CN117252099A

CN117252099A - Damage monitoring method, system, equipment and medium based on digital twinning

Info

Publication number: CN117252099A
Application number: CN202311226095.4A
Authority: CN
Inventors: 李雄; 倪晓升; 张易东; 熊宇涵; 蒋燕梅; 吕雅丽; 秦小营
Original assignee: Sun Yat Sen University
Current assignee: Sun Yat Sen University
Priority date: 2023-09-21
Filing date: 2023-09-21
Publication date: 2023-12-19
Anticipated expiration: 2043-09-21
Also published as: CN117252099B

Abstract

The invention discloses a damage monitoring method, system, equipment and medium based on digital twinning, and relates to the technical field of transportation equipment. A sensing layer is provided in the housing of the device that is capable of identifying the housing damage and altering the signal. And establishing an initial simulation model of the sensing layer, and correcting and simulating the initial simulation model by adopting the physical damage state of the sensing layer and experimental data of signals to obtain simulation data. And constructing a digital twin model corresponding to the equipment by adopting a digital twin technology. And importing the experimental data and the simulation data into a digital twin model for model training to obtain a damage identification model. And transmitting the sensing layer signals corresponding to the shell to the damage identification model in real time to carry out shell damage identification, obtaining shell damage monitoring data and updating the digital twin model, so that the digital twin model can timely reflect the state of the shell. The device can effectively avoid incomplete and untimely manual inspection, can timely acquire the state of the shell and repair the shell, and ensures the normal operation of the device.

Description

Damage monitoring method, system, equipment and medium based on digital twinning

Technical Field

The invention relates to the technical field of transportation equipment, in particular to a damage monitoring method, system, equipment and medium based on digital twinning.

Background

Along with the improvement of production efficiency and the acceleration of life rhythm, the dependence of people on various devices and equipment is continuously increased, and the normal use of the device is greatly related to the smooth progress of production operation and life use. For some savings devices, the integrity of the housing ensures that the savings function can be used properly. Typically, some visual inspection can only identify some significant problems on the reservoir device, but missing inspection or some hidden injury can lead to leakage of the device or further serious breakage problems.

Transportation equipment such as automobiles, trains, airplanes and the like are closely related to living and production and manufacturing activities of people, and the transportation equipment is used as a barrier for the safety of people living and property when accidents or accidents occur in the high-speed environment due to the fact that the transportation equipment is outside the transportation equipment in the high-speed running process. However, the existing detection and maintenance means are aimed at maintenance and research of the drivability and comfort, neglect detection of damage to the outer shell of the vehicle, and lead people to have insufficient safety guarantee on the transportation equipment running at high speed. Although the aircraft can be subjected to the winding inspection before landing and taking off, due to the fact that the aircraft is large in size and has more curved surfaces, accurate and comprehensive maintenance is difficult to achieve. In addition, in the spacecraft, the spacecraft is required to be isolated from the severe environment of the outer space by utilizing a shell, and the integrity of the shell is a key place for ensuring the normal operation of space operation.

In the military field, the armor protection performance of a tank, an armored vehicle and other combat vehicles directly affects the combat performance of the combat vehicles on the battlefield. At present, the maintenance of the armor is mainly carried out in the maintenance and the relevant detailed detection, but the preliminary judgment can only be realized by naked eyes in daily use, so that some damages or hidden injuries on the armor can be missed, and the anti-striking capability of a war chariot is greatly reduced.

Disclosure of Invention

The invention provides a damage monitoring method, a damage monitoring system, damage monitoring equipment and damage monitoring media based on digital twinning, which solve the technical problem that the state of a shell of equipment is difficult to perceive and monitor in the prior art.

The invention provides a damage monitoring method based on digital twinning, which comprises the following steps:

when the equipment to be monitored is determined, a sensing layer capable of identifying the damage of the shell and changing the signal is arranged in the shell of the equipment;

establishing an initial simulation model of the sensing layer, and correcting and simulating the initial simulation model by adopting the physical damage state of the sensing layer and experimental data of signals to obtain simulation data;

adopting a digital twin technology to construct a digital twin model corresponding to the equipment;

Importing the experimental data and the simulation data into the digital twin model for model training to obtain a damage identification model;

and transmitting the sensing layer signals corresponding to the shell to the damage identification model in real time to carry out shell damage identification, obtaining shell damage monitoring data and updating the digital twin model.

Optionally, the step of providing a sensing layer in the housing of the device capable of identifying housing damage and altering the signal comprises:

constructing a shell of the equipment by adopting a shell layer, an interlayer and a sensing layer;

the sensing layer is attached to the inner side of the shell layer, the sensing layer can change signals according to the state of the shell layer, and the sensing layer comprises a grid-shaped wire layer;

the interlayer is positioned between the wire layer and the shell layer and is an insulator;

and a stay wire layer is arranged between the shell layer and the interlayer.

Optionally, the step of establishing an initial simulation model of the sensing layer, and correcting and performing simulation calculation on the initial simulation model by adopting experimental data of physical damage states and signals of the sensing layer to obtain simulation data includes:

converting the conductor geometric shape corresponding to the sensing layer into a finite element model;

Performing grid division on the finite element model to obtain a grid model;

updating the grid model by adopting material properties and boundary conditions corresponding to the perception layer to obtain an initial simulation model of the perception layer;

performing electric signal comparison on the initial simulation model by adopting the physical damage state of the sensing layer and experimental data of signals to obtain comparison data;

performing parameter adjustment and compensation on the initial simulation model according to the comparison data to obtain a target simulation model;

and carrying out collision simulation on the target simulation model by adopting preset collision object data, and calculating the resistance characteristics under different damage deformation conditions to obtain simulation data.

Optionally, the step of importing the experimental data and the simulation data into the digital twin model to perform model training to obtain a damage identification model includes:

constructing a data set by adopting the experimental data and the simulation data;

importing the data set into the digital twin model, and training the judgment model by adopting a naive Bayes method to obtain the judgment model;

and carrying out model training on the judging model by adopting an initial convolutional neural network to obtain a damage identification model.

Optionally, the step of training the judgment model by using a naive bayes method to obtain the judgment model includes:

dividing the data set into a first training set and a first verification set;

calculating category prior probability and feature condition probability corresponding to the first training set by adopting a naive Bayes method;

establishing model parameters by adopting the category prior probability and the characteristic conditional probability to obtain a naive Bayes model;

performing model evaluation on the naive Bayes model by adopting the first verification set to obtain an evaluation result;

and adjusting model parameters of the naive Bayes model according to the evaluation result to obtain a judgment model.

Optionally, the step of performing model training on the judgment model by using an initial convolutional neural network to obtain a damage identification model includes:

dividing the data set into a second training set, a second verification set and a test set;

inputting the second training set into an initial convolutional neural network for parameter training to obtain training data;

calculating an initial loss function between the training data and the corresponding real label;

calculating a gradient between the initial loss function and network parameters of the initial convolutional neural network by adopting a back propagation algorithm;

Based on the gradient, updating the network parameters by adopting a chain rule and an optimization algorithm in sequence to obtain a target loss function;

verifying the initial convolutional neural network corresponding to the target loss function by adopting the second verification set to obtain verification data;

verifying and adjusting the initial convolutional neural network by adopting verification data to obtain an intermediate convolutional neural network;

testing and data adjustment are carried out on the middle convolutional neural network by adopting the test set, so that a target convolutional neural network is obtained;

and updating the judging model by adopting the target convolutional neural network to obtain a damage identification model.

