Intelligent operation and maintenance method and system based on intelligent cabinet

Abstract

The present application provides an intelligent operation and maintenance method and system based on an intelligent cabinet, which relates to the field of intelligent operation and maintenance technology. By performing finite element analysis on the operating parameters and axis status information of the intelligent cabinet, a three-dimensional finite element model of the intelligent cabinet is obtained, and a node information matrix is determined based on the discretized structure of the tetrahedral unit of the intelligent cabinet; the node information matrix is hierarchically clustered to obtain a cluster center set, and then a similarity response is performed on the three-dimensional finite element model through the cluster center set to obtain a response value of each cluster center; the historical response data of the intelligent cabinet and all response values are used to construct a mechanical response field of the intelligent cabinet, and a digital twin model of the intelligent cabinet is constructed based on a deep neural network; the operation and maintenance instructions of the intelligent cabinet are determined based on the mechanical response field and the digital twin model, and then the operation and maintenance instructions are used to control the intelligent cabinet for remote operation and maintenance. The present application can perform accurate twin prediction of the operating status of the intelligent cabinet to realize intelligent operation and maintenance of the server cabinet.

Description

Technical Field

[0001] The present application relates to the field of intelligent operation and maintenance technology, and more specifically, to an intelligent operation and maintenance method and system based on intelligent cabinets. Background Art

[0002] As an important development direction in the field of equipment operation and maintenance, intelligent operation and maintenance technology has gradually been applied to multiple scenarios such as electricity, communications, transportation, and data centers. Intelligent operation and maintenance relies on technical means such as multi-source data collection, modeling analysis, machine learning, and expert rules to achieve real-time monitoring, intelligent diagnosis, anomaly prediction, and remote control of the operating status of key equipment, effectively reducing operation and maintenance costs and improving system reliability and stability.

[0003] As infrastructure supporting core components such as servers, power modules, and communications equipment, the stability of smart cabinet operation directly impacts the business continuity of the entire system. Demand for intelligent O&M (O&M) for smart cabinets is increasing, and related research is advancing. Existing smart cabinet-based intelligent O&M methods typically rely on sensors to collect operational data such as device temperature, current, voltage, and switch status. These methods implement status assessment and anomaly alerts by setting thresholds or using simple models. However, these methods lack structural response modeling, making it difficult to comprehensively assess potential operational hazards. They also fail to construct a digital twin model that reflects the cabinet's dynamic behavior and cannot dynamically integrate structural response with operational data, limiting prediction accuracy. Furthermore, they lack effective dimensionality reduction and clustering mechanisms for high-dimensional node data, resulting in low model computational efficiency. Furthermore, these models rely on fixed thresholds or simple rule-based judgments, making them unsuitable for complex O&M scenarios with multiple operating conditions, models, and structures. Therefore, accurately predicting the operating status of smart cabinets through twin predictions to achieve intelligent O&M for server cabinets remains a challenging task for the industry. Summary of the Invention

[0004] This application provides an intelligent operation and maintenance method and system based on smart cabinets, which can perform accurate twin predictions on the operating status of smart cabinets to realize intelligent operation and maintenance of server cabinets.

[0005] In a first aspect, the present application provides an intelligent operation and maintenance method and system based on an intelligent cabinet, wherein the operation and maintenance method comprises the following steps:

[0006] Collect the operating parameters of the intelligent cabinet and the axis status information of the intelligent cabinet rolling door;

[0007] Performing finite element analysis on the operating parameters and the axis state information to obtain a three-dimensional finite element model of the intelligent cabinet, and determining a node information matrix based on a discretized structure of tetrahedral units of the intelligent cabinet;

[0008] Performing hierarchical clustering on the node information matrix to obtain a cluster center set, and then performing similarity response on the three-dimensional finite element model through the cluster center set to obtain a response value of each cluster center;

[0009] Acquire historical response data of the smart cabinet, construct a mechanical response field of the smart cabinet using the historical response data and all response values, and build a digital twin model of the smart cabinet based on a deep neural network;

[0010] The operation and maintenance instructions of the smart cabinet are determined based on the mechanical response field and the digital twin model, and the operation and maintenance instructions are then used to control the smart cabinet to perform remote operation and maintenance.

[0011] In this embodiment, the operating parameters of the intelligent cabinet are collected by a load sensor, and the axis status information of the rolling door of the intelligent cabinet is collected by an angle sensor.

[0012] In this embodiment, performing finite element analysis on the operating parameters and the axis state information to obtain a three-dimensional finite element model of the intelligent cabinet specifically includes:

[0013] determining boundary conditions and load conditions of the intelligent cabinet through the axis state information;

[0014] Finite element solution is performed on the operating parameters, and then a three-dimensional finite element model of the intelligent cabinet is constructed based on the result of the finite element solution, the boundary conditions and the load conditions.

[0015] In this embodiment, determining the node information matrix based on the discretized structure of the intelligent cabinet tetrahedron unit specifically includes:

[0016] Obtain the spatial coordinate information of the tetrahedron unit of the intelligent cabinet;

[0017] Determine a discrete node set of the intelligent cabinet tetrahedron unit based on the spatial coordinate information;

[0018] The discrete node sets are combined and arranged to obtain a node information matrix.

