A field bus virtual diagnosis method based on digital twinning and related device

Abstract

The application provides a field bus virtual diagnosis method based on digital twinning and related devices, including the following steps: feature extraction is performed on collected multi-physical field parameters, and clock synchronization is performed to obtain a processed feature vector; real-time anomaly detection is performed on the processed feature vector, wherein: when an anomaly is detected, based on the processed feature vector, the boundary conditions of a multi-physical field coupling digital twinning model are updated in real time, the model is driven to perform multi-physical field coupling simulation; a fault type is configured through a virtual fault injection engine, and parameters of the multi-physical field coupling digital twinning model are modified to safely inject electrical faults, mechanical faults and / or protocol layer faults, and simulate fault response data. The application solves the technical problem that field bus faults are difficult to reproduce and verify in a real environment, the fault simulation accuracy is above 92%, and a safe, low-cost virtual experiment platform is provided for new diagnosis algorithm research and development and maintenance personnel training.

Description

Technical Field

[0001] This invention belongs to the field of industrial automation control technology, and in particular relates to a fieldbus virtual diagnostic method and related devices based on digital twins. Background Technology

[0002] Industrial fieldbuses, as high-speed digital communication links connecting field devices and control systems, are the nerve center of modern intelligent manufacturing systems. With the rapid development of Industry 4.0 and intelligent manufacturing technologies, fieldbus systems based on protocols such as Profibus, Profinet, and EtherCAT are widely used in process industries such as thermal power plants, petrochemicals, and metallurgical manufacturing. Their communication reliability directly affects production safety and efficiency. However, fieldbuses face technical challenges in actual operation, including difficulty in capturing intermittent faults, unclear mechanisms of complex and correlated faults, and difficulties in fault reproduction.

[0003] In existing technologies, two main methods are used for fieldbus fault diagnosis: One approach is the hardware-in-the-loop (HIL) physical fault injection method, which simulates faults by introducing electrical interference or physical disconnection into the real bus. This method carries the risk of damaging physical devices and is difficult to simulate multi-physics coupling faults under extreme operating conditions. The second method is a virtual diagnostic method based on pure software simulation. This method simulates the protocol layer by establishing a simplified bus communication model. However, this method cannot accurately reflect the coupling relationship between the electrical characteristics, mechanical vibration characteristics, and thermal characteristics of the physical layer, resulting in a significant deviation between the simulation results and the actual physical phenomena.

[0004] Third, existing digital twin technologies are mostly applied to lumped-parameter electrical equipment such as motors and transformers, or to network layer routing optimization. While simulation analyses exist for general transmission line losses (skin effect, dielectric loss), there is a lack of multi-physics real-time coupling modeling methods that combine the distributed parameter characteristics of transmission lines with the contact nonlinearity of fieldbus connectors, as well as edge-cloud collaborative architectures that support millisecond-level bidirectional synchronization. The contact nonlinearity includes the nonlinear characteristics of connector contact resistance varying with contact pressure, temperature, and vibration load, and the nonlinear behavior of the contact surface friction coefficient varying with relative sliding velocity. Summary of the Invention

[0005] This invention proposes a fieldbus virtual diagnostic method and related device based on digital twins. It adopts a three-layer architecture design, including a physical bus layer, an edge computing layer, and a cloud simulation layer. Through multi-physics coupling modeling and millisecond-level data synchronization mechanism, it constructs a digital twin environment that is highly consistent with the physical bus system. This solves the problem of how to build a high-fidelity virtual diagnostic environment that can accurately reflect the multi-physics coupling characteristics of the fieldbus system while avoiding the risk of damage to physical equipment. This addresses the technical challenges of complex and correlated faults being difficult to reproduce, having unclear mechanisms, and existing simulation methods being unable to accurately characterize the electrical-mechanical-thermal coupling effect.

[0006] To achieve the above objectives, the technical solution adopted by the present invention is as follows: In a first aspect, the present invention provides a fieldbus virtual diagnostic method based on digital twins, comprising the following steps: Feature extraction is performed on the collected multiphysics parameters, and clock synchronization is performed to obtain the processed feature vector; Real-time anomaly detection is performed on the processed feature vectors, where: When an anomaly is detected, the boundary conditions of the multiphysics coupled digital twin model are updated in real time based on the processed feature vector, driving the model to perform multiphysics coupled simulation. By configuring fault types through a virtual fault injection engine and modifying the parameters of a multiphysics coupled digital twin model, electrical faults, mechanical faults, and / or protocol layer faults can be safely injected to simulate fault response data.

[0007] Preferably, the multiphysics parameters include electrical operating parameters, mechanical vibration parameters, and thermal characteristic parameters.

[0008] Secondly, the present invention provides a fieldbus virtual diagnostic system based on digital twins, comprising: an edge computing layer and a cloud simulation layer, wherein: The edge computing layer is used to extract features from the collected multi-physics parameters, synchronize the clock, obtain a processed feature vector, and perform real-time anomaly detection on the processed feature vector, wherein: When an anomaly is detected, the processed feature vector is output to the cloud simulation layer; The cloud simulation layer is used to update the boundary conditions of the multiphysics coupled digital twin model in real time based on the processed feature vectors, driving the model to perform multiphysics coupled simulation; at the same time, it configures the fault type through the virtual fault injection engine and modifies the parameters of the multiphysics coupled digital twin model to safely inject electrical faults, mechanical faults and / or protocol layer faults to simulate fault response data.

[0009] Preferably, the edge computing layer includes an edge data acquisition node and an edge computing gateway. The edge data acquisition node is used to acquire multi-physics parameters of fieldbus physical nodes in real time. The edge computing gateway is used to extract features from the acquired multi-physics parameters, synchronize the clock, obtain a processed feature vector, perform real-time anomaly detection on the processed feature vector, and output the processed feature vector to the cloud simulation layer when an anomaly is detected.

