Intelligent operation and maintenance method and system for visual equipment

Through the three-dimensional imaging of the hydraulic pump and the network processing of the detailed characteristics, the visual display and status monitoring of the hydraulic pump are realized, solving the problem of description and monitoring of the complex working status of the hydraulic pump, and improving operation and maintenance efficiency and equipment utilization.

CN119046900AInactive Publication Date: 2024-11-29SHANGHAI HANGSHU INTELLIGENT TECH CO LTD
View PDF 5 Cites 0 Cited by

Patent Information

Application Number
CN202411217401.2
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-09-02
Publication Date
2024-11-29
Estimated Expiration
Not applicable · inactive patent

AI Technical Summary

Technical Problem

The working conditions of hydraulic pumps are complex and changeable, and the operational differences between different individuals of the same model are large, so it is difficult to describe their working status with mathematical models. The monitoring signals generated are multi-source, multi-dimensional and multi-modal, requiring effective analysis and processing methods. .

Method used

Through modeling software, the hydraulic pump is imaged in three-dimensionally, and the three-dimensional model is obtained, and the dashboard is built. The sensors are used to obtain the operating data for visual display. Detailed features are extracted in combination with image processing and deep learning technology, and the detailed feature fusion network is constructed to generate deep fusion points. Cloud data realizes visual display and status monitoring of hydraulic pumps, outputs operation and maintenance decision results and automatically issues alarms.

Benefits of technology

Real-time monitoring and fault warning diagnosis of hydraulic pumps are realized, which enhances the intuitive understanding of equipment operation and maintenance personnel, optimizes resource allocation, improves equipment utilization, and realizes the transformation of hydraulic systems from automation to intelligent operation and maintenance.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119046900A_ABST
    Figure CN119046900A_ABST
Patent Text Reader

Abstract

The invention relates to the technical field of equipment operation and maintenance, in particular to a visual equipment intelligent operation and maintenance method and system.The visual equipment intelligent operation and maintenance method comprises the following steps that firstly, three-dimensional imaging is conducted on a hydraulic pump through modeling software, a three-dimensional model of the hydraulic pump is obtained, a hydraulic pump instrument panel is constructed, and operation data of the hydraulic pump are obtained through a sensor; the method comprises the following steps: firstly, acquiring operation data of a hydraulic pump through a sensor, displaying the operation data in a hydraulic pump instrument panel, performing visual display on the hydraulic pump according to a three-dimensional model of the hydraulic pump and the hydraulic pump instrument panel, secondly, acquiring the operation data of the hydraulic pump through the sensor, displaying the operation data in the hydraulic pump instrument panel, performing visual display on the hydraulic pump, and finally, analyzing the visual display data and outputting an operation and maintenance decision result. And according to an operation and maintenance decision result, an alarm is automatically given and related operation and maintenance personnel are notified, resource configuration is optimized, and the equipment utilization rate is improved.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the technical field of equipment operation and maintenance, and in particular to a method and system for intelligent operation and maintenance of visualized equipment. Background Art

[0002] The development of technologies such as big data, deep learning, and cloud platforms has provided new theoretical and technical support for the intelligent operation and maintenance of equipment. "Condition-based maintenance" and "predictive maintenance" have become the development direction of modern equipment operation and maintenance systems. With the cross-integration of intelligent operation and maintenance theory and hydraulic technology, combined with the characteristics of mobile equipment hydraulic systems, such as wide distribution of operating points, mobile operations, and harsh operating environments, how to achieve condition monitoring and intelligent operation and maintenance of hydraulic systems while ensuring that the equipment meets the operating requirements has become a common scientific problem in the equipment manufacturing industry.

[0003] For example, the Chinese patent with the authorization announcement number CN104614247B discloses a visual triaxial test system, in which a transparent pressure chamber is supplied with oil by a manual pump, a hydraulic cylinder is arranged directly above the transparent pressure chamber, and the hydraulic cylinder is supplied with oil by a variable plunger hydraulic pump, a loading head is arranged at the lower end of the hydraulic cylinder piston, and the loading head is directly opposite to the pressure rod of the transparent pressure chamber, a linear differential transformer is arranged directly above the hydraulic cylinder, and the probe of the linear differential transformer is in contact with the upper end of the hydraulic cylinder piston; four cameras and three video cameras are arranged on the periphery of the transparent pressure chamber. The invention has a simple and compact structure, is easy to assemble, has low cost and good stability; the hardware-controlled test has high precision, good reliability and long load-holding time; the radial deformation of the specimen can be observed intuitively, and the radial deformation of the specimen can be more accurately measured by shooting equipment combined with image processing system software.

[0004] The above patents all have the problems raised by this background technology: the working conditions of hydraulic pumps are complex and changeable, and the operating differences between different individuals of the same model are large. It is difficult to describe the working state of the hydraulic pump with a mathematical model. The large number of monitoring signals generated during the operation stage of the hydraulic pump show multi-source, multi-dimensional and multi-modal characteristics. An accurate and effective method is needed to analyze and process the signals. In order to solve the above problems, this application designs a visual equipment intelligent operation and maintenance method and system. Summary of the invention

[0005] The technical problem to be solved by the present invention is to address the deficiencies in the prior art and provide a method and system for intelligent operation and maintenance of visualized equipment. First, a three-dimensional image of the hydraulic pump is performed through modeling software to obtain a three-dimensional model of the hydraulic pump, and a hydraulic pump dashboard is constructed. The hydraulic pump operating data is obtained through sensors and displayed on the hydraulic pump dashboard. The hydraulic pump is visualized based on the hydraulic pump three-dimensional model and the hydraulic pump dashboard. Secondly, the hydraulic pump operating data is obtained through sensors and displayed on the hydraulic pump dashboard. The hydraulic pump is visualized. Finally, the visualized display data is analyzed to output the operation and maintenance decision results, and an alarm is automatically issued and the relevant operation and maintenance personnel are notified based on the operation and maintenance decision results.