[0019] In this embodiment, performing hierarchical clustering on the node information matrix to obtain a cluster center set specifically includes:

[0020] Normalizing the eigenvalue of each node in the node information matrix to obtain normalized node data;

[0021] Determining the similarity of the node information matrix through the normalized node data;

[0022] A node hierarchical clustering tree is constructed based on the similarity, and multiple cluster centers are extracted through the node hierarchical clustering tree to obtain a cluster center set.

[0023] In this embodiment, performing similarity response on the three-dimensional finite element model through the cluster center set to obtain the response value of each cluster center specifically includes:

[0024] Extracting the response characteristic parameters of nodes based on the three-dimensional finite element model;

[0025] Performing a similarity comparison between the response feature parameters and the center features of the cluster center set to obtain a response matching degree between the node and each cluster center;

[0026] A similar mapping is performed on each cluster center using the response matching degree to obtain a response value of each cluster center.

[0027] In this embodiment, historical response data of the smart cabinet is obtained based on a displacement sensor and a data collector on a rolling door shaft of the smart cabinet.

[0028] In this embodiment, the deep neural network is a fully connected neural network, and the digital twin model is a bidirectional coupling model, which is used for mechanical response monitoring and operation status synchronization.

[0029] In this embodiment, the smart cabinet includes: a tetrahedron unit, a rolling door, a temperature control unit, a sensor unit, and a control unit.

[0030] In a second aspect, the present application provides an intelligent operation and maintenance system based on an intelligent cabinet, for executing an intelligent operation and maintenance method based on an intelligent cabinet, the operation and maintenance system comprising:

[0031] A data acquisition module is used to collect the operating parameters of the intelligent cabinet and the axis status information of the intelligent cabinet rolling door;

[0032] a finite element analysis module, configured to perform finite element analysis on the operating parameters and the axis state information to obtain a three-dimensional finite element model of the intelligent cabinet, and determine a node information matrix based on a discretized structure of tetrahedral units of the intelligent cabinet;

[0033] A cluster response module is used to perform hierarchical clustering on the node information matrix to obtain a cluster center set, and then perform similarity response on the three-dimensional finite element model through the cluster center set to obtain a response value of each cluster center;

[0034] A digital twin module is used to obtain historical response data of the smart cabinet, construct a mechanical response field of the smart cabinet using the historical response data and all response values, and build a digital twin model of the smart cabinet based on a deep neural network;

[0035] The remote operation and maintenance module is used to determine the operation and maintenance instructions of the smart cabinet based on the mechanical response field and the digital twin model, and then control the smart cabinet to perform remote operation and maintenance according to the operation and maintenance instructions.

[0036] The technical solutions provided by the embodiments disclosed in this application have the following beneficial effects:

[0037] By collecting the operating parameters of the smart cabinet and the axis status information of the smart cabinet rolling door; performing finite element analysis on the operating parameters and the axis status information to obtain a three-dimensional finite element model of the smart cabinet, and determining the node information matrix based on the discretized structure of the smart cabinet tetrahedron unit; hierarchically clustering the node information matrix to obtain a cluster center set, and then performing similarity response on the three-dimensional finite element model through the cluster center set to obtain the response value of each cluster center; obtaining the historical response data of the smart cabinet, constructing the mechanical response field of the smart cabinet through the historical response data and all response values, and constructing a digital twin model of the smart cabinet based on a deep neural network; determining the operation and maintenance instructions of the smart cabinet based on the mechanical response field and the digital twin model, and then controlling the smart cabinet to perform remote operation and maintenance by the operation and maintenance instructions.

[0038] It can be seen that in this application, intelligent operation and maintenance of server cabinets can be realized. First, by collecting the operating parameters of the intelligent cabinet and the status information of the rolling door axis, a comprehensive perception of the actual operating environment of the cabinet can be achieved, providing basic data for building physical models and response analysis, and ensuring the authenticity and integrity of the model input; and by discretizing the cabinet structure into tetrahedral units, establishing its accurate three-dimensional physical response model, forming a node-level response matrix, providing structural support for subsequent response dimensionality reduction modeling, node response extraction and clustering analysis, and ensuring the engineering physics foundation of twin modeling; secondly, by hierarchical clustering of the node information matrix, obtaining the cluster center set, and performing similarity response mapping, it can be realized. The high-dimensional node response data is effectively reduced in dimension, representative node response features are extracted, and the computational burden of modeling is reduced. At the same time, the overall response trend is retained through the similarity response mechanism, which is conducive to building a twin model that takes both accuracy and efficiency into consideration. Then, by fusing historical response data with clustering results, the time dimension modeling of the structural state is realized, and with the help of deep neural networks, a nonlinear mapping relationship between the operating state and the structural response is established to build a digital twin system with sustainable learning and prediction capabilities. Finally, by combining the predicted response results with the state synchronization results, operation and maintenance decision instructions based on the actual structural situation are generated to realize remote control and fault warning, conduct remote intelligent operation and maintenance, and improve the level of intelligent operation and maintenance.