Optionally, the step of transmitting the sensing layer signal corresponding to the shell to the damage identification model in real time to identify the damage to the shell, obtaining shell damage monitoring data and updating the digital twin model includes:

extracting features of the sensing layer signals corresponding to the shell to obtain time domain features, frequency domain features and bit domain features;

transmitting the time domain features, the frequency domain features and the bit domain features to the damage identification model to identify the damage of the shell, so as to obtain a damage state of the shell;

And importing the shell damage state into preset simulation calculation software to perform stress analysis and fatigue life prediction, obtaining shell damage monitoring data and updating the digital twin model.

The invention also provides a damage monitoring system based on digital twinning, which comprises:

the device comprises a shell construction module, a shell detection module and a shell detection module, wherein the shell construction module is used for arranging a sensing layer capable of identifying shell damage and changing signals in a shell of equipment to be monitored when the equipment to be monitored is determined;

the simulation data obtaining module is used for establishing an initial simulation model of the sensing layer, correcting and simulating the initial simulation model by adopting the physical damage state of the sensing layer and experimental data of signals, and obtaining simulation data;

the digital twin model construction module is used for constructing a digital twin model corresponding to the equipment by adopting a digital twin technology;

the damage identification model obtaining module is used for importing the experimental data and the simulation data into the digital twin model to carry out model training so as to obtain a damage identification model;

and the shell damage monitoring data obtaining module is used for transmitting the sensing layer signals corresponding to the shell to the damage identification model in real time to carry out shell damage identification, obtaining shell damage monitoring data and updating the digital twin model.

The invention also provides an electronic device comprising a memory and a processor, wherein the memory stores a computer program, and the computer program when executed by the processor causes the processor to execute the steps of implementing the damage monitoring method based on digital twinning.

The present invention also provides a computer readable storage medium having stored thereon a computer program which when executed implements a digital twinning based damage monitoring method as in any one of the above.

From the above technical scheme, the invention has the following advantages:

according to the invention, the sensing layer capable of recognizing the damage of the shell and changing the signal is arranged in the shell of the equipment, and the output signal of the sensing layer is changed through deformation, penetration and the like of the shell, so that the sensing and signal capturing capacity of the shell is enhanced. And constructing a digital twin model according to the equipment, mapping the equipment in the actual physical space into a virtual digital space, and realizing the comprehensive grasp of the digital space to the physical space. And importing experimental data of the physical damage state and the sensing layer signal into the digital twin model to ensure that the sensing layer can accurately reflect the physical space in the digital space. By performing simulation calculation on the sensing layer, simulation data of the damage states of the sensing layer and the shell and signals of the sensing layer are obtained, and limited experimental data can be used for calibrating a model and simultaneously producing a large amount of simulation data which accords with physical space and is real and reliable. And then training the digital twin model to obtain a damage identification model, performing effective intelligent training by experimental data and simulation data, and obtaining a sensing layer signal change which can accurately reflect the shell state from a large amount of data training to realize the accurate mapping relation between the sensing layer signal and the shell state. The sensing layer signals are transmitted to the damage identification model in real time, and the state of the shell can be known in time by monitoring the sensing layer signals in real time. The damage recognition model is used for analyzing and comparing the signals of the sensing layer, so that the damage of the shell can be rapidly recognized, and the digital twin model corresponding to the shell is updated by adopting the damage state, so that the digital twin model can timely reflect the state of the shell, and subsequent simulation calculation and state evaluation are convenient. The invention can comprehensively and real-timely monitor the operation process of the equipment and timely know the state of the shell and the damage type. Through the perception layer that sets up in the casing, can realize similar to the technical effect of response nerve in the casing, can carry out accurate location and analysis to the state of casing. The method can effectively avoid incomplete and untimely manual inspection, can realize investigation of hidden injury, timely acquire the state of the shell and repair the shell, reduce loss, ensure normal operation of equipment, and solve the technical problem that the state of the shell of the equipment is difficult to perceive and monitor in the prior art.

Drawings

In order to more clearly illustrate the embodiments of the invention or the technical solutions of the prior art, the drawings which are used in the description of the embodiments or the prior art will be briefly described, it being obvious that the drawings in the description below are only some embodiments of the invention, and that other drawings can be obtained from these drawings without inventive faculty for a person skilled in the art.

Fig. 1 is a flowchart of a digital twin-based damage monitoring method according to an embodiment of the present invention;

fig. 2 is a flow chart of steps of a digital twinning-based damage monitoring method according to a second embodiment of the present invention;

fig. 3 is a schematic diagram of a housing structure according to a second embodiment of the present invention;

fig. 4 is a schematic diagram of a sensing layer structure according to a second embodiment of the present invention;

fig. 5 is a schematic diagram of a damaged state structure of a sensing layer according to a second embodiment of the present invention;

fig. 6 is a schematic view of a curved surface structure of a housing according to a second embodiment of the present invention;

fig. 7 is a schematic diagram of a door structure according to a second embodiment of the present invention;

fig. 8 is a block diagram of a damage monitoring system based on digital twinning according to a third embodiment of the present invention.

The reference numerals in fig. 3-7 are:

1. a wire layer; 2. an interlayer; 3. a shell layer; 4. and a pull wire layer.

Detailed Description

The embodiment of the invention provides a damage monitoring method, a damage monitoring system, damage monitoring equipment and damage monitoring media based on digital twinning, which are used for solving the technical problem that the state of a shell of equipment is difficult to perceive and monitor in the prior art.

In order to make the objects, features and advantages of the present invention more comprehensible, the technical solutions in the embodiments of the present invention are described in detail below with reference to the accompanying drawings, and it is apparent that the embodiments described below are only some embodiments of the present invention, but not all embodiments of the present invention. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.

Referring to fig. 1, fig. 1 is a flowchart illustrating a digital twinning-based damage monitoring method according to an embodiment of the invention.

The first embodiment of the invention provides a damage monitoring method based on digital twinning, which comprises the following steps:

s101, when the equipment to be monitored is determined, a sensing layer capable of identifying damage of the shell and changing signals is arranged in the shell of the equipment.

In the embodiments of the present invention, the housing of the apparatus refers to an outer wall of the apparatus, which serves as an outer wall of the apparatus, which serves to isolate the inside from the outside, to protect the inside conditions, such as an outer wall of some storage equipment, an outer wall of an automobile or train or aircraft or spacecraft, armor of armored equipment, etc. The shell can be provided with a layer of layered structure capable of changing the signal state along with the deformation of the shell, such as a sensing layer, and the signal change can be caused according to the deformation caused by the damage of the shell. The sensing layer may be a capacitive layer, which recognizes the deformation by a capacitance change as an electrical signal, as in the case of the currently prevailing touch screens. Further, the propagation of light in the optical fiber can be utilized, deformation can be identified by the optical signal, and generation of the optical signal under pressure deformation can be realized by optical fiber deformation resulting in an optical path difference.