[0010] Preferably, the edge computing gateway includes a data preprocessing module, a time synchronization module, and an edge inference module, wherein: The data preprocessing module is used to filter and reduce noise and extract features from the collected multi-physics parameters to obtain feature vectors; The time synchronization module is used to synchronize the obtained feature vector with a clock to obtain the processed feature vector; The edge reasoning module is used to perform real-time anomaly detection on the processed feature vector. When an anomaly is detected, the processed feature vector is output to the cloud simulation layer.

[0011] Preferably, the cloud simulation layer includes a multiphysics coupled digital twin model and a virtual fault injection engine. The multiphysics coupled digital twin model is used to update the boundary conditions of the multiphysics coupled digital twin model in real time based on the processed feature vectors, driving the model to perform multiphysics coupled simulation. The virtual fault injection engine is used to configure fault types and modify the parameters of the multiphysics coupled digital twin model to safely inject electrical faults, mechanical faults, and / or protocol layer faults to simulate fault response data.

[0012] Preferably, the multiphysics coupled digital twin model includes an electromagnetic field simulation sub-model, a mechanical stress simulation sub-model, and a heat conduction simulation sub-model, wherein: The electromagnetic field simulation sub-model is used to calculate the distributed inductance, distributed capacitance, and electromagnetic interference field strength of the physical bus cable. The mechanical stress simulation sub-model is used to calculate the stress distribution and fatigue life of the fieldbus physical node under vibration load; The heat conduction simulation sub-model is used to calculate the temperature field distribution and thermal stress coupling effect of the fieldbus physical node.

[0013] Preferably, the virtual fault injection engine includes an electrical fault injection module, a mechanical fault injection module, and a protocol fault injection module, wherein: The electrical fault injection module is used to inject cable breakage, short circuit, abnormal grounding impedance, and electromagnetic pulse interference into the electromagnetic field simulation sub-model. The mechanical fault injection module is used to inject connection loosening, mechanical resonance and fatigue crack propagation into the mechanical stress simulation sub-model; The protocol fault injection module is used to inject frame loss, frame duplication, timing errors, and CRC check errors into the data link layer.

[0014] Thirdly, the present invention provides an electronic device including a processor and a memory, wherein the memory stores computer instructions, and when the computer instructions are executed by the processor, the electronic device performs the method described thereon.

[0015] Fourthly, the present invention provides a computer program product, the computer program product including computer-executable instructions, which, when executed, implement the method described.

[0016] Compared with the prior art, the beneficial effects of the present invention are: This invention provides a fieldbus virtual diagnostic method based on digital twins. By extracting features and synchronizing the clock on acquired multi-physics parameters, a processed feature vector is obtained, which is then used for real-time anomaly detection. This solves the problem of "unclear mechanisms of complex correlated faults" mentioned in the background art. Through the synchronous acquisition and processing of multi-physics parameters, high-fidelity input data is provided for subsequent simulations. When an anomaly is detected, this method updates the boundary conditions of the multi-physics coupled digital twin model in real time based on the processed feature vector, driving the model to perform multi-physics coupled simulation. This invention addresses the deficiency in the background technology of "lack of real-time multi-physics coupling modeling methods," achieving coupled simulation of electromagnetic, mechanical, and thermal multi-physics fields. It accurately reflects the coupling relationship between the electrical, mechanical vibration, and thermal characteristics of the physical layer, overcoming the inability of pure software simulation to accurately reflect the coupling effect of multi-physics fields. Finally, by configuring fault types through a virtual fault injection engine and modifying the parameters of the multi-physics coupled digital twin model, electrical faults, mechanical faults, and / or protocol layer faults can be safely injected to simulate fault response data. This completely avoids the risk of physical equipment damage caused by physical fault injection mentioned in the background technology, achieving safe injection and simulation of various faults in a purely digital environment.

[0017] In summary, this invention, through the technical approach of "edge-side anomaly detection triggering + real-time updating of boundary conditions of multi-physics coupling model + virtual fault injection," effectively solves the technical challenges mentioned in the background art, such as the difficulty in reproducing and verifying fieldbus faults in real environments, the inability of existing simulation methods to accurately reflect multi-physics coupling relationships, and the risk of equipment damage from physical fault injection. It provides technical support for achieving high-fidelity, low-risk, and real-time fieldbus virtual diagnosis. Attached Figure Description

[0018] Figure 1 This is a system architecture diagram according to an embodiment of the present invention; Figure 2 This is a schematic diagram illustrating the principle of the multiphysics coupling model in an embodiment of the present invention; Figure 3 This is a flowchart illustrating the millisecond-level data synchronization timing of an embodiment of the present invention. Figure 4 The virtual fault injection engine architecture and cross-domain coupling diagram of this invention embodiment; Figure 5 This is a flowchart illustrating the workflow of the diagnostic algorithm verification platform according to an embodiment of the present invention. Figure 6 This is a schematic diagram of the energy flow coupling of a port Hamiltonian system according to an embodiment of the present invention. Detailed Implementation

[0019] In the following description, specific details such as particular system architectures and techniques are set forth for illustrative purposes and not for limitation, in order to provide a thorough understanding of the embodiments of this application. However, those skilled in the art will understand that this application may also be implemented in other embodiments without these specific details. In other instances, detailed descriptions of well-known systems, apparatuses, circuits, and methods have been omitted so as not to obscure the description of this application with unnecessary detail.

[0020] It should be understood that, when used in this application specification and the appended claims, the term "comprising" indicates the presence of the described features, integrals, steps, operations, elements and / or components, but does not exclude the presence or addition of one or more other features, integrals, steps, operations, elements, components and / or a collection thereof.

