A device digital twin modeling and intelligent perception method and related components

Abstract

The application discloses a device digital twin modeling and intelligent sensing method and related components, and relates to the field of fault identification, wherein a simulation model is constructed according to actual operation data of a heat exchanger; flow velocity field data and temperature field data output by the simulation model are obtained according to an inlet temperature, an inlet flow velocity of a liquid input to a pipeline of the heat exchanger corresponding to the simulation model, and a fouling thickness inside the pipeline of the heat exchanger corresponding to the simulation model; a corresponding relationship between the fouling thickness and the temperature field data is determined according to flow velocity field data and temperature field data with a data amount proportion of a first preset proportion; and the fouling thickness of the heat exchanger is determined according to the actual liquid temperature of the pipeline of the heat exchanger and the corresponding relationship between the fouling thickness and the temperature field data. The simulation model is constructed according to the actual heat exchanger, the obtained fouling data is more accurate, and in addition, the flow velocity field data and the temperature field data with the data amount proportion of the first preset proportion randomly collected can reduce the calculation time of the data.

Description

Technical Field

[0001] This invention relates to the field of fault identification, and in particular to a digital twin modeling and intelligent sensing method and related components for equipment. Background Technology

[0002] Among the heat exchange equipment widely used in the petrochemical and power drive fields, shell-and-tube heat exchangers are highly regarded for their excellent heat transfer performance, simple and practical scaling design, and outstanding adaptability and operability. Due to their widespread application and crucial role, the safe operation of heat exchangers has become increasingly important, thus demanding higher reliability. If a heat exchange system experiences a decline in performance and the cause is not detected and diagnosed promptly and accurately, it can lead to serious consequences. This decline in performance is often caused by faults, especially soft faults in the heat exchange circulation system, such as scaling, corrosion, and biofilm, which are gradually accumulated by various physical factors and are difficult to detect, analyze, and diagnose in a timely manner. Furthermore, scaling often hides inside the equipment, making it difficult to detect or accurately perceive in a timely manner. A decrease in the heat exchanger's performance increases the pump's load capacity and can even lead to the paralysis of the entire circulation system, resulting in wasted energy and materials and significant losses to the industrial economy. Therefore, identifying internal faults in heat exchangers is an urgent problem to be solved. Summary of the Invention

[0003] The purpose of this invention is to provide a digital twin modeling and intelligent sensing method and related components for equipment. A simulation model is constructed based on the actual heat exchanger, and the liquid flow rate and scaling data are continuously adjusted to obtain more accurate scaling data. In addition, the flow rate field data and temperature field data are randomly collected in proportion to a first preset ratio, which can reduce the data calculation time.

[0004] To address the aforementioned technical problems, this invention provides a method for digital twin modeling and intelligent sensing of equipment, comprising:

[0005] A simulation model is constructed based on the actual operating data of the heat exchanger, including the dimensions of the heat exchanger's pipes.

[0006] Based on the inlet temperature and inlet velocity of the liquid in the pipes of the heat exchanger corresponding to the simulation model, and the scale thickness inside the pipes of the heat exchanger corresponding to the simulation model, the flow velocity field data and temperature field data output by the simulation model are obtained. The flow velocity field data includes the liquid flow velocity at each monitoring point in the simulation model, and the temperature field data includes the temperature at each monitoring point in the simulation model. The monitoring point represents a preset position inside the pipes of the heat exchanger corresponding to the simulation model.

[0007] The flow velocity field data and the temperature field data are randomly collected in proportion to a first preset ratio;

[0008] The correspondence between the scale thickness and the temperature field data is determined based on the collected flow field data and temperature field data in a first preset ratio.

[0009] The scale thickness of the heat exchanger pipes is determined based on the correspondence between the inlet temperature, outlet temperature, scale thickness, and temperature field data of the heat exchanger pipes.

[0010] On the other hand, the actual operating data also includes the density of the liquid inside the heat exchanger and the number of baffles inside the heat exchanger;

[0011] A simulation model is constructed based on the actual operating data of the heat exchanger, including:

[0012] A simulation model is constructed based on the dimensions of the heat exchanger's pipes, the density of the liquid inside the heat exchanger, and the number of baffles inside the heat exchanger.