Specifically, after the device to be monitored is determined, a shell of the device is constructed by adopting the shell layer 3, the interlayer 2 and the sensing layer. Wherein, the perception layer laminating shell layer 3 inboard, perception layer can be according to shell layer 3 state change signal, and perception layer includes latticed wire layer 1. The interlayer 2 is positioned between the wire layer 1 and the shell layer 3, and the interlayer 2 is an insulator. A stay wire layer 4 is arranged between the shell layer 3 and the interlayer 2.

S102, an initial simulation model of the sensing layer is established, and the initial simulation model is corrected and simulated by adopting the physical damage state of the sensing layer and experimental data of signals, so that simulation data are obtained.

In an embodiment of the invention, the conductor geometry corresponding to the sensing layer is converted into a finite element model. And carrying out grid division on the finite element model to obtain a grid model. And updating the grid model by adopting material properties and boundary conditions corresponding to the perception layer to obtain an initial simulation model of the perception layer. And carrying out electric signal comparison on the initial simulation model by adopting the physical damage state of the sensing layer and experimental data of the signals to obtain comparison data, and carrying out parameter adjustment and compensation on the initial simulation model according to the comparison data to obtain the target simulation model. And carrying out collision simulation on the target simulation model by adopting preset collision object data, and calculating the resistance characteristics under different damage deformation conditions to obtain simulation data.

The simulation calculation of the sensing layer is to calculate the resistance characteristics of the sensing layer under different damage deformation conditions. The content of the simulation calculation comprises the calculation of resistance of the sensing layer under various deformation, and further temperature can be considered. The simulation calculation is mainly physical simulation calculation of the resistor, so that corresponding current and voltage changes are generated under the condition of specific power supply input, and the simulation calculation is realized in simulation calculation or finite element analysis. The simulation data are mainly used for intelligent prediction judgment in the future. The simulation data are obtained by performing simulation calculation on the sensing layer, so that the time and cost for obtaining the data can be reduced compared with a mode of obtaining the data based on an actual physical space experiment, the simulation calculation accuracy of the resistor is high, and the reliability of the simulation data is ensured.

S103, constructing a digital twin model corresponding to the equipment by adopting a digital twin technology.

In the embodiment of the invention, when a digital twin model is constructed, a geometric model is firstly established, and the assembly connection relation of all parts in the geometric model is consistent with a physical space. For convenience of description, the physical space herein refers to the real space, and the virtual space refers to the space in which the digital twin model is located. Sometimes, in order to simplify the digital twin equipment model, a certain limit is set for the digital twin equipment model in a virtual space, so that the digital twin equipment model meets a certain production operation function, and an integrated digital twin equipment model with topological characteristics is formed. According to the established geometric model, physical attributes, such as the process requirements of some parts, constraints on connecting parts, material attribute information and the like, are added to the digital twin model in the virtual space, so that the digital twin model is more in line with equipment in the physical space. In the digital twin model, the influence of the environment can be fully considered, so that the interactivity of the environment and equipment in the digital twin model is realized.

The shell of the digital twin model fully reflects the state of the physical space, namely the outer wall of the digital twin model comprises a sensing layer for the analog calculation of the subsequent sensing layer. And then constructing the wire layer 1 in a shell of the digital twin model of the equipment, and endowing the wire layer 1 with physical parameters such as resistivity, temperature coefficient of resistance and the like, so that model data can be directly obtained from the simulation model of the wire layer 1.

S104, importing experimental data and simulation data into a digital twin model to perform model training, and obtaining a damage identification model.

In the embodiment of the invention, experimental data and simulation data are adopted to construct a data set. And importing the data set into a digital twin model, and training the judgment model by adopting a naive Bayes method to obtain the judgment model. And then, carrying out model training on the judging model by adopting an initial convolutional neural network to obtain a damage identification model.

S105, transmitting the sensing layer signals corresponding to the shell to the damage identification model in real time to carry out shell damage identification, obtaining shell damage monitoring data and updating the digital twin model.

In the embodiment of the invention, the sensing layer signals corresponding to the shell are subjected to feature extraction to obtain the time domain features, the frequency domain features and the bit domain features. And transmitting the time domain features, the frequency domain features and the bit domain features to a damage recognition model to recognize the damage of the shell, so as to obtain a damage state of the shell. And importing the shell damage state into preset simulation calculation software to perform stress analysis and fatigue life prediction, obtaining shell damage monitoring data and updating a digital twin model.

In the embodiment of the invention, after the equipment to be monitored is determined, a sensing layer capable of identifying the damage of the shell and changing the signal is arranged in the shell of the equipment. And establishing an initial simulation model of the sensing layer, and correcting and simulating the initial simulation model by adopting the physical damage state of the sensing layer and experimental data of signals to obtain simulation data. And constructing a digital twin model corresponding to the equipment by adopting a digital twin technology. And importing the experimental data and the simulation data into a digital twin model for model training to obtain a damage identification model. And transmitting the sensing layer signals corresponding to the shell to the damage identification model in real time to carry out shell damage identification, obtaining shell damage monitoring data and updating the digital twin model. The sensing layer is utilized to realize timely feedback of the change of the shell state, an accurate damage identification model is obtained through a digital twin technology, the damage identification model is utilized to identify the signal of the sensing layer and timely obtain the shell state, the real-time monitoring can be carried out on equipment in the operation process of the equipment, the shell state and the damage type are accurately judged, the situation that manual inspection is incomplete and untimely is avoided, the investigation of hidden damage can be realized, the shell state is timely known and repaired, the loss is reduced, the normal operation of the equipment is ensured, the technical problem that the effective monitoring on the shell state of the equipment cannot be realized at present is solved, and the integral comprehensive monitoring of the equipment can be ensured by combining the detection of the functional parts of the equipment. In some special operations, the damage factor of the shell can be judged through the obtained state characteristics of the shell, the type and degree of attack can be judged in the field of some armored equipment, and the coping capacity of the armored equipment in military operations is improved.

Referring to fig. 2, fig. 2 is a flowchart illustrating a digital twinning-based damage monitoring method according to a second embodiment of the present invention.

Another digital twinning-based damage monitoring method provided in the second embodiment of the present invention includes:

step 201, when determining the device to be monitored, a sensing layer capable of recognizing damage of the shell and changing signals is arranged in the shell of the device.

Further, step 201 may include the following sub-steps S11-S14:

s11, constructing a shell of the equipment by adopting the shell layer 3, the interlayer 2 and the perception layer.

S12, a sensing layer is attached to the inner side of the shell layer 3, the sensing layer can change signals according to the state of the shell layer 3, and the sensing layer comprises a grid-shaped conducting wire layer 1.

S13, the interlayer 2 is positioned between the wire layer 1 and the shell layer 3, and the interlayer 2 is an insulator.

S14, a stay wire layer 4 is arranged between the shell layer 3 and the interlayer 2.