[0021] It should also be understood that the term “and / or” as used in this application specification and the appended claims means any combination of one or more of the associated listed items and all possible combinations, and includes such combinations.

[0022] As used in this application specification and the appended claims, the term "if" may be interpreted, depending on the context, as "when," "once," "in response to determination," or "in response to detection." Similarly, the phrase "if determined" or "if detected [the described condition or event]" may be interpreted, depending on the context, as meaning "once determined," "in response to determination," "once detected [the described condition or event]," or "in response to detection [the described condition or event]."

[0023] Furthermore, in the description of this application and the appended claims, the terms "first," "second," "third," etc., are used only to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0024] References to "one embodiment" or "some embodiments" as described in this specification mean that one or more embodiments of this application include a specific feature, structure, or characteristic described in connection with that embodiment. Therefore, the phrases "in one embodiment," "in some embodiments," "in other embodiments," "in still other embodiments," etc., appearing in different parts of this specification do not necessarily refer to the same embodiment, but rather mean "one or more, but not all, embodiments," unless otherwise specifically emphasized. The terms "comprising," "including," "having," and variations thereof mean "including but not limited to," unless otherwise specifically emphasized.

[0025] Example 1 This embodiment provides a fieldbus virtual diagnostic method based on digital twins, which includes the following steps: Feature extraction is performed on the collected multiphysics parameters, and clock synchronization is performed to obtain the processed feature vector; Real-time anomaly detection is performed on the processed feature vectors, where: When an anomaly is detected, the boundary conditions of the multiphysics coupled digital twin model are updated in real time based on the processed feature vector, driving the model to perform multiphysics coupled simulation. By configuring fault types through a virtual fault injection engine and modifying the parameters of a multiphysics coupled digital twin model, electrical faults, mechanical faults, and / or protocol layer faults can be safely injected to simulate fault response data.

[0026] Example 2 This embodiment provides a fieldbus system fault diagnosis system based on digital twin technology. The system adopts a three-layer architecture design, including a physical bus layer, an edge computing layer, and a cloud simulation layer. Through multi-physics coupling modeling and millisecond-level data synchronization mechanism, a digital twin environment highly consistent with the physical bus system is constructed.

[0027] The physical bus layer includes at least one fieldbus physical node, which is connected to the industrial control network via a physical bus cable to form an actual production site communication system.

[0028] The edge computing layer is deployed on the physical bus side and includes edge data acquisition nodes and edge computing gateways. The edge data acquisition nodes are used to acquire multi-physics parameters of the fieldbus physical nodes in real time. The multi-physics parameters include electrical operating parameters, mechanical vibration parameters, and thermal characteristic parameters. The edge computing gateway is used to preprocess, synchronize, and perform edge inference on the received multi-physics parameters in sequence.

[0029] The cloud simulation layer is used to construct a multi-physics coupled digital twin model and safely inject faults through a virtual fault injection engine to achieve real-time mapping of physical bus status, fault simulation, diagnostic algorithm verification, and remaining lifetime prediction.

[0030] The cloud-based simulation layer includes a multiphysics-coupled digital twin model and a virtual fault injection engine, among which: The multiphysics coupled digital twin model is constructed based on the finite element method and the multibody dynamics method. It is used to update the boundary conditions of the multiphysics coupled digital twin model based on the output of the edge computing layer and to perform multiphysics coupled simulation.

[0031] The virtual fault injection engine is used to configure fault types and modify the parameters of the multiphysics coupled digital twin model to safely inject various faults.

[0032] Example 3 Based on Example 2, this example provides a fieldbus system fault diagnosis system based on digital twin technology. The electrical operating parameters include the transient current waveform, voltage amplitude and spectral characteristics of the bus cable, and the sampling frequency is not less than 10MHz to capture high-frequency electromagnetic interference.

[0033] The mechanical vibration parameters include triaxial acceleration signals and strain signals, with a sampling frequency of not less than 20kHz to cover the mechanical resonance frequency band.

[0034] The thermal characteristic parameters include surface temperature field distribution and ambient temperature and humidity, with a spatial resolution of not less than 320×240 pixels.

[0035] Example 4 Based on Embodiment 2, this embodiment provides a fieldbus system fault diagnosis system based on digital twin technology. The edge computing gateway includes a data preprocessing module, a time synchronization module, and an edge inference module, wherein: The data preprocessing module is used to filter and reduce noise and extract features from the collected multi-physics parameters to obtain multi-data fault features, which include mechanical fault features, electrical fault features and thermal fault features. The time synchronization module is used to realize clock synchronization of multi-source data based on the IEEE 1588 precision time protocol, with a synchronization accuracy better than 100 nanoseconds, ensuring the consistency of timestamps for electrical, mechanical, and thermal data. The edge inference module is used to perform real-time anomaly detection on preprocessed data based on a lightweight neural network model. If the detection result is an anomaly, the detection result and the received multiphysics parameters are transmitted to the cloud simulation layer. In this embodiment, the lightweight neural network model is obtained by compressing a large model from the cloud using knowledge distillation technology, enabling millisecond-level inference on the edge device, and packaging the detection result with the original data for transmission to the cloud simulation layer.

[0036] Example 5 Based on Example 2, this example provides a fieldbus system fault diagnosis system based on digital twin technology. The multiphysics coupled digital twin model includes an electromagnetic field simulation sub-model, a mechanical stress simulation sub-model, and a heat conduction simulation sub-model, wherein: The electromagnetic field simulation sub-model is constructed based on Maxwell's equations and is used to calculate the distributed inductance, distributed capacitance, and electromagnetic interference field strength of the physical bus cable using the finite-difference time-domain (FDTD) method combined with the processed eigenvectors, taking into account the conductor skin effect and dielectric loss.