[0039] To sum up, the technical solution adopted in this application can perform accurate twin predictions on the operating status of the smart cabinet to realize intelligent operation and maintenance of the server cabinet. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the following is a brief introduction to the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only the embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative labor.

[0041] Figure 1 This is a flow chart of an intelligent operation and maintenance method based on an intelligent cabinet provided in this application;

[0042] Figure 2 is an exemplary flow chart for determining a three-dimensional finite element model of a smart cabinet according to the present application;

[0043] Figure 3 is an exemplary flow chart for determining a cluster center set provided by the present application;

[0044] Figure 4 It is a module structure diagram of the operation and maintenance system provided by this application. DETAILED DESCRIPTION

[0045] The following will be combined with the drawings in the embodiments of this application to clearly and completely describe the technical solutions in the embodiments of this application. Obviously, the embodiments described are only part of the embodiments of this application, not all of the embodiments. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of this application.

[0046] The embodiment of the present application provides an intelligent operation and maintenance method and system based on a smart cabinet, the core of which is to collect the operating parameters of the smart cabinet and the axis status information of the rolling shutter door of the smart cabinet; perform finite element analysis on the operating parameters and the axis status information to obtain a three-dimensional finite element model of the smart cabinet, and determine the node information matrix based on the discretization structure of the smart cabinet tetrahedron unit; hierarchically cluster the node information matrix to obtain a cluster center set, and then perform similarity response on the three-dimensional finite element model through the cluster center set to obtain the response value of each cluster center; obtain historical response data of the smart cabinet, construct a mechanical response field of the smart cabinet through the historical response data and all response values, and construct a digital twin model of the smart cabinet based on a deep neural network; determine the operation and maintenance instructions of the smart cabinet based on the mechanical response field and the digital twin model, and then control the smart cabinet to perform remote operation and maintenance by the operation and maintenance instructions.

[0047] Example 1: In order to better understand the above technical solution, the above technical solution will be described in detail below with reference to the accompanying drawings and specific implementation methods. Figure 1As shown in FIG, this figure is an exemplary flow chart of an intelligent operation and maintenance method based on an intelligent cabinet according to this embodiment of the present application, and the operation and maintenance method includes the following steps:

[0048] In step S1, the operating parameters of the intelligent cabinet and the axis status information of the intelligent cabinet rolling door are collected.

[0049] During specific implementation, the operating parameters of the smart cabinet can be collected through a load sensor, wherein the load sensor is a sensor preset in the smart cabinet, which is used to collect the operating parameters of the equipment during operation in real time, and the operating parameters include: stress changes, current load fluctuations and other load data; in addition, during specific implementation, the axis state information of the rolling shutter door of the smart cabinet can be collected through an angle sensor, wherein the angle sensor refers to a rotation angle sensor installed on one side of the rotating axis of the rolling shutter door of the smart cabinet. As a preferred embodiment, the rotation angle sensor can use a Hall encoder, which is conducive to detecting the opening and closing state, angular velocity or rotation angle change of the rolling shutter door of the smart cabinet. The acquisition frequency of the angle sensor can be set between 10-50Hz, which is convenient for short-term storage of the collected axis state information data. The axis state information is information reflecting the physical action of the rolling shutter door of the smart cabinet. The axis state of the rolling shutter door includes: opening and closing angles, rotation speed, opening and closing times and other state variables.

[0050] It should be noted that the intelligent cabinet in this application includes: tetrahedron units, rolling shutter doors, temperature control units, sensor units, and control units, wherein the tetrahedron units are basic grid units for finite element solution, which are used to describe the distribution of physical quantities such as displacement and stress in the cabinet structure; the rolling shutter doors are used to realize cabinet opening and closing control, and provide protection for the physical safety and environmental isolation of servers or communication equipment. The opening and closing state of the rolling shutter doors can directly affect the air circulation, temperature distribution and load transfer inside the cabinet, and is a key component for structural response prediction; the temperature control unit includes fans, heaters, air conditioning modules, etc. The temperature control unit The element can achieve environmental adaptive adjustment by responding to and operating and maintenance instructions; the sensor unit refers to the perception component deployed in different positions of the smart cabinet (such as rolling door shaft, internal four-sided frame, power supply, etc.). The sensor unit may include load sensors, displacement sensors, temperature and humidity sensors, angle encoders, etc., which are used to collect operating parameters and structural status information to provide data support for subsequent twin modeling; the control unit is used to receive operation and maintenance instructions, and control the execution actions of various functional modules of the cabinet (such as rolling door, temperature control unit) through operation and maintenance instructions to achieve remote opening and closing, alarm prompts, temperature adjustment and other functions. The control unit can be implemented based on an embedded PLC.

[0051] In step S2, a finite element analysis is performed on the operating parameters and the axis state information to obtain a three-dimensional finite element model of the intelligent cabinet, and a node information matrix is determined based on the discretized structure of the tetrahedral unit of the intelligent cabinet.