In the embodiment of the invention, a shell layer 3, an interlayer 2 and a perception layer are adopted to construct a shell of the equipment. The perception layer laminating casing inboard, perception layer can be according to shell layer 3 state change signal. The housing is used for protecting or isolating transportation equipment and some tank bodies, and generally has certain pressure bearing and impact resistance. The shell layer 3 is a part having a shell protection or isolation function, and may be a general shell, or may be reduced in thickness while considering the thickness of the sensing layer, and the sensing layer is attached to the inner side of the shell layer 3. Inside here refers to the interior space protected or isolated with respect to the housing, the protection or isolation of the housing being focused on the protection or isolation of the interior space. The sense layer is attached to the shell layer 3 to ensure that deformation or damage of the shell layer 3 can affect the state of the sense layer, and further, the sense layer signal is changed, so that the sense and transmission of the shell state are performed.

Preferably, the sensing layer is a grid wire layer 1 structure in a grid shape, and the signal is an electric signal. The resistance state of the wire layer 1 can be changed when the shell is deformed and penetrated through by attaching the wire layer 1 of the shell, and then the electric signal of the wire layer 1 is changed, so that the real-time corresponding relation between the shell state and the signal of the wire layer 1 is ensured, a large amount of simulation data can be accurately and efficiently obtained by the simulation calculation of the resistance and the electric signal thereof, and the accuracy of the damage identification model is ensured.

The network-shaped structure formed by connecting the wires is used as a sensing layer, the shape of the structure of the grid wire layer 1 is changed through deformation of the shell, or the structure of the grid wire layer 1 is damaged, so that the change of resistance can be realized, the resistance has a great relation with the shape, the length and the connection of the wires, and the resistance of the structure of the grid wire layer 1 can be easily calculated under the condition of determining the shape and the resistivity. Through weak current in the grid wire layer 1, the resistance of the grid wire layer 1 can be obtained through current or voltage measurement, and the instant state of the structure of the grid wire layer 1 can be clear through reverse engineering. Wherein the voltage can also be further detected by means of the hall effect, and the longitudinal voltage caused by the transverse current can also be used as an effective signal.

The material of a typical wire determines the resistivity ρ of the conductor, and its specific size is also proportional to the length L of the wire and inversely proportional to the cross-sectional area S of the wire. In the classical case, the resistance can be calculated by the formula r=ρl/S. After the wires are connected into a grid, the parallel connection and the series connection of the equivalent resistance are equivalent, and the total equivalent resistance can be calculated by using a formula of parallel connection and series connection. In fig. 4, the wires are connected to each other in a net structure, and only one direction is usually in the vertical and horizontal directions in the plane, and current or voltage can be conducted in this direction, while voltage measurement can be conducted in the other direction, so that the hall effect is utilized to obtain the hall voltage which can also be used as the electric signal of the sensing layer. Specifically, an xy rectangular coordinate system can be established in the plane of the grid wires, and the power supply is turned on along the x direction, i.e. the voltage and current can be measured in the x direction, and the hall voltage can be measured in the y direction. Alternatively, the current lines may be formed in a grid shape from one or more points without restricting the current to a single direction.

In fig. 5, the shape of the sensing layer is greatly changed in this case as the housing is broken, and the equivalent resistance is also changed, so that the current and voltage signals are changed. At this time, the line passing through the broken position is suddenly changed, and other lines not passing through the broken position can be correspondingly influenced due to the parallel connection characteristic in the grid-shaped line. The sensing layer signal under the deformation is reflected, and the related signal characteristics are extracted from the sensing layer signal, so that the sensing layer signal can be fully used for judging and identifying the damage of the shell.

As shown in fig. 3, the housing further comprises a barrier layer 2 between the sensing layer and the shell layer 3, the barrier layer 2 being an insulator. For the shell layer 3, most of the material is metal, and the interlayer 2 can prevent the current of the sensing layer from entering the shell layer 3, so that the signal of the sensing layer is inaccurate, and the normal use of the shell is influenced. When the separator 2 is thin, the sensing layer is closely attached to the shell layer 3 via the separator 2, and when the separator 2 is thick, the separator 2 is considered to have already exerted part of the function of the shell layer 3, and the sensing layer is closely attached to the separator 2.

Preferably, a pull wire layer 4 is further arranged between the shell layer 3 and the interlayer 2, the pull wire layer 4 is composed of a plurality of tensioning wires, and each tensioning wire is provided with a corresponding pull force sensor, and the pull force sensors transmit data to the digital twin model to obtain the stress of the shell. The plurality of wires are not staggered and can be arranged into parallel pull wires, and two ends or one end of each pull wire is connected with the pull force sensor. Wherein the pull wire is embedded in the housing and is close to the housing, and the pull sensor is not embedded in the inside of the housing and is located at both side positions of the housing. The interference of metals on signals can be effectively avoided through the design of the stay wire layer 4, the problem that the capacitance and the capacitance sensor are applied to metal devices is avoided, and the sensor is not required to be embedded into a shell through stay wires, so that the replacement and maintenance in the future are greatly facilitated. When the shell of the shell is subjected to elastic deformation or smaller under pressure and does not trigger the sensing layer electric signal, the sensing layer electric signal acts on the stay wire, so that the tightness degree of the stay wire is changed, the tension sensor directly displays the change of the stay wire in a numerical value, and the stressed shell can be judged.

As shown in fig. 6, in the case with curved surfaces, the sensing layers have specific signal characteristics, so that the distinction of curved surfaces can be effectively achieved. On this basis, the deformation or breakage features on the curved surface will have more prominent signal features on the sensing layer. And under the condition that the inner and outer orientations of the damaged deformation cannot be distinguished on the plane, the sensing layer of the curved surface shell can distinguish the inner and outer orientations, so that the sensing layer has outstanding signal characteristics under different shell states.

Specifically, the sensing layer is used on the body of the vehicle, the body is a shell at the place covered by the plate, and the sensing layers closely attached to the shell layer 3 are arranged inside the body, and are an example of a vehicle door in fig. 7. In the running process of the automobile, when a scratch accident occurs, the sensing layer timely feeds back the damage condition of the automobile body to the digital twin model through signals, and the digital twin model is fed back vividly and intuitively. The protection capability of the vehicle body can be calculated, so that the following driving and maintenance can be planned. Particularly for vehicles with bulletproof requirements, the perception layer can timely acquire the vehicle damage information under the condition of avoiding passenger getting off and checking, so that risk judgment and further strategy analysis are facilitated.

Step 202, an initial simulation model of the sensing layer is established, and the initial simulation model is corrected and simulated by adopting the physical damage state of the sensing layer and experimental data of signals, so as to obtain simulation data.

Further, step 202 may comprise the following sub-steps S21-S26:

s21, converting the conductor geometric shape corresponding to the sensing layer into a finite element model.

S22, carrying out grid division on the finite element model to obtain a grid model.

S23, updating the grid model by adopting material properties and boundary conditions corresponding to the sensing layer to obtain an initial simulation model of the sensing layer.

S24, performing electric signal comparison on the initial simulation model by adopting the physical damage state of the sensing layer and experimental data of the signals to obtain comparison data.

S25, performing parameter adjustment and compensation on the initial simulation model according to the comparison data to obtain a target simulation model.

S26, carrying out collision simulation on the target simulation model by adopting preset collision object data, and calculating resistance characteristics under different damage deformation conditions to obtain simulation data.

In the embodiment of the invention, when the initial simulation model of the sensing layer, namely the wire layer 1 is constructed, the conductor geometry, the material property, the boundary condition and the like which need to be subjected to resistance simulation calculation are determined. The conductor geometry is converted to a finite element model using Computer Aided Design (CAD) software or mesh generation software. Finite element models describe the geometry and structure of conductors in discrete nodes and cells.