[0037] The mechanical stress simulation sub-model is constructed based on multibody dynamics theory. It uses the Lagrange equation and the processed eigenvectors to describe the kinematic relationship of fieldbus physical nodes under vibration load, calculate stress distribution and fatigue life, and consider bolt preload relaxation and contact nonlinearity.

[0038] The heat conduction simulation sub-model is constructed based on Fourier's heat conduction law and computational fluid dynamics equations. It is used to calculate the temperature field distribution and thermal stress coupling effect of fieldbus physical nodes using the finite volume method (FVM) combined with the processed eigenvectors, taking into account Joule heat generation and convective heat dissipation.

[0039] The three sub-models achieve energy flow coupling and bidirectional data interaction through the port Hamiltonian system method. The energy port variables include the Poynting vector of the electromagnetic field, the stress-strain power of the mechanical domain, and the heat flux density of the thermal domain. The energy conservation and numerical stability of the multiphysics coupling model are ensured by power-preserving interconnection conditions.

[0040] In this embodiment, the port Hamiltonian system method is adopted, and the transmission line distributed parameters (distributed inductance, capacitance, and resistance) are used as the output of the electromagnetic field sub-model. The connector contact nonlinearity (contact resistance and friction coefficient) is used as the boundary condition of the mechanical stress sub-model. By defining the cross-domain energy port variables (Poynting vector, stress power, and heat flux density), the electrical-mechanical-thermal three-field coupling is realized, and a unified state-space model is constructed.

[0041] Example 6 Based on Example 2, this example provides a fieldbus system fault diagnosis system based on digital twin technology. The edge computing gateway establishes a data synchronization channel based on the Time-Sensitive Networking (TSN) protocol between itself and the cloud simulation layer, achieving millisecond-level state synchronization between the physical bus layer and the multi-physics coupled digital twin model. Specifically: The edge computing gateway uploads the pre-processed electrical, mechanical, and thermal parameters to the cloud at a cycle of no more than 1 millisecond. The cloud simulation layer updates the boundary conditions of the multiphysics coupled digital twin model in real time based on the received processed feature vectors, drives the simulation model forward, and after completing a simulation step of no more than 5 milliseconds, it downloads the virtual state data to the edge computing gateway for comparison and verification with the physical measured data. In this embodiment, the virtual state data includes electrical signal waveforms, mechanical vibration responses, and temperature distributions predicted by the model.

[0042] This two-way synchronization mechanism ensures state consistency between the digital twin model and the physical system, enabling a virtual diagnostic environment at the hardware-in-the-loop simulation level.

[0043] Example 7 Based on Embodiment 2, this embodiment provides a fieldbus system fault diagnosis system based on digital twin technology. The virtual fault injection engine includes an electrical fault injection module, a mechanical fault injection module, and a protocol fault injection module, wherein: The electrical fault injection module can inject cable breakage, short circuit, abnormal grounding impedance, and electromagnetic pulse interference into the electromagnetic field simulation sub-model by modifying the circuit topology and boundary conditions. The mechanical fault injection module can inject connection loosening, mechanical resonance, and fatigue crack propagation into the mechanical stress simulation sub-model by modifying the stiffness matrix and damping coefficient. The protocol fault injection module can inject frame loss, frame duplication, timing errors, and CRC check errors into the data link layer by modifying the communication protocol stack parameters. The data link layer is the data link layer simulation module of the fieldbus communication protocol stack integrated in the cloud simulation layer.

[0044] Example 8 Based on Example 2, this example provides a fieldbus system fault diagnosis system based on digital twin technology. The virtual fault injection engine also includes a fault correlation analysis module, which establishes a mapping relationship matrix between electrical layer faults, mechanical layer faults and protocol layer faults. When a single type of fault is injected, the coupling effect of the fault on other physical domains is calculated based on the matrix, and the boundary conditions of the corresponding physical domains are updated synchronously in the multi-physics coupled digital twin model to realize fault propagation simulation across physical domains.

[0045] The electrical layer faults refer to abnormal electrical characteristics of the physical layer, including cable breaks, short circuits, abnormal grounding impedance, and electromagnetic pulse interference; the mechanical layer faults refer to abnormal mechanical structures, including loose connections, mechanical resonance, and fatigue crack propagation; the protocol layer faults refer to abnormal communication protocols, including frame loss, frame duplication, timing errors, and CRC check errors.

[0046] Example 9 Based on Embodiment 2, this embodiment provides a fieldbus system fault diagnosis system based on digital twin technology. The fault diagnosis system further includes a diagnostic algorithm verification platform, comprising a diagnostic algorithm module to be verified, a performance evaluation module, and a training data generation module, wherein: The diagnostic algorithm module to be verified is used to receive fault response data output by the multi-physics coupled digital twin model and execute fault classification, location and prediction algorithms. The performance evaluation module is used to calculate indicators such as fault detection rate, false alarm rate and isolation accuracy, and compare them with preset thresholds; The training data generation module is used to control the virtual fault injection engine to perform batch fault injection experiments according to the preset fault mode library, automatically collect fault response data and label it, and generate a standardized dataset for training machine learning algorithms. The fault mode library includes single fault mode, concurrent fault mode and intermittent fault mode.

[0047] Example 10 Referring to the accompanying drawings, this embodiment provides a fieldbus virtual diagnostic system based on digital twins. This embodiment is applied in a thermal power plant's DCS (Distributed Control System) and includes: Implementation of the physical bus layer: In the Profibus-DP fieldbus network of the boiler control system of a thermal power plant, 10 fieldbus physical nodes are selected as monitoring objects. In this embodiment, the 10 fieldbus physical nodes include 8 distributed IO stations and 2 master stations.

[0048] Deploy one set of edge data acquisition nodes at each fieldbus physical node.