[0052] Preferably, in this embodiment, reference Figure 2 As shown in FIG. 1 , this figure is an exemplary flow chart for determining a three-dimensional finite element model of an intelligent cabinet in an embodiment of the present application. In this embodiment, finite element analysis is performed on the operating parameters and the axis state information to obtain a three-dimensional finite element model of the intelligent cabinet, which can be specifically achieved by the following steps:

[0053] First, in step S21, the boundary conditions and load conditions of the intelligent cabinet are determined according to the axis state information;

[0054] Then, in step S22, a finite element method is performed on the operating parameters, and a three-dimensional finite element model of the intelligent cabinet is constructed based on the result of the finite element method, the boundary conditions, and the load conditions.

[0055] In the specific implementation, first, the opening and closing angle, rotation speed and opening and closing times of the rolling shutter door collected by the angle sensor are used through the boundary condition calibration technology to set the rotational freedom and friction damping parameters of the hinge node in the finite element model. The friction damping parameters can be used as the load conditions of the intelligent cabinet to simulate the energy dissipation of the hinge node, and the rotational freedom is used as the boundary conditions of the intelligent cabinet to reflect the force and motion restrictions of the hinge node. Among them, the closed-loop calibration can be used to fine-tune the parameters of the axis state information fed back by the angle sensor in real time to reduce the error between the numerical model and the physical cabinet; then, the operating parameters can be solved by finite element method through the solver of ABAQUS software, and the finite element model can be obtained based on the results of the finite element solution, the boundary conditions and the load conditions through the modeling function of the ABAQUS software, and the finite element model is outputted as the three-dimensional finite element model of the intelligent cabinet.

[0056] It should be noted that the boundary condition calibration technology in this embodiment refers to the combination of sensor calibration and simulation feedback to determine the degree of freedom constraints and friction parameters that best represent the actual hinge behavior; by simulating the stress, strain, displacement and other mechanical behaviors of the smart cabinet under different boundary conditions and load conditions, the structural response characteristics of different nodes in the smart cabinet under working conditions or abnormal working conditions can be quantitatively characterized, which can provide predictive support for subsequent intelligent operation and maintenance.

[0057] In this embodiment, the node information matrix is determined based on the discretized structure of the intelligent cabinet tetrahedron unit in the following manner:

[0058] Obtain the spatial coordinate information of the tetrahedron unit of the intelligent cabinet;

[0059] Determine a discrete node set of the intelligent cabinet tetrahedron unit based on the spatial coordinate information;

[0060] The discrete node sets are combined and arranged to obtain a node information matrix.

[0061] In specific implementation, first, the spatial coordinate information of the tetrahedral unit can be obtained through three-dimensional scanning technology, wherein the spatial coordinate information includes the three-dimensional position of each node in the cabinet coordinate system; as a preferred embodiment, the assembly drawing of the intelligent cabinet and the three-dimensional scanning data can be combined to improve the coordinate accuracy through a registration algorithm; then, the nodes with the same position and high repetition in the spatial coordinate information are merged, and the nodes of the tetrahedral unit of the intelligent cabinet are renumbered to obtain a discrete node set: finally, the three-dimensional coordinates of each node are used as matrix columns, and all nodes are filled into the matrix rows in a predefined order (such as from the bottom to the top), so that the filled matrix is used as the node information matrix. Optionally, a column of topological relationship weights (such as distance) between nodes is added to the matrix to facilitate the calculation of the similarity of node states, so as to improve the accuracy of node state response.

[0062] It should be noted that the node information matrix in this application is a data table structure that organizes the multi-dimensional attributes of each discrete node in the three-dimensional finite element model of the intelligent cabinet into rows and columns. Each row of the node information matrix corresponds to a tetrahedral mesh node, and each column corresponds to the characteristic value of the mesh node (such as three-dimensional coordinates and topological relationship weights, etc.). By integrating spatial, structural and mechanical multi-source information, the geometric position and structural importance of the node can be considered at the same time.

[0063] In step S3, the node information matrix is hierarchically clustered to obtain a cluster center set, and then similarity response is performed on the three-dimensional finite element model through the cluster center set to obtain a response value of each cluster center.

[0064] Preferably, in this embodiment, reference Figure 3 As shown in FIG, this figure is an exemplary flow chart for determining a cluster center set in an embodiment of the present application. In this embodiment, the node information matrix is hierarchically clustered to obtain a cluster center set, which can be specifically implemented by the following steps:

[0065] First, in step S31, the eigenvalue of each node in the node information matrix is normalized to obtain normalized node data;

[0066] Then, in step S32, the similarity of the node information matrix is determined using the normalized node data;

[0067] Finally, in step S33, a node hierarchical clustering tree is constructed based on the similarity, and then multiple cluster centers are extracted through the node hierarchical clustering tree to obtain a cluster center set.