[0013] On the other hand, the velocity field data and the temperature field data are U = [u1, u2, ..., u n There are a total of n velocity field data and temperature field data, u i For the i-th velocity field data and temperature field data;

[0014] in x m Let m be the location of the m-th monitoring point in the i-th velocity field data or temperature field data. The velocity or temperature is the velocity or temperature at the m-th monitoring point in the i-th velocity field data or temperature field data.

[0015] On the other hand, the flow velocity field data and the temperature field data, which are randomly collected in a proportion of a first preset ratio, include:

[0016] The velocity field data and temperature field data are collected according to a first preset ratio based on the Latin hypercube sampling method, and the velocity field data and temperature field data with a data volume ratio of the first preset ratio are used as the first weight w. i ;

[0017] Based on the velocity field data and the temperature field data, after randomly collecting velocity field data and temperature field data at a first preset ratio, the method further includes:

[0018] Determine the rate of change of the velocity field data and the temperature field data. The rate of change is the difference between the i-th data and the (i-1)-th data. The total number of the velocity field data and the temperature field data is n, where 1 < i ≤ n, and i and n are both integers.

[0019] Based on the data change rate, velocity field data and temperature field data are collected in a proportion that is a second preset ratio, from high to low. The velocity field data and temperature field data with the second preset ratio are used as the second weight w. p ;

[0020] The third weight w is obtained by adding the flow velocity field data and the temperature field data with a data volume ratio of the first preset ratio and adding the weights of the flow velocity field data and the temperature field data with a data volume ratio of the second preset ratio. m =w i +w p ;

[0021] The flow field data and temperature field data are collected in a third preset ratio according to the third weight, from high to low.

[0022] On the other hand, after collecting flow field data and temperature field data in a third preset ratio according to the third weight from high to low, it also includes:

[0023] Novel Equations for Unidirectional Fluid-Structure Coupled Systems This refers to the flow velocity field data and the temperature field data;

[0024] in, Represents all N i ×N i The set of real matrices, A 11 For the velocity field data, A 22 For the temperature field data, Characterizes the coupling relationship between the velocity field data and the temperature field data. It is the solution of two subsystems. It is the input vector and x i Related matrices i = 1, 2, It is the set of natural numbers;

[0025] Determine the A 11 and A 22 low-order approximation and The lower-order approximation satisfies The lower-order approximation is data with lower accuracy than the flow velocity field data and the temperature field data, and with a smaller data volume than the flow velocity field data and the temperature field data. This is an approximate solution to the lower-order approximation, and the lower-order approximation relation is: and

[0026] According to the POD-RBF order reduction algorithm, A 11 low-order approximation Represented as A 22 low-order approximation Represented as

[0027] Among them, P 1 Representative composition A 11 The solution depends on the input parameter vector, P. 1 It consists of p elements, p 1 =(p1,...,p i ,...p p ), p i P represents the internal scale thickness of the pipes in the heat exchanger corresponding to the simulation model, the inlet temperature of the pipes in the heat exchanger corresponding to the simulation model, and the inlet flow velocity. 2 Representative composition A 22 Solving for the input parameter vector P that depends on 2 It consists of p+M elements. The first element is P 1 , Characterizing the velocity field data, where j = 1, 2, ..., M, B 11 Let f be the coefficient matrix of the velocity field data. 11 B is the radial basis function for the velocity field data. 22 f is the coefficient matrix of the temperature field data. 22 is the radial basis function for the temperature field data.

[0028] On the other hand, according to the POD-RBF order reduction algorithm, A 11 low-order approximation Represented as A 22 low-order approximation Represented as Following that, it also includes:

[0029] Setting an autoencoder will input vector Mapped to the intermediate vector F = σ(Wp) 2 +b);

[0030] in, P 1 +1+M>n, It is a weight matrix. σ is the bias vector, and σ is the activation function of the coding layer.

[0031] The intermediate vector F = σ(Wp) 2 +b) is mapped to the output vector

[0032] in, P 1 +1+M>n, It is a weight matrix. It is a bias vector. It is the activation function of the decoding layer;

[0033] Determine the reconstructed vector and the original input vector P 2 Reconstruction error between the two E represents the reconstruction error;

[0034] When the reconstruction error is minimized, F is used instead. P in 2 ;

[0035] Based on the replaced lower-order approximation and The scale thickness of the heat exchanger is determined by the correspondence between the scale thickness and the temperature field data.