[0049] In this embodiment, the edge data acquisition node includes a high-frequency current transformer, a triaxial accelerometer, and an infrared thermal imager, wherein: The high-frequency current transformer adopts a Rogowski coil structure with a frequency response range of 0.1Hz to 30MHz. It is installed outside the shielding layer of the bus cable and is used for non-invasive measurement of bus current waveform. The sampling frequency is set to 20MHz and the resolution is 16 bits.

[0050] The triaxial accelerometer uses MEMS technology, has a range of ±50g, a frequency response range of 0.5Hz to 10kHz, is mounted on the metal housing of the fieldbus connector, and is used to collect mechanical vibration signals. The sampling frequency is set to 25.6kHz.

[0051] The infrared thermal imager uses an uncooled focal plane detector, operates in the 8 to 14 micrometer band, and has a thermal sensitivity (NETD) of no more than 50 mK. It is mounted on a bracket 0.5 meters away from the bus node and is used to monitor the node temperature distribution. The frame rate is set to 30 Hz, and each frame is transmitted to the edge computing gateway for thermal feature extraction.

[0052] All sensors are connected to the edge computing gateway via shielded cables, with cable lengths not exceeding 2 meters, to reduce signal attenuation.

[0053] Implementation of the edge computing layer: The edge computing gateway adopts an industrial-grade embedded computer, equipped with an Intel Core i7 processor, 16GB DDR4 memory and 512GB solid-state storage, and runs a real-time Linux operating system.

[0054] The edge computing gateway includes a data preprocessing module, a time synchronization module, and an edge inference module, wherein: The data preprocessing module implements a multi-rate signal fusion algorithm, downsampling high-frequency electrical signals to 100kHz; interpolating low-frequency temperature signals to 100Hz to unify the time reference; using a wavelet packet decomposition algorithm to perform three-level decomposition on the vibration signal and extracting the energy of each frequency band as mechanical fault features; using a fast Fourier transform (FFT) to perform spectral analysis on the current signal and extracting harmonic content as electrical fault features; and using a temperature gradient algorithm to calculate thermal imager data and extract hotspot locations and temperature rise rates as thermal fault features.

[0055] The time synchronization module implements the IEEE 1588v2 precise time protocol, marks the sending and receiving time of each data packet with hardware timestamps, selects the synchronization source using the Best Master Clock (BMC) algorithm, and measures network latency through a delay request-response mechanism, achieving clock synchronization accuracy between edge nodes better than 50 nanoseconds.

[0056] The edge inference module deploys a lightweight convolutional neural network model based on TensorFlow Lite. This model is compressed from a deep residual network (ResNet-50) trained in the cloud using knowledge distillation technology, resulting in a model size that is only 5% of the original model. Its inference latency is less than 2 milliseconds, enabling real-time detection of abnormal data patterns. When an anomaly is detected, the edge computing gateway triggers a flag and uploads it along with the original data to the cloud via the TSN network.

[0057] Construction and implementation of multiphysics coupled digital twin model: The cloud simulation layer is deployed on a high-performance computing server cluster and is implemented by coupling the multiphysics simulation software COMSOL Multiphysics with the self-developed port Hamiltonian system solver.

[0058] The construction process of the electromagnetic field simulation sub-model is as follows: A three-dimensional geometric model of the Profibus-DP bus cable is established, including the internal copper conductor, insulation layer, shielding layer, and sheath. The copper conductor has a diameter of 2.5 mm, the insulation layer has a thickness of 1.2 mm, and the shielding layer is a composite structure of aluminum foil and braided mesh.

[0059] Based on Maxwell's equations, establish the time-harmonic electromagnetic field control equations corresponding to the three-dimensional geometric model:

[0060] In the formula, is the magnetic vector potential (Wb / m). ρ is the magnetic permeability (H / m). The conductivity is expressed in S / m. is the dielectric constant (F / m). ω is the angular frequency (rad / s). The external current density is (A / m²).

[0061] The solution domain was discretized using the finite element method. The mesh size was refined to 0.1 mm on the conductor surface and gradually coarsened to 5 mm further away from the conductor, with a total of approximately 1.2 million elements. The distributed inductance per unit length was calculated. and distributed capacitance The values ​​were 0.65 μH / m and 85 pF / m, respectively, with errors of less than 3% compared to the measured values.

[0062] Considering the skin effect, the AC resistance of a conductor varies with frequency as follows:

[0063] In the formula, The resistance is AC (Ω / m). The resistance is DC (Ω / m). is the skin effect coefficient (s^0.5), and f is the frequency (Hz).

[0064] Based on the relationship between conductor AC resistance and frequency, conductor loss (skin effect loss) during high-frequency signal transmission is calculated, and this loss is used as a heat source. Input the thermal conductivity submodel to achieve electromagnetic-thermal coupling.

[0065] The construction process of the mechanical stress simulation sub-model is as follows: Establish a multibody dynamics model for the fieldbus connector, including the plug, socket, bolt, and PCB board.

[0066] The system dynamics are described using the Lagrange equations:

[0067] In the formula, Let T be the system kinetic energy (J) and V be the system potential energy (J), where... , ,in M For the quality matrix, K Here is the stiffness matrix. It is gravitational potential energy. For generalized coordinates (m or rad). It is a generalized force (N or N·m).

[0068] Bolted connections are simulated using spring-damped elements with a preload of 50 N. Considering contact nonlinearity, a Coulomb friction model with a friction coefficient of 0.15 is employed. Under external vibration excitation, the relative displacement and contact pressure of the connector contact surfaces are calculated to assess the risk of connection loosening. This assessment is then used to predict connector contact failure, providing early warning of poor electrical contact or communication interruption caused by mechanical loosening, thus achieving coupled mechanical-electrical fault analysis.