[0006] To achieve the above object, the present invention provides the following technical solutions:

[0007] A visual equipment intelligent operation and maintenance method, comprising the following steps;

[0008] S1: obtaining the original information and current structural information of the hydraulic pump, performing three-dimensional imaging of the hydraulic pump through modeling software, and obtaining a three-dimensional model of the hydraulic pump;

[0009] S2: constructing a hydraulic pump dashboard, obtaining the hydraulic pump operation data through sensors, displaying the data on the hydraulic pump dashboard, and visually displaying the hydraulic pump based on the hydraulic pump three-dimensional model and the hydraulic pump dashboard;

[0010] S3: Output the operation and maintenance decision results by analyzing the visualized display data;

[0011] S4: Establish an alarm system to automatically issue alarms and notify relevant operation and maintenance personnel based on the operation and maintenance decision results.

[0012] The original information includes the hydraulic pump technical drawings and specification information, the current structural information includes the hydraulic pump image, and the specific steps of performing three-dimensional imaging of the hydraulic pump by the modeling software are as follows:

[0013] S1.1: Obtain the main structure of the hydraulic pump in the technical drawing, and visualize the main body of the hydraulic pump through the modeling software according to the main structure of the hydraulic pump and the specification information to obtain a main model of the hydraulic pump;

[0014] S1.2: Preprocessing the hydraulic pump image to extract detail features of the hydraulic pump image by image processing technology, wherein the preprocessing includes gradient map conversion, gradient response elimination and median filtering, and the detail features include hole features, thread features and concave-convex features;

[0015] S1.3: Construct a detail feature fusion network, use the detail features as input parameters of the detail feature fusion network, train the detail feature fusion network, output deep fusion point cloud data, import the deep fusion point cloud data into point cloud processing software for processing, map it on the hydraulic pump main body model, and output a three-dimensional model of the hydraulic pump.

[0016] The hole feature is reconstructed by multi-scale convolution downsampling on the filtered hydraulic pump image, and the hole feature is calculated by morphological opening operation according to the reconstruction error function compensation;

[0017] The thread feature extracts a thread skeleton image through confidence judgment and skeleton connection, calculates a thread descriptor according to corner point features and grayscale centroid of the thread skeleton image, converts the thread descriptor into high-dimensional data through an MLP network, and adds it to the grayscale centroid to generate a thread matching vector. The generating thread matching vector includes an MLP network, comparing the thread matching vector with a standard thread library, determining the specific type and size of the thread, and generating the thread feature in combination with the thread coordinates;

[0018] The concave-convex feature extracts the surface contour of the hydraulic pump through an edge detection algorithm, identifies the raised and recessed parts of the hydraulic pump, matches the raised and recessed parts of the hydraulic pump with a preset hydraulic pump component template through a template matching algorithm, identifies the hydraulic pump component information, and labels the hydraulic pump component information according to the preset hydraulic pump component template to obtain the concave-convex feature.

[0019] The constructing of the detail feature fusion network includes:

[0020] An input layer, establishing a feature sequence according to the input parameters, wherein the feature sequence includes a hole feature sequence, a thread feature sequence and a concave-convex feature sequence;

[0021] A feature extraction layer, used to extract feature vectors of feature sequences, the feature extraction layer comprising a geometric feature extraction module, a surface feature extraction module and a space feature extraction module, the feature vectors comprising surface features, space features and geometric features;

[0022] The feature fusion layer is used to fuse surface features, spatial features and geometric features to generate deep fusion point cloud data.

[0023] The hydraulic pump operation data includes a pump vibration signal, a hydraulic pump pressure, a hydraulic pump flow rate and a hydraulic pump temperature. The construction of a hydraulic pump dashboard also includes selecting a data visualization tool, importing the hydraulic pump operation data into the data visualization tool, creating a dashboard layout, and adding a data graph. The visual display of the hydraulic pump includes integrating a hydraulic pump three-dimensional model with a hydraulic pump dashboard to provide interactive operations.

[0024] The analysis of the visualized display data includes pressure state analysis, temperature state analysis, flow state analysis and vibration signal analysis, and the output operation and maintenance decision results are used to formulate corresponding operation and maintenance strategies according to the analysis results, and the operation and maintenance strategies include preventive maintenance, corrective maintenance and optimization maintenance;

[0025] The pressure state analysis is used to monitor the pressure data of the inlet and outlet of the hydraulic pump according to the pressure sensor installed on the hydraulic pump, clean and pre-process the collected pressure data, obtain the real-time pressure change law through statistical analysis, compare it with the historical pressure change trend, and output the pressure state analysis results of the inlet and outlet;

[0026] The temperature state analysis is used to monitor the temperature data of the hydraulic oil according to the temperature sensor installed in the hydraulic pump, clean and pre-process the collected temperature data, draw a frequency distribution histogram of the temperature data, and output the temperature state analysis result of the hydraulic oil according to the frequency distribution histogram;

[0027] The flow state analysis is used to monitor the flow data of the hydraulic pump according to the flow sensor installed on the hydraulic pump, clean and pre-process the collected flow data, compare the flow data with the historical flow change trend, and output the flow analysis result of the hydraulic pump.