[0036] On the other hand, the correspondence between the scale thickness and the temperature field data is determined based on the collected flow field data and temperature field data at the first preset ratio, including:

[0037] The temperature field data and the internal scale thickness of the pipes of the heat exchanger corresponding to the simulation model of the temperature field data are input into the BP neural network to obtain the correspondence between the scale thickness output by the BP neural network and the temperature field data.

[0038] The scale thickness of the heat exchanger pipes is determined based on the correspondence between the inlet and outlet temperatures of the heat exchanger pipes and the scale thickness and temperature field data, including:

[0039] The inlet and outlet temperatures of the heat exchanger pipes are input into the BP neural network to obtain the scale thickness of the heat exchanger pipes.

[0040] To address the aforementioned technical problems, the present invention also provides a device digital twin modeling and intelligent sensing system, comprising:

[0041] The model building unit is used to build a simulation model based on the actual operating data of the heat exchanger, including the dimensions of the heat exchanger's pipes.

[0042] The data determination unit is used to obtain the flow velocity field data and temperature field data output by the simulation model based on the inlet temperature and inlet flow velocity of the liquid input to the pipe of the heat exchanger corresponding to the simulation model and the scale thickness inside the pipe of the heat exchanger corresponding to the simulation model. The flow velocity field data includes the liquid flow velocity at each monitoring point in the simulation model, and the temperature field data includes the temperature at each monitoring point in the simulation model. The monitoring point represents a preset position inside the pipe of the heat exchanger corresponding to the simulation model.

[0043] The acquisition unit is used to randomly acquire the flow velocity field data and the temperature field data at a first preset ratio.

[0044] The correspondence determination unit is used to determine the correspondence between the scale thickness and the temperature field data based on the collected flow field data and temperature field data, which are in a first preset ratio of the data volume.

[0045] The scale thickness determination unit is used to determine the scale thickness of the heat exchanger pipes based on the inlet temperature, outlet temperature of the heat exchanger pipes and the correspondence between the scale thickness and temperature field data.

[0046] To address the aforementioned technical problems, the present invention also provides a device for digital twin modeling and intelligent sensing of equipment, comprising:

[0047] Memory, used to store computer programs;

[0048] A processor is used to implement the steps of the above-described digital twin modeling and intelligent sensing method for the device when executing the computer program.

[0049] To address the aforementioned technical problems, the present invention also provides a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the steps of the aforementioned device digital twin modeling and intelligent sensing method.

[0050] This invention discloses a digital twin modeling and intelligent sensing method and related components for equipment. A simulation model is constructed based on the actual operating data of a heat exchanger. Based on the inlet temperature and velocity of the liquid in the pipes of the heat exchanger corresponding to the simulation model, and the internal scale thickness of the pipes, flow velocity field data and temperature field data are obtained from the simulation model. The flow velocity field data includes the liquid flow velocity at each monitoring point in the simulation model, and the temperature field data includes the temperature at each monitoring point. The correspondence between the scale thickness and temperature field data is determined based on a first preset proportion of the collected flow velocity field data and temperature field data. The scale thickness of the heat exchanger is determined based on the correspondence between the actual liquid temperature in the heat exchanger pipes and the scale thickness and temperature field data. By constructing a simulation model based on the actual heat exchanger and continuously adjusting the liquid flow velocity and scale data, the obtained scale data becomes more accurate. Furthermore, randomly collecting flow velocity field data and temperature field data at a first preset proportion reduces the data computation time. Attached Figure Description

[0051] To more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings used in the prior art and embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0052] Figure 1 A flowchart illustrating a device digital twin modeling and intelligent sensing method provided by the present invention;

[0053] Figure 2 This is a schematic diagram of the structure of a device digital twin modeling and intelligent sensing system provided by the present invention;

[0054] Figure 3 This is a schematic diagram of the structure of a digital twin modeling and intelligent sensing device for equipment provided by the present invention. Detailed Implementation

[0055] The core of this invention is to provide a digital twin modeling and intelligent sensing method and related components for equipment. A simulation model is constructed based on the actual heat exchanger, and the liquid flow rate and scaling data are continuously adjusted to obtain more accurate scaling data. In addition, the flow rate field data and temperature field data collected in a first preset proportion can reduce the data calculation time.

[0056] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0057] Figure 1 A flowchart of a device digital twin modeling and intelligent sensing method provided by the present invention includes:

[0058] S11: Construct a simulation model based on the actual operating data of the heat exchanger, including the dimensions of the heat exchanger's pipes;

[0059] To make the simulation model more closely resemble the actual operation of the heat exchanger, the actual operating data of the heat exchanger must be referenced during the simulation model construction process. The actual operating data includes the dimensions of the heat exchanger's pipes, such as length, inner diameter, and outer diameter.