[0069] The stress distribution is obtained by discretizing the Lagrange equations using the finite element method and combining it with multibody dynamics calculations; fatigue life is calculated using cumulative damage theory or crack propagation law.

[0070] The construction process of the heat conduction simulation sub-model is as follows: Establish a three-dimensional thermal model of the bus node, including connectors, cables, terminating resistors, and housing.

[0071] The governing equations for heat conduction are established based on the law of conservation of energy.

[0072] In the formula, The density of the material is (kg / m³). is the specific heat capacity (J / (kg·K)), T is the temperature (K), and k is the thermal conductivity (W / (m·K)). The current distribution, calculated from electromagnetic field simulation, represents the Joule heat generation rate (W / m³). ,in The value is the current density (A / m²). The electric field strength is expressed in V / m.

[0073] Considering both natural convection and radiative heat dissipation, the convective heat transfer coefficient h is taken as 5 to 25 W / (m²·K), and the radiative emissivity... The value is set to 0.8. Simulation results show that under a rated current of 2A, the temperature rise of the terminal resistor does not exceed 15K, which is consistent with the measured infrared thermal image data. The heat dissipation boundary is defined in the heat conduction boundary conditions based on natural convection and radiation heat dissipation, and is specifically applied to the boundary terms of the governing equations. ,in h The convective heat transfer coefficient is... It represents the radiative emissivity.

[0074] The temperature field is calculated by solving the governing equations of heat conduction using the finite volume method (FVM); thermal stress is calculated using structural mechanics equations. Calculations are used to achieve thermo-mechanical coupling.

[0075] Implementation of the port Hamiltonian system coupling method: To achieve electromagnetic-mechanical-thermal coupling, a state-space model corresponding to the port Hamiltonian system is constructed. This state-space model includes defining state variables. ,in This refers to the electromagnetic field state (magnetic flux and charge). For mechanical states (displacement and momentum). Thermal state (entropy); Input variables Represents external input power; output variable This indicates the system response.

[0076] Hamiltonian functions of each subsystem Represents the total energy of the system, satisfying:

[0077] In the formula, Power dissipation (W) is generated by resistance, damping, and thermal resistance. H For Hamiltonian functions, T is Matrix transpose symbol 。

[0078] By defining the interconnection matrix J and the dissipation matrix R, the dynamic equations corresponding to the port Hamiltonian system are:

[0079] In the formula, g(x) is the input matrix.

[0080] The coupling between the electromagnetic field and the thermal field is achieved through Joule heating: As a heat source, input thermal model; The coupling between mechanical and electromagnetic fields is achieved through electromagnetic force: As a mechanical model for external force input; The coupling between the thermal field and the mechanical field is achieved through thermal stress: To correct mechanical stress, where, It is Young's modulus (Pa). The coefficient of thermal expansion (K) - ¹), The temperature difference is K. This power-maintaining interconnect ensures energy conservation and numerical stability of the coupled system. The simulation step size is set to 2 milliseconds to meet real-time requirements.

[0081] Implementation of millisecond-level data synchronization mechanism: The edge computing gateway and the cloud simulation layer are connected via gigabit industrial Ethernet, employing the IEEE 802.1AS Time-Sensitive Networking (TSN) protocol to ensure deterministic data transmission and low latency. The data synchronization process is as follows: At the start of each control cycle (1 millisecond), the edge computing gateway collects the current data from all sensors and adds a precise timestamp. By using the gated scheduling mechanism of TSN, data packets are allocated to the highest priority time-sensitive queue to ensure that the transmission delay is less than 500 microseconds; After receiving the data, the cloud simulation layer immediately updates the boundary conditions of the multiphysics coupled digital twin model, driving the simulation to advance by one step (2 milliseconds). After the simulation is completed, the virtual state data (including the predicted bus voltage waveform, connector stress distribution and temperature field) is encapsulated into a data packet and transmitted back to the edge computing gateway via TSN. The transmission delay is also less than 500 microseconds. Edge computing gateways compare virtual data with physical measured data, calculate residuals, and use them for online model correction.

[0082] The total latency of the entire bidirectional synchronous closed loop is less than 4 milliseconds, achieving near real-time synchronization between the physical system and the virtual model.

[0083] Implementation of the virtual fault injection engine: The virtual fault injection engine provides a graphical user interface that allows users to configure fault parameters and trigger injection.

[0084] Examples of electrical fault injection include: in an electromagnetic field simulation sub-model, adjusting the resistivity of a section of bus cable from its normal value. Modified to To simulate high-resistance contact faults; Add a capacitor to ground in the circuit topology To simulate cable insulation degradation; injection amplitude Ascent time The pulse voltage is used to simulate electromagnetic interference.

[0085] Examples of mechanical fault injection include: gradually reducing bolt preload from 50N to 10N to simulate connection loosening; superimposing a sinusoidal excitation with the connector's natural frequency (approximately 150Hz) onto the vibration signal to simulate mechanical resonance; and setting the crack propagation rate based on Paris's law in fatigue crack propagation simulation.

[0086] In the formula, a is the crack length (m), N is the number of cycles, and C and m are material constants. The range of stress intensity factor (MPa·√m).

[0087] Examples of protocol fault injection include: setting the random drop probability at the data link layer to 10% to simulate frame loss; modifying the standard bit time (1.5 bit time) to 1.0 or 2.0 bit time to simulate timing errors; and modifying the CRC check polynomial to simulate CRC errors. The fault correlation analysis module establishes a 3×3 fault mapping matrix. When an electrical fault (such as a cable break) is injected, it automatically calculates the resulting changes in mechanical stress (due to the loss of electrodynamic force) and thermal distribution (due to current interruption), and synchronously updates the corresponding boundary conditions in the multiphysics coupled digital twin model, achieving a comprehensive simulation of the fault propagation effect.

