Metaaction unit-based electromechanical system digital twinborn model construction method

Through the digital twin model construction method based on meta-action units, the problem that digital twins in the electromechanical system are difficult to accurately reflect the state of physical entities is solved, and the high-fidelity reflection and real-time response capabilities are improved, simplifying the update process.

CN118940469APending Publication Date: 2024-11-12CHONGQING UNIV
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202410929011.1
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-07-11
Publication Date
2024-11-12

AI Technical Summary

Technical Problem

The digital twins of electromechanical systems are difficult to accurately reflect the state of physical entities, lack real-time response capabilities, and lack unified construction standards.

Method used

The digital twin model construction method based on meta-action units is adopted, and the meta-action units are obtained through FMA functional decomposition, and the physical twin model is constructed and updated in real time. The digital twin behavior model is used to correct random errors to achieve high-fidelity reflection.

Benefits of technology

It improves the real-time response capability of digital twins, realizes high-fidelity reflection of the state of physical entities, shortens the response differences between physical twin models and physical entities, simplifies the update process, and improves accuracy and efficiency.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118940469A_ABST
    Figure CN118940469A_ABST
Patent Text Reader

Abstract

The invention relates to an electromechanical system digital twinning model construction method based on meta-action units, which comprises the following steps: decomposing an electromechanical system into a plurality of meta-action units based on a meta-action unit theory, and constructing a physical twinning model and a digital twinning behavior model for the meta-action units based on an operation state; the digital twinborn behavior model is used for dynamically sensing the operation state of the element action unit, correcting a random error in a subsequent physical twinborn model updating process and updating the physical twinborn model in real time; therefore, firstly, the response difference between the physical twinborn model and the physical entity is shortened by updating the physical twinborn model in real time; secondly, compensating errors existing in the physical twinborn model by using the digital twinborn behavior model so as to correct the response of the physical twinborn model at the current moment; according to the method, real-time reflection of the virtual model to the physical model is realized, so that high fidelity of digital twinning is realized.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention belongs to the technical field of digital twin of electromechanical systems, and specifically relates to a method for constructing a digital twin model of an electromechanical system based on a meta-action unit. Background Art

[0002] Digital Twin technology is an integrated multi-physical, multi-scale, and multi-probability simulation process that completes mapping in virtual space to reflect the entire life cycle of the corresponding physical equipment. It combines sensor data, machine learning models, and software algorithms to create dynamic virtual models of physical objects or systems; these models can reflect the operating status of the entity in real time, allowing users to remotely monitor, analyze, and predict. Digital Twin technology is now widely used in fields such as fault prediction and diagnosis, intelligent maintenance, and system optimization.

[0003] Among them, when constructing digital twins in electromechanical systems, digital twins are generally constructed for the entire system, key components in the electromechanical system, and key parts. These analysis objects have some shortcomings. There are mainly three aspects: 1) It is difficult for the twin to clearly express the relationship between the components and the whole machine, and it can only reflect the single independent state of the whole or the component; 2) There are many parts, and their structure, quantity and operating status are very different. The digital twin construction methods of the parts of different electromechanical equipment will be different, so it is difficult to formulate a unified digital twin system construction standard; 3) The twin constructed from the whole machine is difficult to go deep into the system to reflect the component status; and the construction from the component is labor-intensive and difficult to achieve. Therefore, it is necessary to seek new digital twin construction objects for electromechanical systems to overcome the above problems.

[0004] In response to the current problem of too large or too small analysis granularity of electromechanical systems, Professor Zhang Genbao's team at Chongqing University proposed a new decomposition method based on the movement and function of the electromechanical system. The electromechanical system is decomposed from top to bottom using the function-motion-action decomposition method (FMA) to obtain the most basic meta-action unit. Currently, meta-action theory is mainly used to analyze the quality, reliability and accuracy of electromechanical systems, and its analytical advantages have been recognized by many scholars in the field of reliability. In "Construction of Digital Twin Model of Drive Transmission System Based on Meta-action Theory", Yang Xin et al. further established a digital twin framework based on meta-action units, further promoting the development of digital twin technology in the application of meta-action theory system.

[0005] However, the above existing research only stays at the construction of the framework and the study of related concepts, lacking the practical application of digital twin technology. For different functional components, how to improve the real-time response capability of digital twins, and reflect the state of physical entities with high fidelity through the measurement results of limited sensors, so that the constructed digital twins can accurately reflect the characteristics, behaviors and environment of physical entities, is the key in the digital twin technology of electromechanical systems, and is also one of the problems that need to be solved urgently. Summary of the invention

[0006] In view of the above-mentioned deficiencies in the prior art, the technical problem to be solved by the present invention is to provide a method for constructing a digital twin model of an electromechanical system based on a meta-action unit, so as to avoid the problem that the digital twin constructed for the electromechanical system is difficult to accurately reflect the state of the physical entity, thereby improving the real-time response capability of the twin to achieve the high-fidelity effect of the digital twin.

[0007] In order to solve the above technical problems, the present invention adopts the following technical solutions:

[0008] A method for constructing a digital twin model of an electromechanical system based on a meta-action unit, characterized in that it comprises the following steps:

[0009] S1. Use the FMA functional decomposition method to decompose the electromechanical system layer by layer according to the "function-motion-action" mapping route to obtain several meta-action units. The meta-action units only contain two types of the most basic motion units, namely, the moving meta-action unit and the rotating meta-action unit;

[0010] S2. Construct a physical twin model for the meta-action unit based on the physical mechanism, and obtain the characterization parameters of the running state of the physical twin model;

[0011] S3, updating the physical twin model obtained in step S2 in real time to obtain a real-time updated physical twin model;

[0012] S4. Obtain historical data during the actual operation of the meta-action unit, and build a digital twin behavior model based on the historical data and the real-time updated physical twin model, wherein the digital twin behavior model is used to correct random errors in the real-time update process of the physical twin model;

[0013] S5. The digital twin behavior model obtained in step S4 is combined with the real-time updated physical twin model to construct the digital twin model. The response value of the real-time data output by the real-time updated physical twin model during the actual operation of the meta-action unit is corrected by the digital twin behavior model, which is the digital twin response value of the digital twin model to the real-time data.