[0028] The specific steps of vibration signal analysis are as follows:

[0029] S7.1: Adaptively decompose the vibration signal, calculate the kurtosis value of each component, sort the components in descending order according to the kurtosis value, select the top ten components for signal reconstruction, and eliminate the vibration signal noise. The calculation formula of the component kurtosis value is:

[0030]

[0031] Among them, I k represents the component kurtosis value, σ represents the standard deviation of the component signal, E s represents the variance of the component signal, x i represents the signal amplitude of the component signal, x m represents the mean of the component signals;

[0032] S7.2: Perform Fourier transform on the reconstructed vibration signal, divide the frequency domain boundary of the Fourier spectrum of the vibration signal by scale space transform, and divide the Fourier spectrum of the vibration signal into frequency bands with different center frequencies. Each frequency band corresponds to an intrinsic mode function u k ;

[0033] S7.3: Iterate each eigenmode function and calculate the optimal solution that satisfies the minimum sum of the bandwidths of each modal component;

[0034] S7.4: Calculate the power spectrum according to the optimal solution of each modal component to obtain the characteristic frequency of the rolling bearing signal;

[0035] S7.5: Compare the characteristic frequency of the rolling bearing signal with the theoretical frequency of the rolling bearing, and output the vibration signal analysis result of the hydraulic pump.

[0036] A visualized equipment intelligent operation and maintenance system, the system comprising a three-dimensional model creation module, a hydraulic pump visualization module, a hydraulic pump monitoring module and a hydraulic pump maintenance module;

[0037] The three-dimensional model creation module is used to obtain the original information and current structural information of the hydraulic pump, perform three-dimensional imaging of the hydraulic pump through modeling software, and create a three-dimensional model of the hydraulic pump;

[0038] The hydraulic pump visualization module is used to construct a hydraulic pump dashboard, obtain hydraulic pump operation data through sensors, display it in the hydraulic pump dashboard, and visualize the hydraulic pump based on the hydraulic pump three-dimensional model and the hydraulic pump dashboard;

[0039] The hydraulic pump monitoring module is used to collect the working status data of the hydraulic pump through the sensor, display it in the visual interface, and analyze the visual display data;

[0040] The hydraulic pump maintenance module is used to output the operation and maintenance decision results according to the analysis results of the visual display data, automatically issue an alarm and notify the relevant operation and maintenance personnel according to the operation and maintenance decision results;

[0041] The three-dimensional model creation module includes:

[0042] The main body creation unit is used to obtain the main structure of the hydraulic pump in the technical drawing, and visualize the main body of the hydraulic pump through the modeling software according to the main structure and specification information of the hydraulic pump to obtain the main model of the hydraulic pump;

[0043] A feature extraction unit is used to preprocess the hydraulic pump image and extract detail features of the hydraulic pump image through image processing technology based on the preprocessed hydraulic pump image;

[0044] A three-dimensional modeling unit is used to construct a detail feature fusion network, use the detail feature as an input parameter of the detail feature fusion network, perform training through the detail feature fusion network, output deep fusion point cloud data, import the deep fusion point cloud data into point cloud processing software for processing, map it on the main model of the hydraulic pump, and output a three-dimensional model of the hydraulic pump;

[0045] The hydraulic pump monitoring module comprises:

[0046] The temperature monitoring unit is used to monitor the temperature data of the hydraulic oil according to the temperature sensor installed in the hydraulic pump, clean and pre-process the collected temperature data, draw a frequency distribution histogram of the temperature data, and output the temperature state analysis result of the hydraulic oil according to the frequency distribution histogram;

[0047] The flow monitoring unit is used to monitor the flow data of the hydraulic pump according to the flow sensor installed on the hydraulic pump, clean and pre-process the collected flow data, compare the flow data with the historical flow change trend, and output the flow analysis result of the hydraulic pump;

[0048] The pressure monitoring unit is used to monitor the pressure data of the hydraulic pump inlet and outlet according to the pressure sensor installed on the hydraulic pump, clean and pre-process the collected pressure data, obtain the real-time pressure change law through statistical analysis, compare it with the historical pressure change trend, and output the pressure status analysis results of the inlet and outlet;

[0049] The vibration signal monitoring unit is used to process the vibration signal, extract the rolling bearing signal corresponding to the hydraulic pump in the vibration signal, compare it with the standard rolling bearing signal, and output the vibration signal analysis result of the hydraulic pump.

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

[0051] 1. The present invention uses three-dimensional modeling technology to visualize the hydraulic pump, and uses the dashboard to monitor the various parameters of the hydraulic pump in real time, thereby enhancing the intuitive understanding of the equipment operation status by the operation and maintenance personnel, optimizing resource allocation, and improving equipment utilization;

[0052] 2. The present invention ensures the orderly production and avoids the occurrence of catastrophic accidents by performing status monitoring and fault early warning diagnosis on the hydraulic pump, thus realizing the transformation of the hydraulic system from automated operation and maintenance to intelligent operation and maintenance, which has important practical significance. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] Other features, objects and advantages of the present invention will become more apparent from the detailed description of non-limiting embodiments made with reference to the following drawings:

[0054] Figure 1 This is a flow chart of a method for intelligent operation and maintenance of visualized equipment according to Embodiment 1 of the present invention;

[0055] Figure 2 This is a schematic diagram of a corner feature determination area according to Embodiment 1 of the present invention;

[0056] Figure 3 This is a detailed feature fusion network structure diagram of Example 1 of the present invention;

[0057] Figure 4This is a module diagram of a visualization equipment intelligent operation and maintenance system according to Example 2 of the present invention. DETAILED DESCRIPTION

[0058] The technical solutions in the embodiments of the present invention will be described clearly and completely below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments.