[0068] In specific implementation, first, the eigenvalue of each node in the node information matrix can be transformed into a Z-score through normalization preprocessing to obtain normalized node data with consistent dimension; then, the similarity between nodes is calculated based on the normalized node data. As a preferred embodiment, the variance of the Euclidean distance between the normalized node data can be used as the similarity of the node information matrix; finally, the node information matrix is traversed from bottom to top based on the similarity using an agglomerative hierarchical clustering algorithm to obtain a node hierarchical clustering tree, and then the node hierarchical clustering tree is divided into multiple clusters through a random forest algorithm, and then all eigenvalues of each cluster are converted into eigenvectors through unique hot encoding, and then the arithmetic mean of all eigenvectors in each cluster is used as the cluster center, and then the set consisting of all cluster centers is used as the cluster center set.

[0069] It should be noted that the node hierarchical clustering tree in this embodiment refers to a tree structure that reflects the merging process between nodes at different clustering granularities, and the cluster center set refers to a set of representative node features extracted from each cluster, which is used for subsequent similarity response mapping of the three-dimensional finite element model. In this embodiment, by constructing a node hierarchical clustering tree and extracting the cluster center set, the mechanical response pattern of the intelligent cabinet under different structural states can be summarized, which is conducive to improving the decision-making accuracy of the entire intelligent operation and maintenance system.

[0070] In this embodiment, similarity response is performed on the three-dimensional finite element model by using the cluster center set to obtain the response value of each cluster center in the following manner, namely:

[0071] Extracting the response characteristic parameters of nodes based on the three-dimensional finite element model;

[0072] Performing a similarity comparison between the response feature parameters and the center features of the cluster center set to obtain a response matching degree between the node and each cluster center;

[0073] A similar mapping is performed on each cluster center using the response matching degree to obtain a response value of each cluster center.

[0074] In specific implementation, all nodes of the three-dimensional finite element model are traversed in sequence through script batch processing to extract the response characteristic parameters of the nodes. The response characteristic parameters include: displacement vector amplitude and stress tensor components. Among them, the displacement vector amplitude reflects the overall motion magnitude of the node and can be used to identify high deformation areas. The stress tensor components contain normal stress and stress information, which can characterize the stress state in the direction of the node. Then, the cosine similarity is used to compare the similarity of the response characteristic parameters of each node with the cluster center, and the similarity comparison result is used as the response matching degree of each node and each cluster center. Finally, the existing nearest neighbor selection calculation formula can be used to select multiple nearest neighbor nodes with high response matching degrees, and then the average value of the response characteristic parameters of the cluster center and all the selected nearest neighbor nodes is used as the response value of the cluster center. The response value of each cluster center can be obtained through the above steps.

[0075] It should be noted that the response value in this application refers to the average physical response index of the node area in the neighborhood of the cluster center under specific load and boundary conditions, which can be obtained by aggregating the response characteristic parameters of similar nodes. The response value includes the displacement amplitude and stress tensor components. The cluster center constructs a model simplification interface through the response mapping mechanism, which can form an efficient bridging mechanism from the structural model to the digital twin training module; in addition, in this embodiment, the typical response distribution pattern of the overall structure is reflected by the cluster center response value to perform similarity response on the three-dimensional finite element model, which can provide low-redundancy and semantically clear input data for subsequent digital twin model training.

[0076] In step S4, historical response data of the smart cabinet is obtained, a mechanical response field of the smart cabinet is constructed using the historical response data and all response values, and a digital twin model of the smart cabinet is constructed based on a deep neural network.

[0077] In specific implementation, the historical response data of the smart cabinet is obtained based on the displacement sensor and data collector on the rolling shutter door axis of the smart cabinet, that is: the axial displacement changes, moving speed and acceleration curve of the rolling shutter door during the opening and closing process can be collected in real time through a high-sensitivity laser displacement sensor, and the collected data is periodically stored through a high-speed data collector, and then the stored data is used as the historical response data of the smart cabinet; it should be noted that the historical response data is a historical response time series data set containing dimensions such as time, displacement, and speed, which can be used as the sample basis input in subsequent mechanical response field reconstruction and digital twin model training.

[0078] In this embodiment, the mechanical response field of the smart cabinet can be constructed using the historical response data and all response values in the following manner:

[0079] Performing data completion and fitting processing on all response values based on the historical response data to obtain fitting response values;

[0080] Reconstructing and mapping the fitted response values to obtain spatial distribution characteristics of nodes of the smart cabinet in the mapping space;

[0081] A mechanical response field of the intelligent cabinet is constructed according to the spatial distribution characteristics and the fitting response values in the mapping space.