[0014] To further improve the above technical solution, the physical twin model includes a meta-action unit trajectory model, a meta-action unit velocity model and a meta-action unit position model;

[0015] The meta-action unit is abstracted into a mass block, and the meta-action unit trajectory model is constructed. The specific expression is as follows:

[0016]

[0017] In formula (3.1), y(t) represents the displacement caused by the force, t represents the running time of the element action unit; C 1 and C 2 is a constant; parameter c u represents the equivalent damping, m u Represents the mass of the mass block, μ n Indicates the vibration frequency at the time of initial normal operation; Y represents the static displacement ratio of the element action unit, F 0 Indicates the force in the vibration direction at the time of initial normal operation, k u represents the equivalent stiffness; λ represents the vibration frequency ratio of the element action unit, μ represents the external force frequency; Indicates phase;

[0018] Introduce the meta-action unit trajectory state representation parameter Θ 1 , which is expressed as follows:

[0019]

[0020] In formula (3.2), Y t represents the imbalance amount when the meta-action unit is running, R t Represents the real part of the first-order vibration at the current moment, I t Represents the imaginary part of the first-order vibration at the current moment, R 0 represents the real part of the initial normal operation time, I 0 represents the imaginary part of the initial normal operation time; λ t Indicates the vibration frequency μ of the element action unit at the current moment t The vibration frequency μ at the initial normal operation 0 The ratio of A t Indicates the vibration amplitude a of the element action unit at the current moment t The vibration amplitude a at the initial normal operation 0 difference;

[0021] The velocity model of the element action unit is expressed by doing work. According to the law of conservation of energy, the kinetic energy increment ΔE in time Δt is equal to the work done by the resultant external force of the element action unit. The specific equation is as follows:

[0022] ΔE=ΔW=Pdt (3.3)

[0023] In formula (3.3), P represents the instantaneous power of the resultant external force. Further expanding formula (3.3), we can get the expression of the velocity model of the elementary action unit, which is as follows:

[0024]

[0025] In formula (3.4), n represents the number of parts that produce motion in the elementary action unit; m n represents the mass of the nth part; O represents the center of mass of the nth part; J On represents the moment of inertia of the nth part relative to the center of mass; v On Represents the velocity vector of the nth part relative to the center of mass; ω n 、F d 、F r 、M d 、M r They represent the angular velocity, driving force, resistance, driving torque and resistance torque of the nth part respectively, all of which are vectors;

[0026] When the elementary action unit is a rotation elementary action unit, based on formula (3.4), the rotation elementary action unit is obtained at t i Angular velocity ω at time o (t i ), the specific expression is as follows:

[0027]

[0028] In formula (3.5), ω(t i-1 ) represents the angular velocity of the rotating element action unit at the last moment; J o Represents the equivalent moment of inertia of the output component of the rotating element action unit, M o It represents the equivalent moment of the output member of the rotating element action unit;

[0029] When the meta-action unit is a mobile meta-action unit, based on formula (3.4), we can get the mobile meta-action unit at t i The speed at the moment v o (t i ), the specific expression is as follows:

[0030]

[0031] In formula (3.6), v(t i-1 ) represents the speed of the mobile action unit at the last moment; F o represents the equivalent force of the output of the mobile action unit, m oIt represents the equivalent mass of the output of the mobile action unit;

[0032] Introduce the velocity state representation parameter Θ of the meta-action unit 2 ,

[0033] When the meta-action unit is a rotational meta-action unit, the specific expression is as follows:

[0034]

[0035] When the meta-action unit is a mobile meta-action unit, the specific expression is as follows:

[0036]

[0037] In formulas (3.7) and (3.8), f t , t They represent the change state of the speed of the element action unit caused by friction, wear and deformation, ω t Represents the angular velocity at the current running time, ω 0 Represents the angular velocity at the time of initial normal operation; π t represents the energy loss of the elementary action unit, T t Indicates the temperature at the current operating time, T 0 Indicates the temperature at the time of initial normal operation; δ Δt Indicates the speed stability of the element action unit within the time period Δt, represents the maximum angular velocity within the time period Δt, represents the minimum angular velocity within the time period Δt; v t Indicates the speed at the current running time, v 0 Indicates the speed at the time of initial normal operation; is the maximum operating speed in the time period Δt, is the minimum operating speed during the time period Δt;

[0038] According to the relationship between speed and position, the meta-action unit position model is constructed based on equations (3.5) and (3.6);

[0039] When the elementary action unit is a rotation elementary action unit, at time t i Location The expression is as follows:

[0040]

[0041] In formula (3.9), Indicates the position of the rotating element action unit at the previous moment;

[0042] When the meta-action unit is a mobile meta-action unit, at time ti The position s(t i ) is as follows:

[0043]

[0044] In formula (3.10), s(t i-1 ) represents the position of the mobile action unit at the previous moment;

[0045] Introduce the position state representation parameter Θ of the meta-action unit 3 , which is expressed as follows:

[0046] Θ 3 ={SD T} (3.11)

[0047] In formula (3.11), SD T Indicates the distance between the position of the meta-action unit at the current moment and the position during normal initial operation;

[0048] The characterization parameter Θ of the running state of the physical twin model is obtained. The specific expression is as follows:

[0049] Θ={Θ 1 ,Θ 2 ,Θ 3} (3.12)

[0050] Furthermore, the step S3 specifically includes:

[0051] The characterization parameters are regarded as random processes and algorithm iteration is used to achieve real-time updates to obtain updated characterization parameter values, which are then updated accordingly in the physical twin model to obtain a real-time updated physical twin model.

[0052] Furthermore, the steps of dynamically updating the characterization parameters are as follows:

[0053] Obtain the real-time data of the actual operation of the meta-action unit, input the real-time data into the physical twin model in step S2 for response, and obtain the consistency function MAU based on the response and real-time data e (f), the specific expression of the consistency function is as follows:

[0054] MAU e (f) = f PE (t)-f VE (t)

[0055] Among them, f VE (t) represents the value of the parameter representing the operation status of the physical twin model over time; f PE (t) represents the actual value of the physical entity at the corresponding time;

[0056] The particle swarm algorithm is used to solve the minimum value of the consistency function, and the characterization parameter value corresponding to the obtained minimum value is used as the updated characterization parameter value to complete the dynamic update of the characterization parameter.

[0057] Furthermore, the step S4 specifically includes:

[0058] S4.1. Obtain historical data of a preset time period connected to the current moment and a time tag corresponding to each historical data, input the historical data into the physical twin model of the current moment after real-time update, and the physical twin model of the current moment responds to obtain a response value corresponding to the physical twin model of the current moment, and then obtain the error between the response value and the historical data; each error corresponds to a time tag;

[0059] S4.2. All errors and corresponding time labels obtained in step S4.1 are used as training data and normalized. The normalized training data is trained and learned using a neural network to obtain a digital twin behavior model, which is used to predict the error value at the current moment.

[0060] Furthermore, the digital twin response value of the digital twin model in step S5 is obtained according to the following steps:

[0061] S5.1. Acquire the real-time data of the actual operation process of the meta-action unit and the time tag corresponding to the real-time data;

[0062] S5.2. Input the real-time data into the physical twin model at the current moment to respond, and obtain the response value of the physical twin model at the current moment;

[0063] S5.3, input the time tag corresponding to the real-time data into the digital twin behavior model obtained in step S4.2 to obtain the predicted value of the output error of the physical twin model at the current moment;

[0064] S5.4. Use the predicted value obtained in step S5.3 to correct the response value obtained in step S5.2. After correction, the response value of the physical twin model at the current moment is the digital twin response value of the digital twin model at the current moment.