[0059] Embodiment 1:

[0060] See also Figure 1 , an embodiment provided by the present invention: a method for intelligent operation and maintenance of visualized equipment, comprising the following steps;

[0061] S1: obtaining the original information and current structural information of the hydraulic pump, performing three-dimensional imaging of the hydraulic pump through modeling software, and obtaining a three-dimensional model of the hydraulic pump;

[0062] S2: constructing a hydraulic pump dashboard, obtaining the hydraulic pump operation data through sensors, displaying the data on the hydraulic pump dashboard, and visually displaying the hydraulic pump based on the hydraulic pump three-dimensional model and the hydraulic pump dashboard;

[0063] S3: Output the operation and maintenance decision results by analyzing the visualized display data;

[0064] S4: Establish an alarm system to automatically issue alarms and notify relevant operation and maintenance personnel based on the operation and maintenance decision results.

[0065] The original information includes the hydraulic pump technical drawings and specification information, the current structural information includes the hydraulic pump image, and the specific steps of performing three-dimensional imaging of the hydraulic pump by the modeling software are as follows:

[0066] S1.1: Obtain the main structure of the hydraulic pump in the technical drawing, wherein the main structure of the hydraulic pump includes a pump housing, a stator and rotor, inlet and outlet pipelines, a sealing device, a bearing, a driving device and a control valve. According to the main structure of the hydraulic pump and the specification information, the main body of the hydraulic pump is visualized by the modeling software to obtain a main model of the hydraulic pump;

[0067] S1.2: preprocessing the hydraulic pump image, and extracting detail features of the hydraulic pump image by image processing technology based on the preprocessed hydraulic pump image, wherein the detail features include hole features, thread features, and concave-convex features;

[0068] S1.3: Construct a detail feature fusion network, use the detail features as input parameters of the detail feature fusion network, train the detail feature fusion network, output deep fusion point cloud data, import the deep fusion point cloud data into point cloud processing software for processing, map it on the hydraulic pump main body model, and output a three-dimensional model of the hydraulic pump.

[0069] The specific steps of extracting the detailed features of the hydraulic pump image by image processing technology are as follows:

[0070] S1.2.1: converting the hydraulic pump image into a hydraulic pump row gradient map by a row gradient operator;

[0071] S1.2.2: Check the gradient response of each pixel in the hydraulic pump row gradient map in the gradient direction, set the response threshold, set the gradient response in the hydraulic pump row gradient map that is less than the response threshold to zero, retain only the gradient response in the hydraulic pump row gradient map that is greater than or equal to the response threshold, and suppress the non-maximum value in the hydraulic pump row gradient map according to the median filter;

[0072] S1.2.3: Based on the filtered hydraulic pump gradient map, the feature is reconstructed through multi-scale convolution downsampling, and the hole feature is calculated through morphological opening operation according to the reconstruction error function compensation. The calculation formula of the hole feature is:

[0073]

[0074] Where, represents the hole feature, RE[·] represents the reconstruction error function, f(·) represents the multi-scale convolution, represents the hydraulic pump line gradient map, represents the bias term, j represents the single Gaussian distribution of the hydraulic pump line gradient map in the mixed Gaussian model, K represents the total number of Gaussian distributions of the hydraulic pump line gradient map in the mixed Gaussian model, ω j represents the weight value of the jth Gaussian distribution, μ(·) represents the probability density function of the Gaussian distribution, M X represents the Gaussian distribution mean of the hydraulic pump line gradient map, V X represents the Gaussian distribution variance of the hydraulic pump row gradient map, ⊙ represents the corrosion operation, represents the expansion operation;

[0075] S1.2.4: Confidence judgment is performed on all pixel points in the filtered hydraulic pump line gradient map, and pixel points greater than the thread confidence threshold are sequentially connected to extract the thread skeleton map;

[0076] S1.2.5: Extract the corner features and grayscale centroid of the thread skeleton image, calculate the thread descriptor based on the angle between the corner features and the grayscale centroid, use the MLP network to convert the thread descriptor into high-dimensional data, and add it to the grayscale centroid to generate a thread matching vector. Compare the thread matching vector with the standard thread library to determine the specific type and size of the thread, and generate the thread feature in combination with the thread coordinates. The calculation formula of the thread descriptor is:

[0077]

[0078] Among them, t d represents the thread descriptor, a represents the serial number of a single corner feature of the thread skeleton graph, A represents the total number of corner features of the thread skeleton graph, Ψ a represents the angle between the ath corner feature and the grayscale centroid, β a represents the weighted value of the i-th corner feature, F(·) represents the Fourier transform function, x c and c Represents the horizontal and vertical coordinates of the grayscale centroid, c a represents the ath corner feature of the thread skeleton graph, c p Represents the average value of the corner point features of the thread skeleton graph;

[0079] S1.2.6: Perform edge detection on the filtered hydraulic pump gradient image using an edge detection algorithm, extract the surface contour of the hydraulic pump based on the edge detection result, smooth, fit and optimize the surface contour of the hydraulic pump using the least squares method, and identify the convex and concave parts of the hydraulic pump from the optimized contour;

[0080] S1.2.7: Match the convex and concave parts of the hydraulic pump with the preset hydraulic pump component template through the template matching algorithm to identify the hydraulic pump component information, and label the hydraulic pump component information according to the preset hydraulic pump component template to obtain the concave and convex features;