[0082] In a specific implementation, first, existing numerical approximation algorithms such as least squares fitting and high-order polynomial fitting can be used to complete and fit all response values through historical response data, and then the fitting results are used as fitted response values, wherein the fitted response value refers to the fitted response index of the node area in the neighborhood of the cluster center; secondly, the fitted response value of the node is projected into the spatial geometric structure through a projection algorithm, and the three-dimensional coordinate system in the spatial geometric structure is used as the spatial distribution feature of the node in the mapping space of the smart cabinet, wherein the node response value and the spatial position can be matched through the Euclidean distance, and the three-dimensional coordinate system in the spatial geometric structure can be obtained. The three-dimensional coordinate system is the information expressed by all three-dimensional coordinates in the spatial geometric structure, including the mapping relationship between spatial position and spatial feature; finally, the fitted response value is used as a reference value for modeling through a finite element mesh model, and a mechanical response field is constructed based on the spatial position and spatial feature expressed by the spatial distribution feature. The mechanical response field can express the spatiotemporal distribution characteristics of physical response quantities such as stress, strain, displacement energy density, etc. of each node of the smart cabinet under different loads and boundary conditions.

[0083] It should be noted that the fitting response value in this embodiment refers to the continuous response data obtained after interpolation, extrapolation and other operations are performed on the measurement data through a mathematical fitting algorithm, which can be used to complete missing data and smooth local data; the spatial distribution feature refers to the geometric position combination information of each node in the finite element model in the three-dimensional space, which is used to reflect the spatial correlation between the structural layout and the response area; the mechanical response field refers to the node-level response distribution formed by the structural model under the action of external loads and boundary conditions, including stress field, strain field, displacement field, etc., which is an important form of expression for constructing a spatial continuous field from physical measurement data.

[0084] In this embodiment, the deep neural network is a fully connected neural network, and the digital twin model is a bidirectional coupling model, which is used for mechanical response monitoring and operation status synchronization.

[0085] In specific implementation, the fully connected neural network is composed of multiple linearly connected neuron layers, including an input layer, several hidden layers and an output layer. The input layer is used to receive multi-source feature data, including spatial coordinates, historical response data, cluster centers, etc. The hidden layer uses a nonlinear activation function for feature extraction and nonlinear transformation; the output layer outputs the structural response prediction results, such as node displacement, stress, strain and other physical response quantities. The fully connected neural network can process high-dimensional and highly continuous structural response data.

[0086] In addition, the digital twin model is constructed using a bidirectional coupling mechanism, that is, it realizes bidirectional information interaction and dynamic fusion between the physical simulation model and the perception data. In the forward coupling direction (bottom-up), the real-time operating parameters and response data collected by the sensor are used as input. After processing by the fully connected neural network, it can be used to predict the physical response behavior of the cabinet structure under the current working conditions and realize online mechanical response monitoring. In the reverse coupling direction (top-down), the simulation response results generated by the finite element model can be fed back to the neural network. The weight or bias of the neural network is adjusted by model parameter correction, residual update, etc., so as to dynamically correct the digital twin model to make it consistent with the actual structural state and realize operation state synchronization. The digital twin model of the smart cabinet can be obtained by training the operation data of the smart cabinet through the above-mentioned fully connected neural network combined with the bidirectional coupling mechanism.

[0087] It should be noted that the bidirectional coupling model in this application is a physical model that integrates physical modeling and data-driven, which is used to perceive the information backflow and complementarity between data and model simulation. Through bidirectional coupling, the digital twin model can have physical interpretability; response monitoring of the mechanical feedback of the smart cabinet through the digital twin model means predicting the response behavior of each part of the structure under stress state through the digital model, such as stress concentration, deformation risk, etc., which is conducive to remote monitoring.

[0088] In step S5, the operation and maintenance instructions of the smart cabinet are determined based on the mechanical response field and the digital twin model, and then the operation and maintenance instructions are used to control the smart cabinet to perform remote operation and maintenance.

[0089] In this embodiment, the operation and maintenance instructions of the smart cabinet may be determined based on the mechanical response field and the digital twin model in the following manner:

[0090] Extracting features of the mechanical response field to obtain state response features of the smart cabinet;

[0091] Performing state prediction on the smart cabinet using the digital twin model, and then determining difference features based on the state prediction results and the state response features;

[0092] An operation and maintenance instruction for the smart cabinet is determined based on the difference characteristics.

[0093] In the specific implementation, first, physical quantities such as stress abnormal distribution area, node displacement mutation point, temperature gradient in temperature control area are extracted, and then all the extracted physical quantities are used as state response characteristics; then, the state of the smart cabinet is predicted through the digital twin model, and then a comparative analysis is performed with the state response characteristics and the prediction results, and then the difference results of the comparative analysis are used as difference features. The difference feature is used to measure the degree of deviation between the actual state and the predicted state, where the deviation calculation can be measured by the existing Euclidean distance; finally, based on the difference feature and the preset operation and maintenance strategy rule library, the operation and maintenance instructions of the smart cabinet are obtained, and the operation and maintenance instructions include temperature adjustment, opening and closing limit, fatigue warning, energy consumption optimization and other operation commands, where the preset operation and maintenance strategy rule library can be set based on the historical experience of server cabinet operation and maintenance.

[0094] It should be noted that the state response characteristics in this application represent the key physical quantity characteristics of the structural state or operating state extracted based on the current mechanical response field; the difference characteristics represent the error indicators or trend deviation characteristics obtained by comparing the predicted state and the response state; among them, the mechanical response field and the digital twin model generate operation and maintenance instructions through the data mapping interface for intelligent operation and maintenance, which can reduce response lag and risk misjudgment.