[0065] Compared with the prior art, the present invention has the following beneficial effects:

[0066] 1. A method for constructing a digital twin model of an electromechanical system based on a meta-action unit of the present invention decomposes the electromechanical system into several meta-action units based on the meta-action unit theory, and constructs a physical twin model and a digital twin behavior model for the meta-action unit based on the operating status. The digital twin behavior model is used to dynamically perceive the operating status of the meta-action unit and correct random errors in the subsequent updating process of the physical twin model, and to update the physical twin model in real time. In this way, firstly, by updating the physical twin model in real time, the response difference between the physical twin model and the physical entity is shortened; secondly, the digital twin behavior model is used to compensate for the errors in the physical twin model, so as to realize the correction of the response of the physical twin model at the current moment; the present invention realizes the real-time reflection of the virtual model to the physical model, thereby realizing the high fidelity of the digital twin.

[0067] 2. The present invention provides a method for constructing a digital twin model of an electromechanical system based on a meta-action unit. Based on the characterization parameters of the operating state of the introduced physical twin model, it is regarded as a random process, and these parameters are subsequently updated in real time using algorithm iteration, thereby realizing real-time update of the physical entity to the physical twin model and shortening the response difference of the physical twin model to real-time data. It has the advantages of short iteration time, high accuracy and simple implementation process.

[0068] 3. The present invention provides a method for constructing a digital twin model of an electromechanical system based on a meta-action unit. According to the characteristic that digital twins can generate a large amount of data, the present invention proposes a data-driven digital twin behavior model; specifically, a bidirectional long short-term memory network is used as the residual learning model of the meta-action unit physical twin model to train and learn the normalized error data, thereby achieving a better error correction effect. BRIEF DESCRIPTION OF THE DRAWINGS

[0069] Figure 1 It is a flow chart of a method for constructing a digital twin model of an electromechanical system based on a meta-action unit of the present invention;

[0070] Figure 2 Schematic diagram of the equivalent vibration model of the meta-action unit;

[0071] Figure 3 Schematic diagram of the equivalent velocity model of the mobile action unit;

[0072] Figure 4 It is a schematic diagram of the equivalent velocity model of the rotation element action unit;

[0073] Figure 5 Flowchart for real-time updating of physical twin models using particle swarm algorithm;

[0074] Figure 6 It is a structural block diagram of the digital twin framework in the present invention;

[0075] Figure 7 This is a decomposition diagram obtained by performing FMA functional decomposition on a certain model of industrial robot in a specific embodiment;

[0076] Figure 8 It is a trajectory response curve diagram of the physical twin model and the physical entity in a specific embodiment;

[0077] Fig. 9 It is a trajectory response curve diagram after being updated by the particle swarm algorithm in a specific embodiment;

[0078] Fig.10 It is a graph showing the error variation of Bi-LSTM in a specific embodiment;

[0079] Fig.11 is a response curve diagram of the digital twin model in a specific embodiment;

[0080] Fig.12 It is a comparison curve diagram before and after Bi-LSTM error compensation in a specific embodiment. DETAILED DESCRIPTION

[0081] In order to make the purpose, technical scheme and advantages of the embodiments of the present invention clearer, the technical scheme in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. The components of the embodiments of the present invention generally described and shown in the drawings here can be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present invention provided in the drawings is not intended to limit the scope of the claimed invention, but merely represents selected embodiments of the present invention. Based on the embodiments in the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work belong to the scope of protection of the present invention.

[0082] It should be noted that similar numbers and letters represent similar items in the following drawings, so once an item is defined in one drawing, it does not need to be further defined and explained in the subsequent drawings. In the description of the present invention, it should be noted that the orientation or position relationship indicated by the terms "center", "upper", "lower", "left", "right", "vertical", "horizontal", "inside", "outside", etc. is based on the orientation or position relationship shown in the drawings, or the orientation or position relationship in which the invention product is usually placed when used, which is only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as a limitation of the present invention. In addition, the terms "first", "second", "third", etc. are only used to distinguish the description, and cannot be understood as indicating or implying relative importance. In addition, the terms "horizontal", "vertical", etc. do not mean that the components are absolutely horizontal or suspended, but can be slightly tilted. For example, "horizontal" only means that its direction is more horizontal than "vertical", and does not mean that the structure must be completely horizontal, but can be slightly tilted. In the description of the present invention, it is also necessary to explain that, unless otherwise clearly specified and limited, the terms "set", "install", "connect", and "connect" should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection, or it can be indirectly connected through an intermediate medium, or it can be the internal communication of two elements. For ordinary technicians in this field, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0083] The specific implementation modes of the present invention will be further described in detail below in conjunction with the accompanying drawings.

[0084] See also Figure 1 , a method for constructing a digital twin model of a mechatronic system based on a meta-action unit, comprising the following steps:

[0085] S1. Use the FMA functional decomposition method to decompose the electromechanical system layer by layer according to the "function-motion-action" mapping route to obtain several meta-action units. The meta-action units only contain two types of the most basic motion units, namely, the moving meta-action units and the rotating meta-action units; and further perform functional analysis on the meta-action units and decompose them into several lowest agreed level units. The lowest agreed level units include input parts, support parts, intermediate parts, fasteners and output parts;

[0086] S2. Construct a physical twin model for the meta-action unit based on the physical mechanism, and obtain the characterization parameters of the running state of the physical twin model;

[0087] S3, updating the physical twin model obtained in step S2 in real time to obtain a real-time updated physical twin model;

[0088] S4. Obtain historical data during the actual operation of the meta-action unit, and build a digital twin behavior model based on the historical data and the real-time updated physical twin model, wherein the digital twin behavior model is used to correct random errors in the real-time update process of the physical twin model;

[0089] S5. The digital twin behavior model obtained in step S4 is combined with the real-time updated physical twin model to construct the digital twin model. The response value of the real-time data output by the real-time updated physical twin model during the actual operation of the meta-action unit is corrected by the digital twin behavior model, which is the digital twin response value of the digital twin model to the real-time data.

[0090] A method for constructing a digital twin model of an electromechanical system based on a meta-action unit of the present invention decomposes the electromechanical system from top to bottom using the function-motion-action decomposition method to obtain the most basic meta-action unit; a physical twin model and a digital twin behavior model are constructed for the meta-action unit based on the operating status; the digital twin behavior model is used to dynamically perceive the operating status of the meta-action unit and correct random errors in the subsequent physical twin model update process, and to update the physical twin model in real time; in this way, firstly, by updating the physical twin model in real time, the response difference between the physical twin model and the physical entity is shortened; secondly, the digital twin behavior model is used to compensate for the errors in the physical twin model to realize the correction of the response of the physical twin model at the current moment; the present invention realizes the real-time reflection of the virtual model on the physical model, thereby realizing the high fidelity of the digital twin.