[0081] See also Figure 2 , a schematic diagram of a corner feature judgment area according to an embodiment of the present invention, wherein the steps of extracting corner features of a thread skeleton graph are as follows:

[0082] S1.2.5.1: Randomly select any pixel point in the thread skeleton image, mark it as point P, and calculate the gray value I of P P ;

[0083] S1.2.5.2: Draw a circle with P as the center and 4 pixels as the radius, and select 20 pixels on the edge of the circle;

[0084] S1.2.5.3: Set the grayscale threshold t. If the grayscale values ​​of 20 consecutive pixels out of 16 pixels are less than IP -t or greater than I P +t, then point P is determined to be a corner feature of the image;

[0085] The grayscale centroid determination method includes: traversing each pixel point in the thread skeleton image, setting a difference threshold, calculating the difference between the grayscale value of each pixel point and each pixel point in the neighborhood, and comparing the accumulated difference with the difference threshold. If the accumulated sum of the grayscale value differences between the pixel point and each pixel point in the neighborhood is greater than the difference threshold, the pixel point is the grayscale centroid, determining the direction of the grayscale centroid, and performing spatial weighted optimization on it;

[0086] The calculation formula of thread matching vector is:

[0087] T i (G) =d i +MLP(G i ),

[0088] Where i represents the thread matching vector number, T (G) represents a high-dimensional feature vector, d i and G i They represent grayscale centroid grayscale value and thread descriptor respectively, and MLP(·) represents the multi-layer perceptron network layer;

[0089] It is used to increase the dimension of the thread descriptor, so that the grayscale centroid and the thread descriptor are added to form a feature vector. Each thread descriptor is combined with its coordinates and confidence. The feature points are embedded in the high-dimensional vector using a multi-layer perceptron. First, the thread descriptor G is used to i Converted to 256-dimensional data, the number of channels of the five-layer perceptron are 3, 32, 64, 128 and 256 respectively;

[0090] The specific steps of the hydraulic pump surface profile are as follows:

[0091] S1.2.6.1: grayscale the filtered hydraulic pump line gradient image, and decompose the hydraulic pump line gradient image into k sub-images of different scales;

[0092] S1.2.6.2: Based on each scale sub-image, calculate the centroid intensity and direction of the scale sub-image;

[0093] S1.2.6.3: Set the centroid intensity of the scaled sub-image to the local maximum value, traverse all pixels in the direction according to the direction, and if the centroid intensity of the pixel point is less than the local maximum value, modify the grayscale value of the pixel point to the average grayscale value to obtain a new scaled sub-image;

[0094] S1.2.6.4: The edge detection results of the k new scale sub-images are fused by weighted average method to obtain a fused edge image, and the fused edge image is enhanced by gradient amplitude enhancement method to extract the surface contour of the hydraulic pump;

[0095] See also Figure 3 , the structure diagram of the detail feature fusion network of the embodiment of the present invention, the construction of the detail feature fusion network includes:

[0096] An input layer, establishing a feature sequence according to the input parameters, wherein the feature sequence includes a hole feature sequence, a thread feature sequence and a concave-convex feature sequence;

[0097] A feature extraction layer, used to extract feature vectors of feature sequences, the feature extraction layer comprising a geometric feature extraction module, a surface feature extraction module and a space feature extraction module, the feature vectors comprising surface features, space features and geometric features;

[0098] The feature fusion layer is used to fuse surface features, spatial features, and geometric features to generate deep fusion point cloud data;

[0099] The input layer is used to accumulate according to the input parameters, establish the grey differential equation, calculate the least squares parameters of the grey differential equation, and calculate the fitting value of the characteristic sequence by discretizing the least squares parameter solution;

[0100] The face feature extraction module includes a face shape feature extraction submodule, a face structure feature extraction submodule and a feature cascade module. The face shape feature extraction submodule includes a multi-layer perceptron network. Each layer first performs a symmetric operation on a feature sequence through average pooling, then describes the shape of the feature sequence through one-dimensional convolution and normalization, and finally activates according to a ReLU activation function to obtain face shape features. The face structure feature extraction submodule includes a group of Gaussian kernels for learning. The feature sequence is learned through the kernel function to obtain the radial distance, azimuth and inclination angle of each face in the feature sequence, and the local structural features of each face are encoded through convolution to obtain face structure features. The feature cascade module cascades the face shape features and the face structure features to obtain face features.

[0101] The geometric feature extraction module obtains the center point position of the feature sequence through the attention mechanism, calculates the neighboring points of the center point according to the K nearest neighbor algorithm, uses the geometric vectors of the center point and the neighboring points as the source of the Gaussian kernel function, outputs the local geometric features of the center point through the Gaussian kernel function, and finally maps all local geometric features to a high-dimensional space through a residual connection network, performs a pooling operation, and outputs the geometric features;

[0102] The spatial feature extraction module trains the feature sequence through a multi-layer perceptron network and outputs the spatial features according to the maximum pooling;

[0103] The feature fusion layer maps the spatial features, surface features and geometric features corresponding to a single feature in the feature sequence into a tuple sequence [(Q1, K1, J1), (Q2, K2, J2), ..., (Q m ,K m ,J m )], where Q m represents the corresponding input of the mth feature in the geometric feature, K m represents the corresponding input of the mth feature in the spatial feature, J m Represent the corresponding input of the mth feature in the face feature, perform dot product on the corresponding input features in each candidate sequence in the tuple sequence, calculate the correlation score after the dot product of the candidate sequence by the Pearson correlation coefficient, compress the correlation score according to the sigmoid function, and compare it with the correlation threshold. If it is greater than the correlation threshold, retain the candidate sequence. If it is less than or equal to the correlation threshold, filter the candidate sequence, collect the local features of the candidate sequence retained in the tuple sequence, and enhance them. Splice the enhanced candidate sequence along the head sequence of the tuple sequence, fuse the local features through residual connection, and perform pooling operation on the feature vector after residual connection fusion to obtain deep fusion point cloud data;