[0088] Implementation of the Diagnostic Algorithm Validation Platform: The diagnostic algorithm validation platform provides a standardized API interface, supporting the integration of diagnostic algorithms written in Python, MATLAB, and C++. The diagnostic algorithm to be validated receives fault response data from a multi-physics coupled digital twin model, including electrical signal waveforms, vibration spectra, and thermogram sequences, and performs feature extraction, fault classification, and localization.

[0089] The performance evaluation module calculates the following metric: Fault Detection Rate (FDR) = ,in The number of faults detected. Total number of injected faults; False Alarm Rate (FAR) = ,in Number of false alarms Number of samples in normal condition; Isolation precision (IA) = ,in To correctly isolate the number of faults down to the specific component.

[0090] The training data generation module controls the virtual fault injection engine to perform batch experiments, such as continuously injecting 1000 different types of faults, automatically generating labeled datasets for training and testing deep learning models. In thermal power plant applications, this invention was used to generate a dataset containing 50 fault modes and a total of 50,000 records. The trained convolutional neural network model achieved a fault detection rate of 96.5% in actual deployment.

[0091] The deployment implementation of this invention's system in a thermal power plant boiler control system involves installing 10 sets of edge data acquisition nodes at key nodes of the boiler room fieldbus, connected via fiber optic Ethernet to an edge computing gateway deployed in the central control room. The cloud simulation layer is deployed on a server cluster in the power plant's information center, connected to the edge layer via an industrial firewall. After system operation, it successfully simulated 12 typical faults, including intermittent poor cable contact, connector vibration and loosening, and overheating of terminal resistors. By comparing the virtual fault injection results with actual historical fault records, the fault phenomena showed a 93% match, verifying the high fidelity of the digital twin model. Maintenance personnel used this invention's system for training, repeatedly practicing fault diagnosis procedures in a virtual environment, shortening the training cycle from the traditional two weeks to three days, and enabling it to be conducted without interrupting production. Furthermore, based on this invention's system, a novel deep learning-based fault prediction algorithm was verified. This algorithm can predict potential faults in the bus system 72 hours in advance, providing important technical support for predictive maintenance in thermal power plants.

[0092] In this embodiment, by employing a multi-physics coupled digital twin modeling method, the electrical, mechanical, and thermal characteristics of the fieldbus system and their coupling relationships are accurately characterized, thus achieving high-fidelity simulation of complex and interconnected faults with a fault simulation accuracy exceeding 92%. The use of an edge-cloud collaborative data synchronization mechanism enables millisecond-level synchronization between the physical system and the virtual model, ensuring the real-time performance and realism of the virtual diagnostic environment. Data synchronization latency is less than 1 millisecond, meeting the real-time control requirements of industrial sites. Furthermore, the use of virtual fault injection technology simulates various typical and extreme faults in a purely digital environment, thus avoiding the risk of physical equipment damage and reducing fault verification costs by over 85%.

[0093] Specific test data shows that in the fieldbus diagnostic application of DCS systems in thermal power plants, the system of this invention has significant advantages over existing technologies: In terms of fault detection rate, the system of this invention achieves 96.5%, while traditional diagnostic methods based on communication log analysis only achieve 78.3%, and existing simulation methods based on simplified communication models achieve 82.1%. Regarding fault location accuracy, the system of this invention can isolate faults to specific bus nodes or cable segments with an accuracy of 94.2%, while traditional methods only achieve 65.4%. In terms of diagnostic response time, the average time from fault occurrence to location completion for the system of this invention is 12 milliseconds, while traditional hardware-in-the-loop testing methods require no less than 500 milliseconds, and although pure software simulation methods have a fast response, they cannot reflect the true physical characteristics. Regarding hardware resource consumption, this invention reduces the edge-side computing load to 30% of traditional centralized processing through an edge-cloud collaborative architecture, while simultaneously improving cloud simulation computing efficiency by more than 3 times through model reduction technology.

[0094] This invention also provides a safe training environment for maintenance personnel. By simulating various emergency situations through virtual fault injection, training effectiveness evaluation shows that maintenance personnel trained using this invention's system experience a 40% reduction in average response time and a 60% reduction in error rate during actual fault handling. Furthermore, this invention supports the verification of predictive maintenance algorithms. Through long-term operational data accumulation, it can predict potential bus system faults up to 72 hours in advance with an accuracy rate of 89%, providing crucial technical support for industrial production.

[0095] Example 11 This embodiment also provides a computing device. The computing device includes a bus, a processor, a memory, and a communication interface. The processor, memory, and communication interface communicate with each other via the bus. The computing device can be a server or a terminal device. It should be understood that this application does not limit the number of processors and memory in the computing device.

[0096] A bus can be a Peripheral Component Interconnect (PCI) bus or an Extended Industry Standard Architecture (EISA) bus, etc. Buses can be categorized as address buses, data buses, control buses, etc. For ease of representation, a bus can include a path for transmitting information between various components of a computing device (e.g., memory, processor, communication interfaces).

[0097] The processor may include any one or more of the following: central processing unit (CPU), graphics processing unit (GPU), tensor processing unit (TPU), application specific integrated circuit (ASIC), field-programmable gate array (FPGA), microprocessor (MP), or digital signal processor (DSP).

[0098] Memory can include volatile memory, such as random access memory (RAM). Processors can also include non-volatile memory. volatile memory, such as read-only memory (ROM). ROM (memory only), flash memory, hard disk drive (HDD), or solid state drive (SSD).

[0099] The memory stores executable program code, which the processor executes to implement the functions of the aforementioned units, thereby achieving, for example, the method described in Embodiment 1. That is, the memory may store instructions for the methods and functions relating to the computing device in any of the above embodiments.