[0091] Wherein, the physical twin model includes a meta-action unit trajectory model, a meta-action unit velocity model and a meta-action unit position model;

[0092] 1) Trajectory modeling of meta-action units

[0093] Since vibration is the main factor causing the actual running trajectory of the meta-action unit to change, the present invention uses a first-order vibration model to represent the trajectory change of the meta-action unit;

[0094] See also Figure 2 , the element action unit is abstracted into a mass block. According to the D'Alembert principle, the vibration equation of the mass block is expressed as follows:

[0095] m u y″(t)+c u y′(t)+k u y(t)=F(t)

[0096] Among them, m u represents the mass of the mass block, c urepresents the equivalent damping, k u represents the equivalent stiffness; t represents the running time of the element action unit; y(t) represents the displacement caused by the influence of force; F(t) represents the force in the vibration direction;

[0097] The vibration equation of the mass block is solved to obtain the trajectory model of the element action unit. The specific expression is as follows:

[0098]

[0099] Among them, C 1 and C 2 is a constant; μ n Indicates the vibration frequency at the time of initial normal operation, F 0 It represents the force in the vibration direction at the time of initial normal operation; λ represents the vibration frequency ratio of the elementary action unit, μ represents the external force frequency; Y represents the static displacement ratio of the elementary action unit; Indicates phase;

[0100] It can be seen from the expression of the meta-action unit trajectory model that the trajectory of the meta-action unit changes continuously with the increase of running time, and the main factors affecting its change include vibration force, vibration frequency, and its own equivalent mass, stiffness and damping. Due to the limitations of measurement conditions and the uncertainty of the internal parts of the specific meta-action unit, the trajectory change of the meta-action unit is difficult to obtain by direct measurement by sensors, and it is also difficult to calculate in real time under running conditions; in response to these problems, the present invention introduces meta-action unit trajectory state characterization parameters to represent the trajectory state of the meta-action unit. Through the analysis of the failure mechanism of the meta-action unit, the main internal factors affecting the running trajectory of the meta-action unit are the wear and cracks of the parts, and the crack propagation. The crack degradation of mechanical parts can be described by the real and imaginary parts of the first-order vibration signal, and the wear between parts is related to the frequency and amplitude of the vibration signal. The meta-action unit trajectory state characterization parameter Θ is obtained 1 , which is expressed as follows:

[0101]

[0102] Among them, Y t represents the imbalance amount when the meta-action unit is running, R t Represents the real part of the first-order vibration at the current moment, I t Represents the imaginary part of the first-order vibration at the current moment, R 0 represents the real part of the initial normal operation time, I 0 represents the imaginary part of the initial normal operation time; λ t Indicates the vibration frequency μ of the element action unit at the current moment t The vibration frequency μ at the initial normal operation0 The ratio of A t Indicates the vibration amplitude a of the element action unit at the current moment t The vibration amplitude a at the initial normal operation 0 difference.

[0103] 2) Speed ​​modeling of meta-action units

[0104] The velocity model of the electromechanical system can usually be expressed by the work done, and the same is true for the elementary action unit. According to the law of conservation of energy, the kinetic energy increment ΔE in time Δt is equal to the work done by the external force of the elementary action unit. The specific equation is as follows:

[0105] ΔE=ΔW=Pdt

[0106] Among them, P represents the instantaneous power of the resultant external force. Further expansion can be used to obtain the expression of the velocity model of the elementary action unit, which is as follows:

[0107]

[0108] Where n represents the number of parts that produce motion in the elementary action unit; m n represents the mass of the nth part; O represents the center of mass of the nth part; J On represents the moment of inertia of the nth part relative to the center of mass; v On Represents the velocity vector of the nth part relative to the center of mass; ω n 、F d 、F r 、M d 、M r They represent the angular velocity, driving force, resistance, driving torque and resistance torque of the nth part respectively, all of which are vectors;

[0109] According to the above meta-action unit speed model, draw the meta-action unit speed equivalent model, such as Figure 3 and Figure 4 As shown, since the meta-action unit only contains two types of the most basic motion units, namely the moving meta-action unit and the rotating meta-action unit, they need to be further analyzed and processed separately;

[0110] When the meta-action unit is a rotational meta-action unit, the rotational meta-action unit is obtained based on the meta-action unit speed model at t i Angular velocity ω at time o (t i ), the specific expression is as follows:

[0111]

[0112] Among them, ω(t i-1 ) represents the angular velocity of the rotating element action unit at the last moment; J oRepresents the equivalent moment of inertia of the output component of the rotating element action unit, M o It represents the equivalent moment of the output member of the rotating element action unit;

[0113] When the meta-action unit is a mobile meta-action unit, the speed of the mobile meta-action unit at t is obtained based on the meta-action unit speed model. i The speed at the moment v o (t i ), the specific expression is as follows:

[0114]

[0115] Among them, v(t i-1 ) represents the speed of the mobile action unit at the last moment; F o represents the equivalent force of the output of the mobile action unit, m o It represents the equivalent mass of the output of the mobile action unit;

[0116] It can be seen here that the speed of the meta-action unit changes continuously with the increase of running time, and the main factors affecting its change are equivalent torque, equivalent moment of inertia, equivalent force and equivalent mass. Although the speed of the meta-action unit can be directly measured by sensors in some cases, the "health value" of its speed state cannot be directly measured by sensors. In the prior art, it is usually only possible to track the health of its current state in real time through the method of "collection → offline processing → output judgment results"; therefore, the present invention introduces the meta-action unit speed state characterization parameter θ 2 To indicate the speed state of the meta-action unit when it is running.

[0117] Likewise,

[0118] When the meta-action unit is a rotational meta-action unit, the specific expression is as follows:

[0119]

[0120] When the meta-action unit is a mobile meta-action unit, the specific expression is as follows:

[0121]

[0122] Among them, f t , t They represent the change state of the speed of the element action unit caused by friction, wear and deformation, ω t Represents the angular velocity at the current running time, ω 0 Represents the angular velocity at the time of initial normal operation; π t represents the energy loss of the elementary action unit, T t Indicates the temperature at the current operating time, T 0Indicates the temperature at the time of initial normal operation; δ Δt Indicates the speed stability of the element action unit within the time period Δt, represents the maximum angular velocity within the time period Δt, represents the minimum angular velocity within the time period Δt; v t Indicates the speed at the current running time, v 0 Indicates the speed at the time of initial normal operation; is the maximum operating speed in the time period Δt, is the minimum operating speed during the time period Δt.

