart, judging whether the speed fluctuation has a long-term drift phenomenon, comparing the actual rotating speed with the predetermined working rotating speed interval, judging whether it is in the best working state, and outputting the rotating speed analysis result, the rotating speed analysis result includes motor overspeed, motor underspeed and motor unstable rotating speed.

[0102] Please refer to Figure 5 The present application provides an embodiment: vibration signal analysis flowchart, the specific steps are as follows:

[0103] S7.1: Adaptive signal decomposition is performed on the vibration signal, the kurtosis value of each component is calculated, the components are sorted in descending order according to the kurtosis value, the top ten components are selected for signal reconstruction, and the noise of the vibration signal is eliminated;

[0104] S7.2: Fourier transform is performed on the vibration signal reconstructed by the signal, the frequency domain boundary of the Fourier spectrum of the vibration signal is divided through scale space transformation, and the Fourier spectrum of the vibration signal is divided into frequency bands with different center frequencies, each frequency band corresponds to an intrinsic modal function u k ;

[0105] S7.3: Each intrinsic modal function is iterated, and the optimal solution that satisfies the minimum sum of bandwidth of each modal component is calculated;

[0106] S7.4: The power spectrum is calculated according to the optimal solution of each modal component, the characteristic frequency of the vibration signal is obtained, the component with the largest energy in the power spectrum is selected as the bearing inner ring signal, the component with the second largest energy in the power spectrum is selected as the bearing outer ring signal, and the component with the third largest energy in the power spectrum is selected as the bearing balance frame signal;

[0107] S7.5: The characteristic frequency of the rolling bearing signal is compared with the vibration frequency under normal state, and the vibration signal analysis result is output, wherein the comparison further includes time domain analysis and pattern recognition.

[0108] The power failure detection submodule includes:

[0109] Power monitoring unit: This unit is responsible for real-time monitoring of the power state of the motor, including current, voltage and other key indicators. It can detect power interruptions, voltage fluctuations and other abnormal conditions, and record the time, duration and degree of occurrence.

[0110] Historical data comparison unit: This unit compares and analyzes the data collected by the power monitoring unit with historical data to identify any abnormalities or deviations from the normal pattern. Through comparative analysis, potential power problems can be predicted to provide a basis for preventive measures.

[0111] Fault location unit: When power anomalies are detected, the fault location unit is used to determine the specific location and cause of the fault. It can distinguish between external power supply problems and problems with the motor itself, helping maintenance personnel to respond quickly and take the correct repair measures

[0112] Although the embodiments of the present application have been shown and described above, it should be understood that the above embodiments are exemplary and should not be construed as limiting the present application. Those skilled in the art can make changes, modifications, replacements and variations to the above embodiments within the scope of the present application.