And performing grid division on the finite element model, and dividing the geometric shape of the conductor into small discrete units to obtain a grid model. Proper meshing is important for accurate resistance calculation. According to the resistivity and other properties of the conductor material, appropriate material characteristics are defined for different grid cells, and an initial simulation model of the wire layer 1 is obtained. These properties will be used in the solution process to calculate the resistance. The resistance problem is converted into a solution problem of the system of equations by selecting an appropriate finite element solver and setting appropriate boundary conditions. The boundary conditions may relate to input current, fixed voltage or fixed potential, etc. And (3) carrying out finite element solution by using numerical calculation software, and obtaining a numerical solution of the conductor resistance by solving an equation set. The solution process will take into account conductor material properties, geometry, boundary conditions, etc. Post-processing and analysis are performed on the calculated resistance results, such as drawing a current density profile, calculating the total resistance of the conductor, and the like. The conductor layer 1 is arranged on the conductor layer 1 in accordance with the actual situation, and the specific required electric signal can be obtained.

In physical space, physical damage to the housing not only causes deformation of the wire layer 1, but also causes further changes in the signal of the wire layer 1, such as further heat generation and air flow changes, which emphasizes the interaction of the device and environment that needs to be considered in the simulation calculation. After experimental data are imported, the physical damage state of the wire layer 1 in the wire layer 1 simulation calculation model is calculated in a simulation mode, and the shape and signal compensation of the wire layer 1 are adjusted for the second time, so that the wire layer 1 fully accords with the actual state.

In order to ensure the simulation reliability of the wire layer 1, an actual physical test needs to be performed on the wire layer 1 to obtain electrical signals of the wire layer 1 under different deformation and damage. The same model is built on the simulation model of the wire layer 1, namely the initial simulation model, electric signal comparison is carried out, the initial simulation model is continuously corrected, and intelligent algorithm parameter adjustment and compensation are introduced when necessary, so that the initial simulation model is fully consistent with the actual model, and the target simulation model is obtained. Because the related simulation calculation of the resistor has higher stability and reliability, a large amount of simulation data can be obtained through the target simulation model.

The simulation data are obtained by deforming or damaging the target simulation model to different degrees and then performing simulation calculation. The deformation or damage of different degrees to the target simulation model can be realized through collision simulation, so that the simulation data of the target simulation model accords with the actual situation. The object, its initial position, velocity, mass, etc. properties to be subjected to the collision simulation are determined, and environmental parameters such as gravitational acceleration, air resistance, etc. are set. And establishing a mathematical model corresponding to the actual object according to the physical characteristics of the geometric shape, the mass, the friction and the like of the object. Common models include newton's mechanical models, elastic collision models, and the like. Objects and environments are divided into discrete fractions, for example using a grid or a particle system. Thus, continuous problems such as movement, deformation and the like of the object can be converted into discrete problems. The motion of the object is solved numerically using numerical methods such as the euler method or the longgrid-kuta method. The position, speed, force, etc. of the object at each time step are calculated based on the physical model and constraints. In each time step, collisions between objects are detected. Common collision detection algorithms include bounding box detection, separation axiom, and the like. If collision between objects is found, the next step is carried out. And calculating the rebound speed, torque, deformation and other changes of the object after collision according to the collision model. Common collision models are full elastic collisions, partial elastic collisions, and the like. And updating the position, speed and other state information of the object according to the collision response result and the numerical solution result. And repeatedly executing the self-consistent calculation until the simulated time process is finished or the required simulation precision is reached. And carrying out post-processing and analysis on the result of the collision simulation. The method can draw graphs of object motion trail, energy loss, deformation condition and the like, and perform result verification and optimization. Thus, target simulation models under different deformation or damage conditions are obtained, and simulation calculation of the electric signals is further carried out, so that simulation data are obtained.

And 203, constructing a digital twin model corresponding to the equipment by adopting a digital twin technology.

In the embodiment of the present invention, the specific implementation process of step 203 is similar to that of step 103, and will not be described herein.

And 204, constructing a data set by adopting experimental data and simulation data.

In an embodiment of the invention, a dataset for training is collected and consolidated, the dataset comprising experimental data and simulation data for the perception layer.

Step 205, importing the data set into a digital twin model, and training the judgment model by adopting a naive Bayes method to obtain the judgment model.

Further, step 205 may include the following substeps S31-S35:

s31, data division is carried out on the data set, and a first training set and a first verification set are obtained.

S32, calculating the category prior probability and the feature condition probability corresponding to the first training set by adopting a naive Bayes method.

And S33, establishing model parameters by adopting category prior probability and feature conditional probability to obtain a naive Bayes model.

And S34, performing model evaluation on the naive Bayes model by adopting the first verification set to obtain an evaluation result.

And S35, adjusting model parameters of the naive Bayes model according to the evaluation result to obtain a judgment model.

In the embodiment of the invention, when model training is performed, each crossed node of the wire layer 1 is marked, and the existence state of the node is judged by using a judgment model. Here, the judgment model training is performed using a naive bayes method, first, a data set for training including experimental data and simulation data of the wire layer 1 is collected and collated. For the judgment model, the input characteristics and the corresponding class labels are needed, wherein the characteristics are characteristics such as signal amplitude and the like, and the classes are time domain characteristics, frequency domain characteristics and bit domain characteristics. According to task requirements and data characteristics, proper characteristics are selected and converted into a form suitable for naive Bayesian hypothesis. The data set is divided into a first training set and a first validation set. The first training set is used for parameter estimation of the model, and the first validation set is used for adjusting the super-parameters of the model and evaluating the performance of the model.

Based on the naive bayes assumption, a class prior probability and feature conditional probabilities of features in each class can be calculated from the samples in the first training set. Specifically, the probability of each feature in each category is calculated from samples of different categories in the first training set. Parameters of the model are calculated and estimated according to a naive bayes formula using the samples in the first training set. And establishing parameters of the model by calculating the category prior probability and the feature condition probability. The first validation set or other evaluation index is used to evaluate the performance of the trained naive bayes model, such as the evaluation index of accuracy, precision, recall, F1 value, etc. And according to the evaluation result, adjusting the hyper-parameters of the model to optimize the performance of the model, and obtaining the judgment model. For naive bayes models, common hyper-parameters include smoothing parameters (e.g., laplace smoothing), etc. The judgment model obtained through the steps can be used for judging a new sample, namely, the existence state of each node can be known by inputting the electric signal of the perception layer.

And 206, performing model training on the judgment model by adopting an initial convolutional neural network to obtain a damage identification model.

Further, step 206 may include the following substeps S41-S49:

s41, data division is carried out on the data set, and a second training set, a second verification set and a test set are obtained.

S42, inputting the second training set into the initial convolutional neural network to perform parameter training, and obtaining training data.

S43, calculating an initial loss function between the training data and the corresponding real label.

S44, calculating the gradient between the initial loss function and the network parameters of the initial convolutional neural network by adopting a back propagation algorithm.

S45, updating network parameters by adopting a chain rule and an optimization algorithm in sequence based on the gradient to obtain a target loss function.