[0123] 3) Motion position modeling of meta-action units

[0124] According to the relationship between speed and position, a meta-action unit position model is constructed;

[0125] When the elementary action unit is a rotation elementary action unit, at time t i Location The expression is as follows:

[0126]

[0127] in, Indicates the position of the rotating element action unit at the previous moment; ω 0 represents the angular velocity at the time of initial normal operation;

[0128] When the meta-action unit is a mobile meta-action unit, at time t i The position s(t i ) is as follows:

[0129]

[0130] Among them, s(t i-1 ) represents the position of the mobile action unit at the last moment; v 0 Indicates the speed at the time of initial normal operation;

[0131] As mentioned above, the position of the meta-action unit can usually be directly measured by a sensor. However, the “health value” of its position needs to be further obtained by calculating the position data. Therefore, the present invention introduces a meta-action unit position state characterization parameter θ 3 , in order to use digital twins for real-time updates, the expression is as follows:

[0132] Θ 3 ={SD T}

[0133] Among them, SD TIndicates the distance between the position of the meta-action unit at the current moment and the position during normal initial operation;

[0134] Based on the above modeling and analysis of the running state of the meta-action unit, the characterization parameter Θ of the running state of the physical twin model is obtained. The specific expression is as follows:

[0135] Θ={Θ 1 ,Θ 2 ,Θ 3}

[0136] On the one hand, these parameter sets are updated in real time through the twin data generated by the sensors in the digital twin system to achieve high fidelity of the physical twin model and the physical entity; on the other hand, these parameter sets also reflect in real time whether the physical entity of the meta-action unit is healthy. By observing the trend of parameter changes and using fault prediction technology, it is possible to predict when a fault will occur and take corresponding measures in advance. It should be noted here that for a specific meta-action unit, the subsequent characterization parameter Θ that needs to be updated is not fixed and uniquely determined, and needs to be established and selected according to the requirements of the actual working conditions and the limited means of monitoring.

[0137] In this way, the meta-action unit is digitally twinned based on the physical mechanism to obtain a physical twin model. The purpose is to perform a dynamic description based on the static description of the meta-action unit and use the twin data to dynamically update the operating status of the meta-action unit.

[0138] Wherein, the step S3 specifically includes:

[0139] The characterization parameters are regarded as random processes and algorithm iteration is used to achieve real-time updates to obtain updated characterization parameter values, which are then updated accordingly in the physical twin model to obtain a real-time updated physical twin model.

[0140] The present invention regards the introduced characterization parameters of the operating state of the physical twin model as a random process, and subsequently uses algorithm iteration to update these parameters in real time, thereby realizing real-time update of the physical entity to the physical twin model and shortening the response difference of the physical twin model to real-time data.

[0141] Generally speaking, due to the complex environment in which physical entities are located, it is difficult to characterize them with virtual simulation models. In the prior art, neural networks, Gaussian models, etc. are used for simulation, and actual data is continuously used for approximation, and finally a complete virtual model is obtained to characterize the physical entity; although the results obtained by this method are accurate, the implementation process is complicated and takes a long time, and when the state of the physical entity changes, it needs to be recalculated, and the workload is very large. Another way is to use existing functions to approximate the mechanism model. This update method has the disadvantages of low accuracy and slow speed. In this way, the characterization parameters are updated using algorithm iteration. In addition to short iteration time and high accuracy, the implementation process is simple, which can effectively overcome the above shortcomings.

[0142] Furthermore, the key lies in the selection of the update algorithm. Based on the modeling and analysis of the operating state of the meta-action unit in the present invention, the steps of dynamically updating the characterization parameters are as follows:

[0143] The real-time data is input into the physical twin model in step S2 for response, and the consistency function MAU is obtained based on the response and the real-time data. e (f), the specific expression of the consistency function is as follows:

[0144] MAU e (f) = f PE (t)-f VE (t)

[0145] Among them, f PE (t) represents the actual value of the physical entity at the corresponding time; f VE (t) represents the value of the parameter representing the running state of the physical twin model over time. The parameter representing the state of the meta-action unit trajectory Θ 1 For example, f VE (t) includes Y t The distance value in ,λ t The ratio of the vibration frequencies in t The difference in vibration amplitude in is then used to obtain the displacement generated by the physical twin model simulation; f PE (t) refers to the actual displacement produced by the physical entity; therefore, the consistency function compares the actual value of the physical entity and the simulated value produced by the physical twin model to see how much difference there is between the two.

[0146] The particle swarm algorithm is used to solve the minimum value of the consistency function, and the characterization parameter value corresponding to the obtained minimum value is used as the updated characterization parameter value to complete the dynamic update of the characterization parameter.

[0147] Since the purpose of digital twins is to reflect the physical entity as realistically as possible, the response of the twin model should be as close as possible to the response of the physical entity, that is, the difference between the two should be as small as possible, thus converting the real-time update problem of digital twins into the real-time update of MAU. e (f) Minimum solution (MAU e (f)>0), that is, the problem of finding the extreme value of a multivariate function. The particle swarm algorithm is used in the present invention to solve the problem. The global search capability of the intelligent algorithm is used to search for the global optimal point in a complex space to update the characterization parameters of the meta-action unit during transient response, and complete the real-time update of the physical twin model. The particle swarm optimization algorithm uses the mutual cooperation and competition between particles to search for the optimal point in the entire space. It has the characteristics of fast speed, convenient solution, and high accuracy. Compared with other evolutionary algorithms, the particle swarm algorithm does not have the operations of evolutionary model operators such as crossover and selection. It moves in multidimensional space at a set particle speed and searches for its own historical optimal position and the optimal position of the field history, and continuously updates and iterates the solution to obtain the final optimal value. In addition, the particle swarm algorithm can dynamically track and correct the search strategy in multidimensional space, that is, it has a unique memory function, which can be well applied to the real-time update of the physical twin model in the present invention.

[0148] When implemented, the function MAU e (f) is used as the corresponding fitness function to obtain the particle position value corresponding to the global minimum error, so as to realize the dynamic update of the characterization parameters. Figure 5 , the detailed process is as follows:

[0149] (1) Initialize the parameters of the particles, including the initialization of the learning factor, search space, number of iterations, and search speed in the particle swarm algorithm, as well as the initialization of the characterization parameters. The initialization of the characterization parameters mainly includes obtaining the mean of the characterization parameters through the initial normal operation state, and then giving the variation range of the evaluation parameters based on engineering experience or experts;

[0150] (2) Evaluate the particles; calculate the fitness value of the particles through the fitness function. If the current fitness value is better than the previous extreme value, update the position and size of the extreme value.

[0151] (3) Update the particles; Update the particles to obtain new position and velocity values;

[0152] (4) Result detection: If the convergence accuracy meets the standard or the maximum number of iterations is reached, stop updating and output the results; otherwise, continue running.