[0104] The hydraulic pump operation data includes a pump vibration signal, a hydraulic pump pressure, a hydraulic pump flow rate and a hydraulic pump temperature. The construction of a hydraulic pump dashboard includes selecting a data visualization tool, importing the hydraulic pump operation data into the data visualization tool, creating a dashboard layout, and adding data graphs. The data graphs include a real-time flow graph, a pressure change curve, a temperature trend graph and a vibration signal graph. The real-time flow graph is used to display the flow change of the hydraulic pump. The pressure change curve is used to display the pressure change of the hydraulic pump inlet and outlet. The temperature trend graph is used to display the temperature of the hydraulic oil. The vibration signal graph is used to display the vibration amplitude of the hydraulic pump. The visual display of the hydraulic pump includes integrating the three-dimensional model of the hydraulic pump and the hydraulic pump dashboard to provide interactive operation.

[0105] The analysis of the visualized display data includes pressure state analysis, temperature state analysis, flow state analysis and vibration signal analysis, and the output operation and maintenance decision results are used to formulate corresponding operation and maintenance strategies according to the analysis results, and the operation and maintenance strategies include preventive maintenance, corrective maintenance and optimization maintenance;

[0106] The pressure state analysis is used to monitor the pressure data of the inlet and outlet of the hydraulic pump according to the pressure sensor installed on the hydraulic pump, clean and pre-process the collected pressure data, obtain the real-time pressure change law through statistical analysis, compare it with the historical pressure change trend, and output the pressure state analysis results of the inlet and outlet;

[0107] The temperature state analysis is used to monitor the temperature data of the hydraulic oil according to the temperature sensor installed in the hydraulic pump, clean and pre-process the collected temperature data, draw a frequency distribution histogram of the temperature data, and output the temperature state analysis result of the hydraulic oil according to the frequency distribution histogram;

[0108] The flow state analysis is used to monitor the flow data of the hydraulic pump according to the flow sensor installed on the hydraulic pump, clean and pre-process the collected flow data, compare the flow data with the historical flow change trend, and output the flow analysis result of the hydraulic pump;

[0109] S7.1: Adaptively decompose the vibration signal, calculate the kurtosis value of each component, sort the components in descending order according to the kurtosis value, select the top ten components for signal reconstruction, and eliminate the vibration signal noise. The calculation formula of the component kurtosis value is:

[0110]

[0111] Among them, I k represents the component kurtosis value, σ represents the standard deviation of the component signal, E s represents the variance of the component signal, x i represents the signal amplitude of the component signal, x m represents the mean of the component signals;

[0112] S7.2: Perform Fourier transform on the reconstructed vibration signal, divide the frequency domain boundary of the Fourier spectrum of the vibration signal by scale space transform, and divide the Fourier spectrum of the vibration signal into frequency bands with different center frequencies. Each frequency band corresponds to an intrinsic mode function u k ;

[0113] S7.3: Iterate each eigenmode function and calculate the optimal solution that satisfies the minimum sum of the bandwidths of each modal component. The calculation formula for the optimal solution of each modal component is:

[0114]

[0115] Wherein, represents the optimal solution of each modal component, argmin(·) represents the minimization condition constraint function, represents the Lagrange penalty factor, K represents the number of intrinsic mode functions, k represents a single intrinsic mode function, represents the Dirac function, * represents the convolution operation, e represents the natural exponent, j represents the imaginary unit, represents the center frequency, t represents the horizontal axis coordinate, cos(·) represents the cosine function, represents the scale transformation penalty factor, represents the approximate coefficient of the empirical wavelet transform, represents the detail coefficient of the empirical wavelet transform, and sin(·) represents the sine function;

[0116] S7.4: Calculate the power spectrum according to the optimal solution of each modal component, obtain the characteristic frequency of the rolling bearing signal, select the component with the largest energy in the power spectrum as the rolling bearing inner ring signal, the component with the second largest energy in the power spectrum as the rolling bearing outer ring signal, the component with the third largest energy in the power spectrum as the rolling bearing rolling element signal, and the component with the second largest energy in the power spectrum as the rolling bearing cage signal;

[0117] S7.5: Compare the characteristic frequency of the rolling bearing signal with the theoretical frequency of the rolling bearing, and output the vibration signal analysis result of the hydraulic pump;

[0118] The pressure state analysis results include normal pressure state, high pressure state and low pressure state;

[0119] The temperature state analysis results include normal temperature state, high temperature state and low temperature state;

[0120] The flow analysis results include normal flow status, blocked flow status and leaked flow status;

[0121] The vibration signal analysis result includes a normal bearing state, a faulty bearing state and an abnormal bearing state.