S46, verifying the initial convolutional neural network corresponding to the target loss function by adopting the second verification set to obtain verification data.

And S47, verifying and adjusting the initial convolutional neural network by adopting verification data to obtain an intermediate convolutional neural network.

S48, testing and data adjustment are carried out on the middle convolutional neural network by adopting the testing set, and the target convolutional neural network is obtained.

And S49, updating the judgment model by using the target convolutional neural network to obtain a damage identification model.

In the embodiment of the invention, the existence state of each node is used as a pixel point, and a convolutional neural network can be used for further model training so as to perform damage identification according to the node state. The structure of the convolutional neural network comprises a plurality of convolutional layers, a pooling layer, a full-connection layer and the like. The data set is divided into a second training set, a second validation set and a test set. The second training set is used for parameter training of the model, the second verification set is used for adjusting super parameters of the model and evaluating performance of the model, and the test set is used for performance evaluation of the final model. The weights and offsets of the convolutional neural network are initialized. And inputting the image data in the second training set into an initial convolutional neural network, and performing forward propagation through operations such as convolution, activation function, pooling and the like to obtain an output result, namely training data. Based on the training data and the real labels, the value of the loss function (such as cross entropy loss) is calculated to obtain an initial loss function. The initial loss function measures the difference between the model output and the real label. The gradient of the initial loss function to the network parameters is calculated using a back propagation algorithm. The gradient is transferred from the latter layer to the former layer by the chain law, updating the network parameters to minimize the loss function. An optimization algorithm (e.g., random gradient descent) is used to update the network parameters. And the optimization algorithm adjusts the value of the network parameter according to the direction and the size of the gradient, so that the loss function is gradually reduced, and the target loss function is obtained. And adopting a second verification set to verify the initial convolutional neural network corresponding to the target loss function to obtain verification data. And according to the performance of the verification data, adjusting the super parameters of the initial convolutional neural network, such as the learning rate, the regularization coefficient, the convolutional kernel size and the like, so as to improve the generalization capability and performance of the model. And repeating the processes of forward propagation, loss calculation, back propagation and parameter optimization until a preset stopping condition is reached (such as maximum iteration number or loss function convergence) to obtain the intermediate convolutional neural network. And performing performance evaluation on the trained middle convolutional neural network by using the test set, and performing data adjustment to obtain the target convolutional neural network. The evaluation index comprises accuracy, precision, recall, F1 value and the like. And updating the judgment model by using the target convolutional neural network to obtain a damage identification model, wherein the damage identification model obtained through the steps can identify the shell damage through the node state.

The generated damage identification model can give the state of the equipment shell according to the input perception layer signal, generally can give the shell state result and can analyze the damage source at the same time, thereby helping related personnel to discover problems in advance and take measures to cope with the problems so as to reduce or reduce loss and casualties.

Step 207, transmitting the sensing layer signals corresponding to the shell to the damage recognition model in real time to recognize the damage of the shell, obtaining the shell damage monitoring data and updating the digital twin model.

Further, step 207 may include the following substeps S51-S53:

and S51, carrying out feature extraction on the sensing layer signals corresponding to the shell to obtain time domain features, frequency domain features and bit domain features.

S52, transmitting the time domain features, the frequency domain features and the bit domain features to a damage recognition model to recognize the damage of the shell, and obtaining a damage state of the shell.

And S53, importing the shell damage state into preset simulation calculation software to perform stress analysis and fatigue life prediction, obtaining shell damage monitoring data and updating the digital twin model.

In the embodiment of the invention, the digital twin model is updated in real time during the running process of the equipment, and part of the running state of the digital twin model can be directly acquired, such as temperature, speed and power consumption. The shell state is fed back through the signal of the sensing layer, and the instant state of the shell can be timely and comprehensively obtained through the signal feedback of the sensing layer. In this process, the sense layer signal is typically transmitted into the digital twin model via a wireless network link.

And identifying and comparing the signal characteristics of the sensing layer through a damage identification model in the digital twin model, thereby obtaining the instant state of the shell. The perceptual layer signal here may be processed and feature extracted in a digital twin model. The signal characteristic processing of the perception layer can be divided into time domain characteristics, frequency domain characteristics and bit domain characteristics, and when the model is trained, the three characteristics are used as input data to carry out model training. The time domain characteristic refers to the time varying nature of the signal. It describes the amplitude, waveform and timing information of the signal at different points in time. Common time domain features include: average value: the average amplitude of the signal over a period of time. Variance: the degree of dispersion of the signal amplitude is measured. Peak value: maximum amplitude of the signal. Peak-to-peak: the difference between the maximum and minimum values of the signal. Pulse width: the duration of the signal pulse. Autocorrelation function: describing the similarity between the signal and its delayed versions at different instants. The frequency domain characteristic refers to the distribution characteristic of the signal over frequency. By converting the signal into the frequency domain, the energy distribution of the signal at different frequencies can be displayed for analysis of the frequency content and spectral characteristics of the signal. Common frequency domain features include: fourier transform: the time domain signal is converted to the frequency domain, showing the energy distribution of the signal at different frequencies. Spectral density: the energy density of the signal at different frequencies is described. The main frequency: the fundamental frequency component of the signal. Harmonic components: an integer multiple of the dominant frequency component in the signal. Frequency bandwidth: bandwidth range of the signal in the frequency domain. The bit domain features refer to the change characteristics of the signals in position, and the electrical signal features are related with the positions, so that coordinate position information can be simply attached to the signal features, and the method can be used for quickly positioning abnormal signals.

Preferably, the feature extraction is carried out on the perception layer signal before the perception layer signal enters the digital twin model, and the processed time domain feature, frequency domain feature and bit domain feature are transmitted into the digital twin model, so that the data transmission quantity can be effectively reduced, and the updating efficiency of the digital twin model is improved.

After the shell damage is identified through the damage identification model, the instant shell protection capability is calculated according to the shell damage state. After the state of the shell is obtained through the sensing layer signal, the digital twin model carries out corresponding state updating, and physical layer simulation calculation can be carried out through a device or the shell in the digital twin model. Because the digital twin model can realize physical mapping of physical space, including structure, materials and more physical properties, the damage state (crack, damage, fatigue damage and the like) of the shell can be obtained through the damage identification model, the damage state of the shell is led into preset simulation calculation software for stress analysis and fatigue life prediction, specific stress condition and damage degree, namely shell damage monitoring data, can be obtained, and further the protection capability (such as compression resistance, impact resistance, life and the like) of the shell is evaluated.

Further, the method also comprises the step of recording the factors caused by the shell damage, recording the signal characteristics of the sensing layer under different damage factors, training, identifying the signals of the real-time sensing layer, and judging the factors caused by the shell damage. While the sensing layer signal is recorded in the whole flow, the characteristics of the sensing layer signal and the shell damage factors are recorded, for example, the forming time and the shell damage states caused by different types of impact and fatigue damage can be greatly distinguished, and the specific shell damage factors can be obtained by effectively distinguishing the time domain characteristics, the frequency domain characteristics and the bit domain characteristics of the sensing layer signal. The time of deformation or damage formation can be obtained from the time domain features, the degree of deformation or damage can be obtained from the frequency domain features, and the position of deformation or damage can be obtained from the position domain features, so that the omnibearing information of deformation or damage of the shell is obtained.