[0100] The communication interface uses transceiver modules such as, but not limited to, network interface cards and transceivers to enable communication between computing devices and other devices or communication networks.

[0101] Example 12 This embodiment also provides a computer-readable storage medium storing computer instructions that, when executed by a processor, cause the processor to perform the methods and functions of the computing device involved in any of the above embodiments.

[0102] Generally, the various embodiments of this disclosure can be implemented in hardware or dedicated circuitry, software, logic, or any combination thereof. Some aspects can be implemented in hardware, while others can be implemented in firmware or software, which can be executed by a controller, microprocessor, or other computing device. Although various aspects of the embodiments of this disclosure are shown and described as block diagrams, flowcharts, or represented using some other illustration, it should be understood that the blocks, apparatuses, systems, techniques, or methods described herein can be implemented as, as non-limiting examples, in hardware, software, firmware, dedicated circuitry or logic, general-purpose hardware or controllers or other computing devices, or some combination thereof.

[0103] The above-described embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application, and should all be included within the protection scope of this application.

Claims

1. A fieldbus virtual diagnostic method based on digital twins, characterized in that, Includes the following steps: Feature extraction is performed on the collected multiphysics parameters, and clock synchronization is performed to obtain the processed feature vector; Real-time anomaly detection is performed on the processed feature vectors, where: When an anomaly is detected, the boundary conditions of the multiphysics coupled digital twin model are updated in real time based on the processed feature vector, driving the model to perform multiphysics coupled simulation. By configuring fault types through a virtual fault injection engine and modifying the parameters of a multiphysics coupled digital twin model, electrical faults, mechanical faults, and / or protocol layer faults can be safely injected to simulate fault response data.

2. The fieldbus virtual diagnostic method based on digital twin according to claim 1, characterized in that, The multiphysics parameters include electrical operating parameters, mechanical vibration parameters, and thermal characteristic parameters.

3. A fieldbus virtual diagnostic system based on digital twins, characterized in that, include: Edge computing layer and cloud simulation layer, wherein: The edge computing layer is used to extract features from the collected multi-physics parameters, synchronize the clock, obtain a processed feature vector, and perform real-time anomaly detection on the processed feature vector, wherein: When an anomaly is detected, the processed feature vector is output to the cloud simulation layer; The cloud simulation layer is used to update the boundary conditions of the multiphysics coupled digital twin model in real time based on the processed feature vectors, driving the model to perform multiphysics coupled simulation; at the same time, it configures the fault type through the virtual fault injection engine and modifies the parameters of the multiphysics coupled digital twin model to safely inject electrical faults, mechanical faults and / or protocol layer faults to simulate fault response data.

4. The fieldbus virtual diagnostic system based on digital twin according to claim 3, characterized in that, The edge computing layer includes edge data acquisition nodes and edge computing gateways. The edge data acquisition nodes are used to acquire multi-physics parameters of fieldbus physical nodes in real time. The edge computing gateway is used to extract features from the acquired multi-physics parameters, synchronize the clock, obtain processed feature vectors, perform real-time anomaly detection on the processed feature vectors, and output the processed feature vectors to the cloud simulation layer when an anomaly is detected.

5. A fieldbus virtual diagnostic system based on digital twins according to claim 4, characterized in that, The edge computing gateway includes a data preprocessing module, a time synchronization module, and an edge inference module, wherein: The data preprocessing module is used to filter and reduce noise and extract features from the collected multi-physics parameters to obtain feature vectors; The time synchronization module is used to synchronize the obtained feature vector with a clock to obtain the processed feature vector; The edge reasoning module is used to perform real-time anomaly detection on the processed feature vector. When an anomaly is detected, the processed feature vector is output to the cloud simulation layer.

6. The fieldbus virtual diagnostic system based on digital twin according to claim 3, characterized in that, The cloud simulation layer includes a multiphysics coupled digital twin model and a virtual fault injection engine. The multiphysics coupled digital twin model is used to update the boundary conditions of the multiphysics coupled digital twin model in real time based on the processed feature vectors, driving the model to perform multiphysics coupled simulation. The virtual fault injection engine is used to configure fault types and modify the parameters of the multiphysics coupled digital twin model to safely inject electrical faults, mechanical faults and / or protocol layer faults to simulate fault response data.

7. A fieldbus virtual diagnostic system based on digital twins according to claim 6, characterized in that, The multiphysics coupled digital twin model includes an electromagnetic field simulation sub-model, a mechanical stress simulation sub-model, and a heat conduction simulation sub-model, wherein: The electromagnetic field simulation sub-model is used to calculate the distributed inductance, distributed capacitance, and electromagnetic interference field strength of the physical bus cable. The mechanical stress simulation sub-model is used to calculate the stress distribution and fatigue life of the fieldbus physical node under vibration load; The heat conduction simulation sub-model is used to calculate the temperature field distribution and thermal stress coupling effect of the fieldbus physical node.

8. A fieldbus virtual diagnostic system based on digital twins according to claim 6, characterized in that, The virtual fault injection engine includes an electrical fault injection module, a mechanical fault injection module, and a protocol fault injection module, wherein: The electrical fault injection module is used to inject cable breakage, short circuit, abnormal grounding impedance, and electromagnetic pulse interference into the electromagnetic field simulation sub-model. The mechanical fault injection module is used to inject connection loosening, mechanical resonance and fatigue crack propagation into the mechanical stress simulation sub-model; The protocol fault injection module is used to inject frame loss, frame duplication, timing errors, and CRC check errors into the data link layer.

9. An electronic device, characterized in that, It includes a processor and a memory, the memory storing computer instructions that, when executed by the processor, cause the electronic device to perform the method of any one of claims 1 to 2.

10. A computer program product, characterized in that, The computer program product includes computer-executable instructions that, when executed, implement the method of any one of claims 1 to 2.

Images