[0122] Embodiment 2:

[0123] See also Figure 4 ,The present invention provides an embodiment: a visualization equipment intelligent operation and maintenance system, the system comprising a three-dimensional model creation module, a hydraulic pump visualization module, a hydraulic pump monitoring module and a hydraulic pump maintenance module;

[0124] The three-dimensional model creation module is used to obtain the original information and current structural information of the hydraulic pump, perform three-dimensional imaging of the hydraulic pump through modeling software, and create a three-dimensional model of the hydraulic pump;

[0125] The hydraulic pump visualization module is used to construct a hydraulic pump dashboard, obtain hydraulic pump operation data through sensors, display it in the hydraulic pump dashboard, and visualize the hydraulic pump based on the hydraulic pump three-dimensional model and the hydraulic pump dashboard;

[0126] The hydraulic pump monitoring module is used to collect the working status data of the hydraulic pump through the sensor, display it in the visual interface, and analyze the visual display data;

[0127] The hydraulic pump maintenance module is used to output the operation and maintenance decision results according to the analysis results of the visual display data, automatically issue an alarm and notify the relevant operation and maintenance personnel according to the operation and maintenance decision results;

[0128] The three-dimensional model creation module includes:

[0129] The main body creation unit is used to obtain the main structure of the hydraulic pump in the technical drawing, and visualize the main body of the hydraulic pump through the modeling software according to the main structure and specification information of the hydraulic pump to obtain the main model of the hydraulic pump;

[0130] A feature extraction unit is used to preprocess the hydraulic pump image and extract detail features of the hydraulic pump image through image processing technology based on the preprocessed hydraulic pump image;

[0131] A three-dimensional modeling unit is used to construct a detail feature fusion network, use the detail feature as an input parameter of the detail feature fusion network, perform training through the detail feature fusion network, output deep fusion point cloud data, import the deep fusion point cloud data into point cloud processing software for processing, map it on the main model of the hydraulic pump, and output a three-dimensional model of the hydraulic pump;

[0132] The hydraulic pump monitoring module comprises:

[0133] The temperature monitoring unit is used to monitor the temperature data of the hydraulic oil according to the temperature sensor installed in the hydraulic pump, clean and pre-process the collected temperature data, draw a frequency distribution histogram of the temperature data, and output the temperature state analysis result of the hydraulic oil according to the frequency distribution histogram;

[0134] The flow monitoring unit is used to monitor the flow data of the hydraulic pump according to the flow sensor installed on the hydraulic pump, clean and pre-process the collected flow data, compare the flow data with the historical flow change trend, and output the flow analysis result of the hydraulic pump;

[0135] The pressure monitoring unit is used to monitor the pressure data of the hydraulic pump inlet and outlet according to the pressure sensor installed on the hydraulic pump, clean and pre-process the collected pressure data, obtain the real-time pressure change law through statistical analysis, compare it with the historical pressure change trend, and output the pressure status analysis results of the inlet and outlet;

[0136] The vibration signal monitoring unit is used to process the vibration signal, extract the rolling bearing signal corresponding to the hydraulic pump in the vibration signal, compare it with the standard rolling bearing signal, and output the vibration signal analysis result of the hydraulic pump.

[0137] Although the embodiments of the present invention have been shown and described above, it is to be understood that the above embodiments are exemplary and are not to be construed as limitations of the present invention. A person skilled in the art may change, modify, replace and vary the above embodiments within the scope of the present invention.

Claims

1. A visual equipment intelligent operation and maintenance method, characterized in that: The steps include: S1: obtaining the original information and current structural information of the hydraulic pump, performing three-dimensional imaging of the hydraulic pump through modeling software, and obtaining a three-dimensional model of the hydraulic pump; S2: constructing a hydraulic pump dashboard, obtaining the hydraulic pump operation data through sensors, displaying the data on the hydraulic pump dashboard, and visually displaying the hydraulic pump based on the hydraulic pump three-dimensional model and the hydraulic pump dashboard; S3: Output the operation and maintenance decision results by analyzing the visualized display data; S4: Establish an alarm system to automatically issue alarms and notify relevant operation and maintenance personnel based on the operation and maintenance decision results.

2. According to claim 1, a visualization equipment intelligent operation and maintenance method is characterized in that: The original information includes the hydraulic pump technical drawings and specification information, the current structural information includes the hydraulic pump image, and the specific steps of performing three-dimensional imaging of the hydraulic pump by the modeling software are as follows: S1.1: Obtain the main structure of the hydraulic pump in the technical drawing, and visualize the main body of the hydraulic pump through the modeling software according to the main structure of the hydraulic pump and the specification information to obtain a main model of the hydraulic pump; S1.2: Preprocessing the hydraulic pump image to extract detail features of the hydraulic pump image by image processing technology, wherein the preprocessing includes gradient map conversion, gradient response elimination and median filtering, and the detail features include hole features, thread features and concave-convex features; S1.3: Construct a detail feature fusion network, use the detail features as input parameters of the detail feature fusion network, train the detail feature fusion network, output deep fusion point cloud data, import the deep fusion point cloud data into point cloud processing software for processing, map it on the hydraulic pump main body model, and output a three-dimensional model of the hydraulic pump.

3. According to claim 2, a visualization equipment intelligent operation and maintenance method is characterized in that: The hole feature is reconstructed by multi-scale convolution downsampling on the filtered hydraulic pump image, and the hole feature is calculated by morphological opening operation according to the reconstruction error function compensation; The thread feature extracts a thread skeleton image through confidence judgment and skeleton connection, calculates a thread descriptor according to corner point features and grayscale centroid of the thread skeleton image, converts the thread descriptor into high-dimensional data through an MLP network, and adds it to the grayscale centroid to generate a thread matching vector. The generating thread matching vector includes an MLP network, comparing the thread matching vector with a standard thread library, determining the specific type and size of the thread, and generating the thread feature in combination with the thread coordinates; The concave-convex feature extracts the surface contour of the hydraulic pump through an edge detection algorithm, identifies the raised and recessed parts of the hydraulic pump, matches the raised and recessed parts of the hydraulic pump with a preset hydraulic pump component template through a template matching algorithm, identifies the hydraulic pump component information, and labels the hydraulic pump component information according to the preset hydraulic pump component template to obtain the concave-convex feature.