During experiments and simulation calculation, the deformation or damage of the shell is classified, namely deformation, puncture, fracture and the like, and corresponds to the deformation or damage of the corresponding sensing layer. Thus, after the deformation or damage is identified by the damage identification model, the damage identification model can be corresponding to the caused factors. The shell protection capability can be obtained by quantifying the influence of deformation or damage on the rigidity of the shell. On armored equipment, one of the characteristics of the protective force is anti-striking capability, different attacks leave different striking marks and damage effects on the shell, and the striking type and the striking degree can be identified by effectively distinguishing the signals of the sensing layers. Further, according to the shell damage, the impact direction can be effectively judged, and then the combat requirement is matched.

In the embodiment of the invention, the sensing layer capable of identifying the damage of the shell and changing the signal is arranged in the shell of the equipment, and the output signal of the sensing layer is changed through the deformation, penetration and the like of the shell, so that the sensing and signal capturing capacity of the shell is enhanced. And constructing a digital twin model according to the equipment, mapping the equipment in the actual physical space into a virtual digital space, and realizing the comprehensive grasp of the digital space to the physical space. And importing experimental data of the physical damage state and the sensing layer signal into the digital twin model to ensure that the sensing layer can accurately reflect the physical space in the digital space. The sensing layer is a wire layer 1, simulation data of the damage state of the wire layer 1 and the shell and signals of the sensing layer are obtained by performing simulation calculation on the wire layer 1, and limited experimental data can be used for calibrating a model and simultaneously producing a large amount of simulation data which accords with a physical space and is real and reliable. And then training the digital twin model to obtain a damage identification model, performing effective intelligent training by experimental data and simulation data, and obtaining a sensing layer signal change which can accurately reflect the shell state from a large amount of data training to realize the accurate mapping relation between the sensing layer signal and the shell state. The sensing layer signals are transmitted to the digital twin model in real time, and the state of the shell can be known in time by monitoring the sensing layer signals in real time. The damage recognition model is used for analyzing and comparing the signals of the sensing layer, so that the damage of the shell can be rapidly recognized, and the damage state is updated into the digital twin model corresponding to the shell, so that the digital twin model can timely reflect the state of the shell, and subsequent simulation calculation and state evaluation are convenient to perform.

Referring to fig. 8, fig. 8 is a block diagram illustrating a digital twinning-based damage monitoring system according to a third embodiment of the present invention.

The third embodiment of the present invention provides a damage monitoring system based on digital twinning, which comprises:

the housing construction module 801 is configured to provide a sensing layer in a housing of the device, which is capable of recognizing a housing damage and changing a signal, when the device to be monitored is determined.

The simulation data obtaining module 802 is configured to establish an initial simulation model of the sensing layer, and correct and simulate the initial simulation model by using the physical damage state of the sensing layer and experimental data of the signal to obtain simulation data.

The digital twin model construction module 803 is configured to construct a digital twin model corresponding to the device by using a digital twin technology.

The damage recognition model obtaining module 804 is configured to import experimental data and simulation data into the digital twin model for model training, so as to obtain a damage recognition model.

The shell damage monitoring data obtaining module 805 is configured to transmit the sensing layer signal corresponding to the shell to the damage identification model in real time to identify the shell damage, obtain shell damage monitoring data, and update the digital twin model.

In the embodiment of the invention, the equipment to be monitored is transportation equipment, such as vehicles, war chariot or airplanes, and the like, and generally has a certain degree of protection or isolation capability, so that goods or personnel in the inner cavity are protected. When the device to be monitored is determined by the housing construction module, a sensing layer formed by the structure of the grid conductor layer 1 is arranged in the housing of the device, and the sensing layer can change along with the deformation of the housing, so that the resistance of the sensing layer changes. The change condition of the sensing layer can be known by collecting the electric signals of the point positions of the sensing layer. The digital twin model construction module adopts a digital twin technology to construct a digital twin model corresponding to the equipment, and can be realized by means of digital twin kits in matlab, ansys and other software. Typically, the digital twin model includes a sensing layer, resulting in a digital twin model that is consistent with the physical model. The damage identification model obtaining module is used for importing experimental data of the sensing layer signals and the shell state into the simulation data obtaining module, wherein the simulation data obtaining module also comprises specific size, shape, structure, materials and the like of the sensing layer and the shell. The simulation data obtaining module is used for simulating and calculating the sensing layer, the resistance under the shape change of different sensing layers can be obtained through finite element analysis and calculation, the corresponding electric signals are obtained through calculation, and the sensing layer state under the corresponding electric signals, namely the shell state, can be obtained through recording. As the simulation data is the resistance calculation of the sensing layers with different deformations, the simulation calculation technology is mature, and the reliability of the simulation data can be effectively ensured. The experimental data and the simulation data are imported into a digital twin model for training to obtain a damage identification model, and methods such as linear regression, polynomial regression, support Vector Machine (SVM) and the like can be used. The damage identification model can realize rapid characteristic identification of signals according to the electric signals transmitted by the real-time module, and accurate prediction and judgment of the electric signals of the sensing layer can be realized. The sensing layer electric signal can be transmitted to the shell damage monitoring data obtaining module in time through wireless transmission by the Wi-Fi module or the embedded wireless chip. The shell damage monitoring data obtaining module calculates by utilizing a damage identification model according to the electric signals of the sensing layer, updates the state in the digital twin model and displays the damage state of the shell in real time.

The simulation data obtaining module and the damage recognition model obtaining module can be nested in the digital twin model, but the digital twin model building module, the damage recognition model obtaining module and the simulation data obtaining module are arranged independently of the digital twin model in general, because the former modules complete the functions after obtaining the stable damage recognition model in general.

In another embodiment, the method can be omitted from the digital twin model under the condition that the sensing layer has little influence on the protection capability of the shell, so that the calculated amount of the digital twin model is reduced. The simulation calculation, training, real-time and calculation related to the perception layer are arranged outside the digital twin model, and then the shell state is transmitted into the digital twin model for updating according to the calculation result, so that the calculation amount in the digital twin model can be reduced, and the calculation of other aspects in the digital twin model is not influenced.

Optionally, the housing construction module 801 may perform the steps of:

a shell layer 3, an interlayer 2 and a sensing layer are adopted to construct a shell of the equipment;

the sensing layer is attached to the inner side of the shell layer 3, the sensing layer can change signals according to the state of the shell layer 3, and the sensing layer comprises a grid-shaped wire layer 1;

The interlayer 2 is positioned between the wire layer 1 and the shell layer 3, and the interlayer 2 is an insulator;

a stay wire layer 4 is arranged between the shell layer 3 and the interlayer 2.

Optionally, the simulation data obtaining module 802 includes:

the finite element model obtaining module is used for converting the conductor geometric shape corresponding to the sensing layer into a finite element model.

The grid model obtaining module is used for carrying out grid division on the finite element model to obtain a grid model.

And the initial simulation model obtaining module is used for updating the grid model by adopting the material attribute and the boundary condition corresponding to the perception layer to obtain the initial simulation model of the perception layer.