[0095] In addition, it should be noted that in the present application, the operation and maintenance instructions can be used to control the smart cabinet for remote operation and maintenance. In specific implementation, the operation and maintenance instructions can be sent to the smart cabinet control unit through the communication interface. The control unit is a computing platform built based on a programmable logic controller, which is used to parse the operation and maintenance instructions and distribute them to the target execution module. The control unit triggers corresponding execution actions based on the type of operation and maintenance instructions. For example: temperature control adjustment instructions can be used to adjust the fan speed, air conditioning power or the working status of the heating unit and rolling door, and the structural limit instruction can limit the maximum opening and closing angle or opening and closing number of the rolling door; the remote operation and maintenance in this application refers to relying on the communication network to issue control instructions to the smart cabinet through the twin system, and to monitor its status in real time and respond to adjustments in an intelligent maintenance process, which is conducive to improving the safety of server cabinet operation and maintenance.

[0096] It can be seen that in this application, intelligent operation and maintenance of server cabinets can be realized. First, by collecting the operating parameters of the intelligent cabinet and the status information of the rolling door axis, a comprehensive perception of the actual operating environment of the cabinet can be achieved, providing basic data for building physical models and response analysis, and ensuring the authenticity and integrity of the model input; and by discretizing the cabinet structure into tetrahedral units, establishing its accurate three-dimensional physical response model, forming a node-level response matrix, providing structural support for subsequent response dimensionality reduction modeling, node response extraction and clustering analysis, and ensuring the engineering physics foundation of twin modeling; secondly, by hierarchical clustering of the node information matrix, obtaining the cluster center set, and performing similarity response mapping, it can be realized. The high-dimensional node response data is effectively reduced in dimension, representative node response features are extracted, and the computational burden of modeling is reduced. At the same time, the overall response trend is retained through the similarity response mechanism, which is conducive to building a twin model that takes both accuracy and efficiency into consideration. Then, by fusing historical response data with clustering results, the time dimension modeling of the structural state is realized, and with the help of deep neural networks, a nonlinear mapping relationship between the operating state and the structural response is established to build a digital twin system with sustainable learning and prediction capabilities. Finally, by combining the predicted response results with the state synchronization results, operation and maintenance decision instructions based on the actual structural situation are generated to realize remote control and fault warning, conduct remote intelligent operation and maintenance, and improve the level of intelligent operation and maintenance.

[0097] To sum up, the technical solution adopted in this application can perform accurate twin predictions on the operating status of the smart cabinet to realize intelligent operation and maintenance of the server cabinet.

[0098] In the second embodiment, the present application provides an intelligent operation and maintenance system based on an intelligent cabinet. Figure 4 As shown in FIG, this figure is a module structure diagram of the operation and maintenance system shown in this embodiment of the present application, and the operation and maintenance system includes:

[0099] The data acquisition module 100 is used to collect the operating parameters of the intelligent cabinet and the axis status information of the intelligent cabinet rolling door;

[0100] A finite element analysis module 200 is configured to perform finite element analysis on the operating parameters and the axis state information to obtain a three-dimensional finite element model of the intelligent cabinet, and determine a node information matrix based on a discretized structure of tetrahedral elements of the intelligent cabinet;

[0101] A cluster response module 300 is configured to perform hierarchical clustering on the node information matrix to obtain a cluster center set, and then perform similarity response on the three-dimensional finite element model using the cluster center set to obtain a response value of each cluster center;

[0102] The digital twin module 400 is used to obtain historical response data of the smart cabinet, construct a mechanical response field of the smart cabinet using the historical response data and all response values, and build a digital twin model of the smart cabinet based on a deep neural network;

[0103] The remote operation and maintenance module 500 is used to determine the operation and maintenance instructions of the smart cabinet based on the mechanical response field and the digital twin model, and then control the smart cabinet to perform remote operation and maintenance according to the operation and maintenance instructions.

[0104] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems) and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 a process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0105] Those skilled in the art will appreciate that all or part of the steps in the various methods of the above embodiments can be completed by instructing related hardware through a program. The program can be stored in a computer-readable storage medium, including a read-only memory (ROM), a random access memory (RAM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), a one-time programmable read-only memory (OTPROM), an electronically erasable programmable read-only memory (EEPROM), a compact disc read-only memory (CD-ROM), or other optical disc storage, magnetic disk storage, or magnetic tape storage, or any other computer-readable medium capable of carrying or storing data.

[0106] It should also be noted that the terms "comprises," "includes," or any other variations thereof are intended to encompass non-exclusive inclusion, such that a process, method, commodity, or apparatus that includes a series of elements includes not only those elements but also other elements not explicitly listed, or includes elements inherent to such process, method, commodity, or apparatus. In the absence of further limitations, an element defined by the phrase "comprises a..." does not exclude the presence of other identical elements in the process, method, commodity, or apparatus that includes the element.