Claims

1. An equipment monitoring visualization system based on digital twin technology, characterized in that, The system comprises a data acquisition module, a virtual model creation module, a virtual-real mapping module and an anomaly detection module; The data acquisition module is configured to acquire appearance data and operation data of the motor, wherein the appearance data comprises motor external images and motor internal point cloud data; The virtual model creation module is configured with a motor modeling strategy, which is configured to pre-process the appearance data, analyze the pre-processed appearance data, and output a motor three-dimensional model, wherein the motor modeling strategy comprises an external feature extraction logic, an internal feature extraction logic and a three-dimensional modeling logic; The external feature extraction logic comprises the following steps: S3.1: denoising the motor external images; S3.2: histogram statistics of the gray value of the denoised motor external images, and cutting out the target region of the motor external image pixel points through statistical learning method; S3.3: smoothing the target region image through edge detection, calculating the Gaussian difference according to the smoothed target region image, extracting the key points of the target region image, calculating the medium intensity of the key points, dividing the target region image into R, G, B and Deep four components for pixel decomposition, and calculating the medium intensity of the decomposed key points; S3.4: according to the medium intensity of the key points and the medium intensity of the four components of the decomposed key points, the medium intensity of the image is matched, and the external structure feature is calculated, and the calculation formula of the external structure feature is: where O c denotes the external structure feature, j denotes a single key point, N denotes the total number of key points, δ denotes the Fourier transform factor, x j denotes the pixel horizontal coordinate of the jth key point, y j denotes the pixel vertical coordinate of the jth key point, max[·] denotes the maximum value of the function, M(·) denotes the medium intensity matching score matrix, m Rk denotes the key point R component medium intensity, m Gk denotes the key point G component medium intensity, m Bk denotes the key point B component medium intensity, m Dk denotes the key point Deep component medium intensity, m k denotes the medium intensity of the key point; The internal feature extraction logic comprises the following steps: S4.1: according to the motor internal point cloud data, pre-processing the motor internal point cloud data, and coordinate transformation of the pre-processed motor internal point cloud data through rigid body transformation; S4.2: equally interval dividing the grid, voxelizing the motor internal point cloud data, projecting the voxel index into the corresponding grid, connecting the points in the voxel, strengthening the local feature information, and obtaining the point cloud voxel feature; S4.3: obtaining motor internal structure standard point cloud data, classifying the point cloud voxel feature through clustering algorithm, merging the grid where the point cloud voxel feature in the same center cluster is located, and calculating the internal point cloud feature; The virtual-real mapping module is configured to construct a motor logic model, map the motor logic model to an information space, and construct a digital twin motor in combination with the motor three-dimensional model; The anomaly detection module is configured to investigate suspicious faulty motors, locate the number and position of the faulty motor, and alarm the fault condition during motor visual monitoring; The anomaly detection module comprises a protection self-checking submodule, a power failure detection submodule and a fault alarm submodule; The protection self-checking submodule is configured to detect fault problems during motor operation; The power failure detection submodule is configured to detect fault problems that cannot be found during motor operation; The fault alarm submodule is configured to alarm the faults detected by the protection self-checking submodule and the power failure detection submodule, and send a notification to the staff; The protection self-checking submodule comprises a temperature detection unit, a vibration detection unit and a rotating speed detection unit; The temperature detection unit comprises temperature analysis logic for monitoring temperature data of the motor according to a temperature sensor arranged on the motor, cleaning and preprocessing the collected temperature data, drawing a frequency distribution histogram of the temperature data, and outputting a temperature state analysis result of the motor, the temperature state analysis result comprising motor overheating and normal motor temperature. The vibration detection unit comprises vibration signal analysis logic including vibration signal preprocessing, vibration signal feature extraction, and vibration signal feature analysis, and outputs a vibration signal analysis result, the vibration signal analysis result comprising motor imbalance, motor bearing damage, and shaft bending. The speed detection unit comprises speed analysis logic for monitoring speed data of the motor according to a speed sensor arranged on the motor, cleaning and preprocessing the collected speed data, drawing a speed change trend graph, determining whether there is a long-term drift phenomenon in the speed fluctuation, comparing the actual speed with a predetermined working speed interval, determining whether it is in the best working state, and outputting a speed analysis result, the speed analysis result comprising motor overspeed, motor underspeed, and motor unstable speed.

2. The equipment monitoring visualization system based on digital twin technology according to claim 1, characterized in that, The virtual model creation module comprises an image processing unit, a point cloud processing unit, and a model modeling unit, the external feature extraction logic is configured in the image processing unit, the internal feature extraction logic is configured in the point cloud processing unit, and the three-dimensional modeling logic is configured in the model modeling unit, the image processing unit is configured to process the motor external image according to the motor external image, extract the external structural features, the point cloud processing unit is configured to process the motor internal point cloud data according to the motor internal point cloud data, extract the internal point cloud features, and the model modeling unit is configured to process the external structural features and internal point cloud features, and output the three-dimensional model of the motor.

3. The equipment monitoring visualization system based on digital twin technology according to claim 2, characterized in that, The three-dimensional modeling logic comprises feature classification sub-logic and motor modeling sub-logic, the feature classification sub-logic comprises constructing a feature classification network, training the feature classification network through the external structural features and internal point cloud features, and outputting the motor component type corresponding to the features, and the motor modeling sub-logic comprises setting rendering parameters of the motor component corresponding to the motor component type through modeling software, rendering the external structural features and internal point cloud features corresponding to the motor component type, and outputting the three-dimensional model of the ultrasonic motor; The feature classification network comprises an input layer, a hidden layer, and an output layer; The input layer is configured to input the external structural features and internal point cloud features, perform logarithmic operation and size reconstruction on the external structural features and internal point cloud features, and obtain input parameters of the hidden layer; The hidden layer is configured to train the input parameters of the hidden layer, and calculate motor component type parameters through an attention mechanism; The output layer is configured to output classification results according to the motor component type parameters.

4. The equipment monitoring visualization system based on digital twin technology according to claim 3, characterized in that, The virtual-real mapping module comprises a data integration unit, a physical mapping unit and a visualization display unit, the data integration unit is used for receiving input from sensors and control systems, converting the input into a unified format, and displaying in a motor instrument panel, the physical mapping unit is used for converting physical properties of the motor into a digital model, and mapping the digital model to an information space, and the visualization display unit is used for providing an interactive page, and constructing a digital twin motor through a motor three-dimensional model.

Images