[0153] In this way, by comparing the actual operating state of the physical entity of the meta-action unit with the response of the physical twin model to real-time data, and continuously updating the characterization parameter Θ in real time using the particle swarm algorithm, the virtual model can reflect the physical model in real time and achieve high fidelity of the digital twin. In addition, the updated characterization parameters can also reflect the degradation trend of the meta-action unit. Further prediction can be carried out to predict the failure time of the meta-action unit in advance so that corresponding measures can be taken in advance.

[0154] Among them, see Figure 1 , the step S4 specifically includes:

[0155] S4.1. Obtain historical data of a preset time period connected to the current moment and a time tag corresponding to each historical data, input the historical data into the physical twin model of the current moment after real-time update, and the physical twin model of the current moment responds to obtain a response value corresponding to the physical twin model of the current moment, and then obtain the error between the response value and the historical data; each error corresponds to a time tag;

[0156] S4.2. All errors and corresponding time labels obtained in step S4.1 are used as training data and normalized. The normalized training data is trained and learned using a neural network to obtain a digital twin behavior model, which is used to predict the error value at the current moment.

[0157] Since the physical twin model is the running state of the meta-action unit of the twin model in the physical dimension, but due to the existence of various random errors, the behavior process of the meta-action unit is not fully described, so the present invention constructs a digital twin behavior model to correct the error of the physical twin model. The digital twin behavior model learns the law of historical data errors, predicts the error at the current moment, and then corrects and compensates the physical twin model to obtain the digital twin response value at the current moment. However, the actual running state of the meta-action unit is very complex and difficult to model, so it is necessary to simplify it. In the prior art, many scholars use proxy models or causal models to track the running behavior of the system, and have achieved relatively good results. According to the characteristics of digital twins that can generate a large amount of data, the present invention proposes a data-driven digital twin behavior model, which is combined with the digital twin model of the meta-action unit based on the physical mechanism to obtain a high-fidelity digital twin.

[0158] During implementation, the present invention adopts a bidirectional long short-term memory network (Bidirectional Long Short Term Memory Network, Bi-LSTM) as the residual learning model of the meta-action unit physical twin model to train and learn the normalized data. Bi-LSTM is composed of two independent long short-term memory networks (Long Short Term Memory Network, LSTM), and the input data is feature extracted in both the forward sequence and reverse sequence directions, and finally spliced ​​together for output. Bi-LSTM has the characteristics of LSTM, it can selectively memorize the characteristics of data, and selectively forget, which can effectively solve the problem of information loss in ordinary recurrent neural networks (Recurrent Neural Network, RNN). In addition, Bi-LSTM has the characteristics of forward and reverse feature extraction, which can make good use of the changing characteristics of the physical twin model before and after the state, and perform bidirectional learning to achieve better error correction effect.

[0159] Among them, the digital twin response value of the digital twin model in step S5 is obtained as follows:

[0160] S5.1. Acquire the real-time data of the actual operation process of the meta-action unit and the time tag corresponding to the real-time data;

[0161] S5.2. Input the real-time data into the physical twin model at the current moment to respond, and obtain the response value of the physical twin model at the current moment;

[0162] S5.3, input the time tag corresponding to the real-time data into the digital twin behavior model obtained in step S4.2 to obtain the predicted value of the output error of the physical twin model at the current moment;

[0163] S5.4. Use the predicted value obtained in step S5.3 to correct the response value obtained in step S5.2. After correction, the response value of the physical twin model at the current moment is the digital twin response value of the digital twin model at the current moment.

[0164] In this way, by building a digital twin model based on the digital twin behavior model and the real-time updated physical twin model, a highly guaranteed digital twin response value can be obtained. The use of the behavior model is further explained here. For example, if the digital twin response value at the current moment 11 is required, when building the digital twin behavior model, the historical data at moments 1-10 are selected to execute steps S4.1 and S4.2; for the current moment 11, the real-time data at the current moment 11 is input into the digital twin behavior model, and the output is the predicted value of the error at the current moment 11 by the digital twin behavior model. The predicted value is combined with the response value of the physical twin model at the current moment 11, which is the digital twin response value at the current moment.

[0165] See also Figure 6 When implemented, the steps S1 and S2 also include building a digital twin framework based on the meta-action unit, and the digital twin framework includes the meta-action unit physical entity MAU PE , Meta-Action Unit Virtual Mirror MAU VM , Meta Action Unit Service MAU SS , Meta-action unit twin data MAU DD And the yuan action unit connection system MAU CN ;

[0166] Meta-action unit physical entity MAU PE It is the true representation of the physical state of the meta-action unit, and is the entity unit obtained by decomposing the whole machine through FMA. According to the definition of meta-action unit, MAU PE In addition to the lowest agreed hierarchical unit of the meta-action unit (five internal parts, input parts, middle parts, output parts, support parts and fasteners), it also includes the influence of the input of the upstream meta-action unit and the influence of the resistance of the downstream meta-action unit; this can achieve high fidelity of the meta-action unit, that is, a complete MAU PE Includes basic information of the five internal components and external excitation information of the meta-action unit. PE It is the foundation of the digital twin of the entire meta-action unit. It generates physical information through the drive of the service system and transmits real-time data through the connection system for analysis and decision-making.

[0167] The meta-action virtual mirror MAU VM MAU PE High-fidelity digital models, based on MAU PE The basic structure and information of MAU VM Including the geometric information, constraint feature information and real-time response behavior of each meta-action unit of the meta-action unit physical entity, meta-action virtual mirror MAU VM It also includes the historical rules of its operation summarized by the state of the operation of the meta-action unit; the meta-action virtual mirror MAUVM For meta-action unit physical entity MAU PE Provide a complete reflection.

[0168] The meta-action unit serves MAU SS Used to provide services to each model during the movement of the meta-action unit; taking meta-action unit monitoring as an example, it includes MAU VM Modeling and simulation services for MAU DD Data related processing services (storage, packaging, mining, etc.) and MAU CN Provide services such as data collection and transmission;

[0169] The meta-action unit twin data MAU DD Includes all data in the digital twin process; in the meta-action unit digital twin framework, MAU DD Specifically include: ①MAU PE Dynamic and static data generated; ②MAU VM Dynamic and static data generated; ③MAU SS Information about the processing methods of the generated algorithms and models; ④MAU SS Algorithm model library and rule library; ⑤MAU DD The derivative data of various data are the data obtained after processing, classification and integration of the above three types of data, which are used for comprehensive utilization.

[0170] The meta-action unit is connected to the system MAU CN Used to implement the meta-action unit physical entity MAU PE , Meta-action virtual image MAU VM , Meta Action Unit Service MAU SS and MAU of twin data of meta-action units DD The data between them are transmitted to each other in real time to form an internal closed-loop system.