Claims

1. An intelligent operation and maintenance method based on an intelligent cabinet, characterized in that: The operation and maintenance method comprises the following steps: Collect the operating parameters of the intelligent cabinet and the axis status information of the intelligent cabinet rolling door; Performing finite element analysis on the operating parameters and the axis state information to obtain a three-dimensional finite element model of the intelligent cabinet, and determining a node information matrix based on a discretized structure of tetrahedral units of the intelligent cabinet; Performing hierarchical clustering on the node information matrix to obtain a cluster center set, and then performing similarity response on the three-dimensional finite element model through the cluster center set to obtain a response value of each cluster center; Acquire historical response data of the smart cabinet, construct a mechanical response field of the smart cabinet using the historical response data and all response values, and build a digital twin model of the smart cabinet based on a deep neural network; The operation and maintenance instructions of the smart cabinet are determined based on the mechanical response field and the digital twin model, and the operation and maintenance instructions are then used to control the smart cabinet to perform remote operation and maintenance.

2. The intelligent operation and maintenance method based on the intelligent cabinet according to claim 1, characterized in that: The operating parameters of the smart cabinet are collected through the load sensor, and the axis status information of the rolling shutter door of the smart cabinet is collected through the angle sensor.

3. The intelligent operation and maintenance method based on the intelligent cabinet according to claim 1, characterized in that: Performing finite element analysis on the operating parameters and the axis state information to obtain a three-dimensional finite element model of the intelligent cabinet specifically includes: determining boundary conditions and load conditions of the intelligent cabinet through the axis state information; Finite element solution is performed on the operating parameters, and then a three-dimensional finite element model of the intelligent cabinet is constructed based on the result of the finite element solution, the boundary conditions and the load conditions.

4. The intelligent operation and maintenance method based on the intelligent cabinet according to claim 1, characterized in that: The node information matrix determined based on the discretization structure of the intelligent cabinet tetrahedron unit specifically includes: Obtain the spatial coordinate information of the tetrahedron unit of the intelligent cabinet; Determine a discrete node set of the intelligent cabinet tetrahedron unit based on the spatial coordinate information; The discrete node sets are combined and arranged to obtain a node information matrix.

5. The intelligent operation and maintenance method based on the intelligent cabinet according to claim 1, characterized in that: The node information matrix is hierarchically clustered to obtain a cluster center set, specifically including: Normalizing the eigenvalue of each node in the node information matrix to obtain normalized node data; Determining the similarity of the node information matrix through the normalized node data; A node hierarchical clustering tree is constructed based on the similarity, and multiple cluster centers are extracted through the node hierarchical clustering tree to obtain a cluster center set.

6. The intelligent operation and maintenance method based on an intelligent cabinet according to claim 1, characterized in that: The three-dimensional finite element model is subjected to similarity response by the cluster center set, and the response value of each cluster center is obtained specifically including: Extracting the response characteristic parameters of nodes based on the three-dimensional finite element model; Performing a similarity comparison between the response feature parameters and the center features of the cluster center set to obtain a response matching degree between the node and each cluster center; A similar mapping is performed on each cluster center using the response matching degree to obtain a response value of each cluster center.

7. The intelligent operation and maintenance method based on an intelligent cabinet according to claim 1, characterized in that: The historical response data of the smart cabinet is obtained based on the displacement sensor and data collector on the rolling door shaft of the smart cabinet.

8. The intelligent operation and maintenance method based on an intelligent cabinet according to claim 1, characterized in that: The deep neural network is a fully connected neural network, and the digital twin model is a bidirectional coupling model, which is used for mechanical response monitoring and operation status synchronization.

9. The intelligent operation and maintenance method based on an intelligent cabinet according to claim 1, characterized in that: The intelligent cabinet includes: a tetrahedron unit, a rolling shutter door, a temperature control unit, a sensor unit, and a control unit.

10. An intelligent operation and maintenance system based on an intelligent cabinet, used to execute an intelligent operation and maintenance method based on an intelligent cabinet according to any one of claims 1 to 9, characterized in that: The operation and maintenance system includes: A data acquisition module is used to collect the operating parameters of the intelligent cabinet and the axis status information of the intelligent cabinet rolling door; a finite element analysis module, configured to perform finite element analysis on the operating parameters and the axis state information to obtain a three-dimensional finite element model of the intelligent cabinet, and determine a node information matrix based on a discretized structure of tetrahedral units of the intelligent cabinet; A cluster response module is used to perform hierarchical clustering on the node information matrix to obtain a cluster center set, and then perform similarity response on the three-dimensional finite element model through the cluster center set to obtain a response value of each cluster center; A digital twin module is used to obtain historical response data of the smart cabinet, construct a mechanical response field of the smart cabinet using the historical response data and all response values, and build a digital twin model of the smart cabinet based on a deep neural network; The remote operation and maintenance module is used to determine the operation and maintenance instructions of the smart cabinet based on the mechanical response field and the digital twin model, and then control the smart cabinet to perform remote operation and maintenance according to the operation and maintenance instructions.

Images