4. According to claim 3, a visualization equipment intelligent operation and maintenance method is characterized in that: The constructing of the detail feature fusion network includes: An input layer, establishing a feature sequence according to the input parameters, wherein the feature sequence includes a hole feature sequence, a thread feature sequence and a concave-convex feature sequence; A feature extraction layer, used to extract feature vectors of feature sequences, the feature extraction layer comprising a geometric feature extraction module, a surface feature extraction module and a space feature extraction module, the feature vectors comprising surface features, space features and geometric features; The feature fusion layer is used to fuse surface features, spatial features and geometric features to generate deep fusion point cloud data.

5. According to claim 1, a visualization equipment intelligent operation and maintenance method is characterized in that: The hydraulic pump operation data includes a pump vibration signal, a hydraulic pump pressure, a hydraulic pump flow rate and a hydraulic pump temperature. The construction of a hydraulic pump dashboard also includes selecting a data visualization tool, importing the hydraulic pump operation data into the data visualization tool, creating a dashboard layout, and adding a data graph. The visual display of the hydraulic pump includes integrating a hydraulic pump three-dimensional model with a hydraulic pump dashboard to provide interactive operations.

6. According to claim 5, a visualization equipment intelligent operation and maintenance method is characterized in that: The analysis of the visualized display data includes pressure state analysis, temperature state analysis, flow state analysis and vibration signal analysis, and the output operation and maintenance decision results are used to formulate corresponding operation and maintenance strategies according to the analysis results, and the operation and maintenance strategies include preventive maintenance, corrective maintenance and optimization maintenance; The pressure state analysis is used to monitor the pressure data of the inlet and outlet of the hydraulic pump according to the pressure sensor installed on the hydraulic pump, clean and pre-process the collected pressure data, obtain the real-time pressure change law through statistical analysis, compare it with the historical pressure change trend, and output the pressure state analysis results of the inlet and outlet; The temperature state analysis is used to monitor the temperature data of the hydraulic oil according to the temperature sensor installed in the hydraulic pump, clean and pre-process the collected temperature data, draw a frequency distribution histogram of the temperature data, and output the temperature state analysis result of the hydraulic oil according to the frequency distribution histogram; The flow state analysis is used to monitor the flow data of the hydraulic pump according to the flow sensor installed on the hydraulic pump, clean and pre-process the collected flow data, compare the flow data with the historical flow change trend, and output the flow analysis result of the hydraulic pump.

7. According to claim 6, a visualization equipment intelligent operation and maintenance method is characterized in that: The specific steps of vibration signal analysis are as follows: S7.1: Adaptively decompose the vibration signal, calculate the kurtosis value of each component, sort the components in descending order according to the kurtosis value, select the top ten components for signal reconstruction, and eliminate the vibration signal noise. The calculation formula of the component kurtosis value is: Among them, I k represents the component kurtosis value, σ represents the standard deviation of the component signal, E s represents the variance of the component signal, x i represents the signal amplitude of the component signal, x m represents the mean of the component signals; S7.2: Perform Fourier transform on the reconstructed vibration signal, divide the frequency domain boundary of the Fourier spectrum of the vibration signal by scale space transform, and divide the Fourier spectrum of the vibration signal into frequency bands with different center frequencies. Each frequency band corresponds to an intrinsic mode function u k ; S7.3: Iterate each eigenmode function and calculate the optimal solution that satisfies the minimum sum of the bandwidths of each modal component; S7.4: Calculate the power spectrum according to the optimal solution of each modal component to obtain the characteristic frequency of the rolling bearing signal; S7.5: Compare the characteristic frequency of the rolling bearing signal with the theoretical frequency of the rolling bearing, and output the vibration signal analysis result of the hydraulic pump.

8. A visualization equipment intelligent operation and maintenance system, used to implement a visualization equipment intelligent operation and maintenance method as claimed in any one of claims 1 to 7, characterized in that: The system includes a three-dimensional model creation module, a hydraulic pump visualization module, a hydraulic pump monitoring module and a hydraulic pump maintenance module; The three-dimensional model creation module is used to obtain the original information and current structural information of the hydraulic pump, perform three-dimensional imaging of the hydraulic pump through modeling software, and create a three-dimensional model of the hydraulic pump; The hydraulic pump visualization module is used to construct a hydraulic pump dashboard, obtain hydraulic pump operation data through sensors, display it in the hydraulic pump dashboard, and visualize the hydraulic pump based on the hydraulic pump three-dimensional model and the hydraulic pump dashboard; The hydraulic pump monitoring module is used to collect the working status data of the hydraulic pump through the sensor, display it in the visual interface, and analyze the visual display data; The hydraulic pump maintenance module is used to output operation and maintenance decision results based on the analysis results of the visual display data, and automatically issue an alarm and notify relevant operation and maintenance personnel based on the operation and maintenance decision results.

Citation Information

Patent Citations

  • Disclosed is substation equipment state monitoring panoramic analysis system

    CN110610542A

  • Pump station start-stop control method and system based on three-dimensional model

    CN111123742A

  • Intelligent damage detection and identification method and system based on acoustic emission signal

    CN116975656A

  • Intelligent operation and maintenance cloud platform for industrial equipment

    CN118154174A

  • Pump equipment monitoring system

    CN212620867U