[0171] Next, a method for constructing a digital twin model of an electromechanical system based on a meta-action unit of the present invention is applied to a certain model of an industrial robot as a specific embodiment for implementation verification, which specifically includes the following steps:

[0172] S1. Perform FMA function decomposition on the industrial robot and obtain 20 meta-action units, such as Figure 7 As shown, taking the J5 belt drive wheel 5 rotation element action A12 (hereinafter referred to as element action unit A12) as an example, the rotation element action unit is further decomposed into five parts, as shown in the following table:

[0173]

[0174]

[0175] Specifically, taking meta-action A12 as an example, a digital twin framework of the meta-action unit is constructed; in this specific embodiment, Solidworks is used to model the meta-action unit to construct the digital twin framework;

[0176] After using Solidworks to build the geometric model of the meta-action unit A12, it was imported into the Simscape Multibody module in the Matlab&Simulink toolbox for simulation; this module not only retains the size, assembly and other information of the three-dimensional entity in Solidworks, but also uses Matlab's powerful data toolbox to build its mechanism model and dynamically simulate it, and can also set the wear state of the internal parts of the meta-action unit, the driving torque and resistance torque, etc.; For the meta-action unit connection, service and data module implementation, the purpose of this section is to monitor its health status, mainly based on acceleration, temperature, vibration, and angle state characteristics for data acquisition and data transmission. The signal measured by the sensor is filtered and then transmitted to the service system. The service system is mainly composed of three-dimensional modeling software Solidworks, data processing software Matlab\Simulink, the robot's real-time servo control system, and real-time data acquisition and display modules; data access and transmission mainly use the OPCUA (OPC Unified Architecture) server and Socket communication method, and are transmitted through the local area network (LAN). The stored data can be called through Matlab;

[0177] S2&S3, due to the high precision requirement of the robot, according to the running state characterization parameters of the meta-action unit combined with the enterprise experience, the characterization parameters that best represent the running state of the unit, that is, the trajectory state characterization parameter Θ={λ t ,a t} to indicate the operating status of the current unit;

[0178] The vibration signal is used to reflect the trajectory state of the element action unit A12. The vibration data generated by the sensor is transmitted to the service system. The real-time data is processed and used as the input of the physical twin model. Thus, the trajectory response of the initial physical twin model and the trajectory response of the physical entity (real-time data) can be obtained, such as Figure 8 As shown, it can be seen that there is a large gap between the response of the physical twin model and the real-time data. At this time, the physical twin model does not have high fidelity, so it needs to be updated. Since the data of the vibration process has a large amount of nonlinearity, this embodiment mainly targets the first-order frequency and amplitude of the vibration signal. Correspondingly, the specific expression of the fitness function updated by the particle swarm algorithm is as follows:

[0179] {MAU e (f)} min =f PE (μ t ,a t )-f VE (μ t ,a t )

[0180] Among them, μ t Indicates the frequency of the current real-time operation of the meta-action unit A12, a t is the amplitude of the current real-time operation of the meta-action unit A12;

[0181] Matlab is used to write the particle swarm algorithm model. The particle group is set to 100, the maximum number of iterations is 300, and the learning factor is 1.5. The update results after the particle swarm algorithm are as follows: Fig. 9 As shown, from Fig. 9 It can be seen that due to the interference of various random errors, the real-time update using the particle swarm algorithm cannot achieve good results and there are still large errors;

[0182] S4. Based on the above steps of building a digital twin behavior model, the digital twin behavior model is added to correct the results after the particle swarm algorithm iteration in real time; the error between the current physical twin model and the historical data is used for Bi-LSTM learning to obtain the following: Fig.10 The error change curve of Bi-LSTM is shown. From the error tracking results, it can be seen that the learning result of Bi-LSTM is good, and the final root mean square error RMSE is less than 0.05;

[0183] S5. The output value of the digital twin model obtained by combining the real-time updated physical twin model with the digital twin behavior model is compared with the actual output (real-time data) of the physical entity. Fig.11 and Fig.12 It can be seen that the digital twin model provided by the present invention can better follow the trajectory state of the physical entity.

[0184] Further, by updating the updated amplitude and frequency in real time, according to Θ 1 The formula is used to calculate the characterization parameters of the trajectory state of the current meta-action unit A12. Some results are shown in the following table:

[0185]

[0186] It can be seen that the values ​​of the trajectory characterization parameters are small and fluctuate within the specified range, indicating that the trajectory state of the meta-action unit is in a healthy operating state. The health of the current meta-action unit can be tracked in real time through the characterization parameters.

[0187] Through the above-mentioned specific implementation examples for the meta-action unit A12, the results obtained show that the method for constructing a digital twin model of an electromechanical system based on a meta-action unit of the present invention has certain practical significance and can achieve high fidelity of the digital twin of the meta-action unit.

[0188] Finally, it should be noted that the above embodiments are only used to illustrate the technical solution of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solution of the present invention can be modified or replaced by equivalents without departing from the purpose and scope of the technical solution of the present invention, which should be included in the scope of the claims of the present invention.

Claims

1. A method for constructing a digital twin model of an electromechanical system based on a meta-action unit, characterized in that: The following steps are involved: S1. Use the FMA functional decomposition method to decompose the electromechanical system layer by layer according to the "function-motion-action" mapping route to obtain several meta-action units. The meta-action units only contain two types of the most basic motion units, namely, the moving meta-action unit and the rotating meta-action unit; S2. Construct a physical twin model for the meta-action unit based on the physical mechanism, and obtain the characterization parameters of the running state of the physical twin model; S3, updating the physical twin model obtained in step S2 in real time to obtain a real-time updated physical twin model; S4. Obtain historical data during the actual operation of the meta-action unit, and build a digital twin behavior model based on the historical data and the real-time updated physical twin model, wherein the digital twin behavior model is used to correct random errors in the real-time update process of the physical twin model; S5. The digital twin behavior model obtained in step S4 is combined with the real-time updated physical twin model to construct the digital twin model. The response value of the real-time data output by the real-time updated physical twin model during the actual operation of the meta-action unit is corrected by the digital twin behavior model, which is the digital twin response value of the digital twin model to the real-time data.