And the comparison data obtaining module is used for carrying out electric signal comparison on the initial simulation model by adopting the physical damage state of the sensing layer and experimental data of the signals to obtain comparison data.

The target simulation model obtaining module is used for carrying out parameter adjustment and compensation on the initial simulation model according to the comparison data to obtain the target simulation model.

The simulation data obtaining sub-module is used for carrying out collision simulation on the target simulation model by adopting preset collision object data, calculating resistance characteristics under different damage deformation conditions and obtaining simulation data.

Optionally, the damage recognition model obtaining module 804 includes:

And the data set construction module is used for constructing a data set by adopting experimental data and simulation data.

The judging model obtaining module is used for importing the data set into the digital twin model, and carrying out judging model training by adopting a naive Bayesian method to obtain the judging model.

The damage recognition model obtaining submodule is used for carrying out model training on the judgment model by adopting the initial convolutional neural network to obtain a damage recognition model.

Alternatively, the judgment model obtaining module may perform the steps of:

dividing the data set into a first training set and a first verification set;

model parameters are established by adopting category prior probability and feature condition probability, and a naive Bayes model is obtained;

performing model evaluation on the naive Bayes model by adopting a first verification set to obtain an evaluation result;

Alternatively, the damage-recognition-model-deriving sub-module may perform the steps of:

calculating an initial loss function between training data and a corresponding real label;

calculating the gradient between the initial loss function and the network parameters of the initial convolutional neural network by adopting a back propagation algorithm;

based on the gradient, updating network parameters by adopting a chain rule and an optimization algorithm in sequence to obtain a target loss function;

adopting a second verification set to verify the initial convolutional neural network corresponding to the target loss function to obtain verification data;

verifying and adjusting data of the initial convolutional neural network by adopting verification data to obtain an intermediate convolutional neural network;

testing and data adjustment are carried out on the intermediate convolutional neural network by adopting a test set, so that a target convolutional neural network is obtained;

and updating the judgment model by using the target convolutional neural network to obtain the damage identification model.

Optionally, the shell damage monitoring data obtaining module 805 includes:

and the feature extraction module is used for carrying out feature extraction on the sensing layer signals corresponding to the shell to obtain time domain features, frequency domain features and bit domain features.

The shell damage state obtaining module is used for transmitting the time domain features, the frequency domain features and the bit domain features to the damage recognition model to carry out shell damage recognition, so as to obtain the shell damage state.

The shell damage monitoring data obtaining submodule is used for guiding the shell damage state into preset simulation calculation software to conduct stress analysis and fatigue life prediction, obtaining shell damage monitoring data and updating a digital twin model.

The embodiment of the invention also provides electronic equipment, which comprises: a memory and a processor, the memory storing a computer program; the computer program, when executed by a processor, causes the processor to perform a digital twinning-based damage monitoring method as in any of the embodiments described above.

The memory may be an electronic memory such as a flash memory, an EEPROM (electrically erasable programmable read only memory), an EPROM, a hard disk, or a ROM. The memory has memory space for program code to perform any of the method steps described above. For example, the memory space for the program code may include individual program code for implementing the various steps in the above method, respectively. The program code can be read from or written to one or more computer program products. These computer program products comprise a program code carrier such as a hard disk, a Compact Disc (CD), a memory card or a floppy disk. The program code may be compressed, for example, in a suitable form. The code, when executed by a computing processing device, causes the computing processing device to perform the steps in the digital twinning-based impairment monitoring method described above.

The embodiment of the invention also provides a computer readable storage medium, on which a computer program is stored, which when executed by a processor implements the damage monitoring method based on digital twinning according to any of the embodiments above.

It will be clear to those skilled in the art that, for convenience and brevity of description, specific working procedures of the above-described systems, apparatuses and units may refer to corresponding procedures in the foregoing method embodiments, which are not repeated herein.

In the several embodiments provided in this application, it should be understood that the disclosed systems, apparatuses, and methods may be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative, e.g., the division of elements is merely a logical functional division, and there may be additional divisions of actual implementation, e.g., multiple elements or components may be combined or integrated into another system, or some features may be omitted, or not performed. Alternatively, the coupling or direct coupling or communication connection shown or discussed with each other may be an indirect coupling or communication connection via some interfaces, devices or units, which may be in electrical, mechanical or other form.

The units described as separate units may or may not be physically separate, and units shown as units may or may not be physical units, may be located in one place, or may be distributed over a plurality of network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

In addition, each functional unit in the embodiments of the present invention may be integrated in one processing unit, or each unit may exist alone physically, or two or more units may be integrated in one unit. The integrated units may be implemented in hardware or in software functional units.

The integrated units, if implemented in the form of software functional units and sold or used as stand-alone products, may be stored in a computer readable storage medium. Based on such understanding, the technical solution of the present invention may be embodied in essence or a part contributing to the prior art or all or part of the technical solution in the form of a software product stored in a storage medium, including several instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the methods of the embodiments of the present invention. And the aforementioned storage medium includes: a U-disk, a removable hard disk, a Read-Only Memory (ROM), a random access Memory (RAM, random Access Memory), a magnetic disk, or an optical disk, or other various media capable of storing program codes.

The above embodiments are only for illustrating the technical solution of the present invention, and are not limiting; although the invention has been described in detail with reference to the foregoing embodiments, it will be understood by those of ordinary skill in the art that: the technical scheme described in the foregoing embodiments can be modified or some technical features thereof can be replaced by equivalents; such modifications and substitutions do not depart from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A digital twinning-based damage monitoring method, comprising:

2. The digital twinning-based impairment monitoring method of claim 1, wherein the step of providing a perception layer in the housing of the device capable of identifying housing impairment and altering signals comprises:

and a stay wire layer is arranged between the shell layer and the interlayer.

3. The method for monitoring damage based on digital twinning of claim 1, wherein the step of establishing an initial simulation model of the sensing layer, correcting and simulating the initial simulation model by using experimental data of physical damage states and signals of the sensing layer, and obtaining simulation data comprises the following steps:

performing grid division on the finite element model to obtain a grid model;

4. The method for monitoring damage based on digital twinning according to claim 1, wherein the step of importing the experimental data and the simulation data into the digital twinning model to perform model training to obtain a damage identification model comprises the steps of:

5. The method for digital twin based injury monitoring according to claim 4, wherein the step of training the judgment model by using a naive bayes method to obtain the judgment model comprises:

dividing the data set into a first training set and a first verification set;

6. The method for monitoring damage based on digital twinning as recited in claim 4, wherein the step of model training the judgment model by using an initial convolutional neural network to obtain a damage recognition model comprises:

7. The method for monitoring damage based on digital twinning according to claim 1, wherein the step of transmitting the sensing layer signal corresponding to the shell to the damage identification model in real time to perform shell damage identification, obtaining shell damage monitoring data and updating the digital twinning model comprises the steps of:

8. A digital twinning-based damage monitoring system, comprising:

9. An electronic device comprising a memory and a processor, the memory having stored therein a computer program which, when executed by the processor, causes the processor to perform the steps of the digital twinning-based impairment monitoring method according to any one of claims 1 to 7.

10. A computer readable storage medium having stored thereon a computer program, wherein the computer program when executed implements the digital twinning-based impairment monitoring method according to any one of claims 1 to 7.