2. According to claim 1, a method for constructing a digital twin model of an electromechanical system based on a meta-action unit is characterized by: The physical twin model includes a meta-action unit trajectory model, a meta-action unit velocity model and a meta-action unit position model; The meta-action unit is abstracted into a mass block, and the meta-action unit trajectory model is constructed. The specific expression is as follows: In formula (3.1), y(t) represents the displacement caused by the force, t represents the running time of the element action unit; C1 and C2 are constants; parameters c u represents the equivalent damping, m u Represents the mass of the mass block, μ n Indicates the vibration frequency at the time of initial normal operation; Y represents the static displacement ratio of the element action unit, F0 represents the force in the vibration direction at the initial normal operation, k u represents the equivalent stiffness; λ represents the vibration frequency ratio of the element action unit, μ represents the external force frequency; Indicates phase; The meta-action unit trajectory state characterization parameter Θ1 is introduced, and its expression is as follows: In formula (3.2), Y t represents the imbalance amount when the meta-action unit is running, R t Represents the real part of the first-order vibration at the current moment, I t represents the imaginary part of the first-order vibration at the current moment, R0 represents the real part at the initial normal operation moment, and I0 represents the imaginary part at the initial normal operation moment; λ t Indicates the vibration frequency μ of the element action unit at the current moment t The ratio of the vibration frequency μ0 to the initial normal operation; A t Indicates the vibration amplitude a of the element action unit at the current moment t The difference between the vibration amplitude a0 at the time of initial normal operation; The velocity model of the element action unit is expressed by doing work. According to the law of conservation of energy, the kinetic energy increment ΔE in time Δt is equal to the work done by the resultant external force of the element action unit. The specific equation is as follows: ΔE=ΔW=Pdt (3.3) In formula (3.3), P represents the instantaneous power of the resultant external force. Further expanding formula (3.3), we can get the expression of the velocity model of the elementary action unit, which is as follows: In formula (3.4), n represents the number of parts that produce motion in the elementary action unit; m n represents the mass of the nth part; O represents the center of mass of the nth part; J On represents the moment of inertia of the nth part relative to the center of mass; v On Represents the velocity vector of the nth part relative to the center of mass; ω n 、F d 、F r 、M d 、M r They represent the angular velocity, driving force, resistance, driving torque and resistance torque of the nth part respectively, all of which are vectors; When the elementary action unit is a rotation elementary action unit, based on formula (3.4), the rotation elementary action unit is obtained at t i Angular velocity ω at time o (t i ), the specific expression is as follows: In formula (3.5), ω(t i-1 ) represents the angular velocity of the rotating element action unit at the last moment; J o Represents the equivalent moment of inertia of the output component of the rotating element action unit, M o It represents the equivalent moment of the output member of the rotating element action unit; When the meta-action unit is a mobile meta-action unit, based on formula (3.4), we can get the mobile meta-action unit at t i The speed at the moment v o (t i ), the specific expression is as follows: In formula (3.6), v(t i-1 ) represents the speed of the mobile action unit at the last moment; F o represents the equivalent force of the output of the mobile action unit, m o It represents the equivalent mass of the output of the mobile action unit; Introduce the velocity state characterization parameter Θ2 of the meta-action unit, When the meta-action unit is a rotational meta-action unit, the specific expression is as follows: When the meta-action unit is a mobile meta-action unit, the specific expression is as follows: In formulas (3.7) and (3.8), f t , t They represent the change state of the speed of the element action unit caused by friction, wear and deformation, ω t represents the angular velocity at the current operating time, ω0 represents the angular velocity at the initial normal operating time; π t represents the energy loss of the elementary action unit, T t represents the temperature at the current operating time, T0 represents the temperature at the initial normal operating time; δ Δt Indicates the speed stability of the element action unit within the time period Δt, represents the maximum angular velocity within the time period Δt, represents the minimum angular velocity within the time period Δt; v t represents the speed at the current running time, and v0 represents the speed at the initial normal running time; is the maximum operating speed in the time period Δt, is the minimum operating speed during the time period Δt; According to the relationship between speed and position, the meta-action unit position model is constructed based on equations (3.5) and (3.6); When the elementary action unit is a rotation elementary action unit, at time t i Location The expression is as follows: In formula (3.9), Indicates the position of the rotating element action unit at the previous moment; When the meta-action unit is a mobile meta-action unit, at time t i The position s(t i ) is as follows: In formula (3.10), s(t i-1 ) represents the position of the mobile action unit at the previous moment; The meta-action unit position state characterization parameter Θ3 is introduced, and its expression is as follows: Θ3={SD T } (3.11) In formula (3.11), SD T Indicates the distance between the position of the meta-action unit at the current moment and the position during normal initial operation; The characterization parameter Θ of the running state of the physical twin model is obtained. The specific expression is as follows: Θ={Θ1,Θ2,Θ3} (3.12).

3. According to claim 2, a method for constructing a digital twin model of an electromechanical system based on a meta-action unit is characterized by: The step S3 specifically includes: The characterization parameters are regarded as random processes and algorithm iteration is used to achieve real-time updates to obtain updated characterization parameter values, which are then updated accordingly in the physical twin model to obtain a real-time updated physical twin model.

4. The method for constructing a digital twin model of an electromechanical system based on a meta-action unit according to claim 3 is characterized in that: The steps for dynamically updating the characterization parameters are as follows: Obtain the real-time data of the actual operation of the meta-action unit, input the real-time data into the physical twin model in step S2 for response, and obtain the consistency function MAU based on the response and real-time data e (f), the specific expression of the consistency function is as follows: MAU e (f)=f PE (t)-f VE (t) Among them, f VE (t) represents the value of the parameter representing the operation status of the physical twin model over time; f PE (t) represents the actual value of the physical entity at the corresponding time; The particle swarm algorithm is used to solve the minimum value of the consistency function, and the characterization parameter value corresponding to the obtained minimum value is used as the updated characterization parameter value to complete the dynamic update of the characterization parameter.

5. A method for constructing a digital twin model of an electromechanical system based on a meta-action unit according to claim 4, characterized in that: The step S4 specifically includes: S4.

1. Obtain historical data of a preset time period connected to the current moment and a time tag corresponding to each historical data, input the historical data into the physical twin model of the current moment after real-time update, and the physical twin model of the current moment responds to obtain a response value corresponding to the physical twin model of the current moment, and then obtain the error between the response value and the historical data; each error corresponds to a time tag; S4.

2. All errors and corresponding time labels obtained in step S4.1 are used as training data and normalized. The normalized training data is trained and learned using a neural network to obtain a digital twin behavior model, which is used to predict the error value at the current moment.

6. The method for constructing a digital twin model of an electromechanical system based on a meta-action unit according to claim 5, characterized in that: The digital twin response value of the digital twin model in step S5 is obtained as follows: S5.

1. Acquire the real-time data of the actual operation process of the meta-action unit and the time tag corresponding to the real-time data; S5.

2. Input the real-time data into the physical twin model at the current moment to respond, and obtain the response value of the physical twin model at the current moment; S5.3, input the time tag corresponding to the real-time data into the digital twin behavior model obtained in step S4.2 to obtain the predicted value of the output error of the physical twin model at the current moment; S5.

4. Use the predicted value obtained in step S5.3 to correct the response value obtained in step S5.

2. After correction, the response value of the physical twin model at the current moment is the digital twin response value of the digital twin model at the current moment.