[0060] Only when the simulation model closely approximates the actual operating data of the heat exchanger will the data obtained from the simulation model be more accurate.

[0061] S12: Based on the inlet temperature and inlet velocity of the liquid in the pipes of the heat exchanger corresponding to the simulation model, and the scale thickness inside the pipes of the heat exchanger corresponding to the simulation model, obtain the velocity field data and temperature field data output by the simulation model. The velocity field data includes the liquid velocity at each monitoring point in the simulation model, and the temperature field data includes the temperature at each monitoring point in the simulation model. The monitoring point represents the preset position inside the pipes of the heat exchanger corresponding to the simulation model.

[0062] Considering that the flow rate of the liquid input to the heat exchanger and the scale thickness in the simulation model affect the liquid flow rate and temperature at various monitoring points of the heat exchanger, the liquid flow rate and scale thickness are adjusted to obtain the flow velocity field data and temperature field data output by the simulation model. The liquid flow rate at each monitoring point constitutes the flow velocity field data, and the temperature at each monitoring point constitutes the temperature field data.

[0063] S13: Randomly collect velocity field data and temperature field data at a first preset ratio;

[0064] S14: Determine the correspondence between the scale thickness and the temperature field data based on the collected flow field data and temperature field data of the first preset ratio;

[0065] Considering the large number of monitoring points and the continuous adjustment of the liquid flow rate and the fouling thickness of the heat exchanger corresponding to the simulation model during the test, a large amount of temperature field data and flow velocity field data were obtained. The subsequent calculation process involved a large amount of data and took a long time.

[0066] Therefore, this application randomly collects data in the velocity field data and temperature field data at a first preset ratio, and then uses the collected data for subsequent calculations.

[0067] Furthermore, although the collected velocity field data and temperature field data at the first preset ratio will reduce the accuracy of the calculation, it will greatly reduce the calculation time, and the error in the calculation accuracy is within the preset range.

[0068] S15: Determine the scale thickness of the heat exchanger pipes based on the correspondence between the inlet temperature, outlet temperature, scale thickness, and temperature field data of the heat exchanger pipes.

[0069] Based on the correlation between temperature field data, scale thickness, and temperature field data, the scale thickness of the heat exchanger can be determined according to the temperature at the monitoring point, so as to facilitate subsequent scale treatment.

[0070] This invention discloses a digital twin modeling and intelligent sensing method and related components for equipment. Based on the inlet temperature and flow velocity of the liquid in the pipes of the heat exchanger corresponding to the simulation model, and the internal scale thickness of the pipes in the simulation model, flow velocity field data and temperature field data output by the simulation model are obtained. The flow velocity field data includes the liquid flow velocity at each monitoring point in the simulation model, and the temperature field data includes the temperature at each monitoring point in the simulation model. The correspondence between the scale thickness and temperature field data is determined based on the collected flow velocity field data and temperature field data at a first preset ratio. The scale thickness of the heat exchanger is determined based on the actual liquid temperature in the heat exchanger pipes and the correspondence between the scale thickness and temperature field data. By constructing a simulation model based on the actual heat exchanger and continuously adjusting the liquid flow velocity and scale data, the obtained scale data becomes more accurate. Furthermore, randomly collecting flow velocity field data and temperature field data at a first preset ratio can reduce the data calculation time.

[0071] Based on the above embodiments:

[0072] In some embodiments, the actual operating data also includes the density of the liquid inside the heat exchanger and the number of baffles inside the heat exchanger;

[0073] A simulation model is constructed based on the actual operating data of the heat exchanger, including:

[0074] A simulation model is constructed based on the dimensions of the heat exchanger's pipes, the density of the liquid inside the heat exchanger, and the number of baffles inside the heat exchanger.

[0075] Considering that heat exchangers achieve liquid flow velocity and direction through the guiding effect of internal baffles, and that these baffles affect the heat transfer efficiency, the number of baffles must also be taken into account in the actual operating data of the heat exchanger. The density of the transferred liquid is related to the scale thickness; if the liquid density is high, it is more likely to adhere to the tube walls of the heat exchanger, leading to scale buildup over time. Therefore, when constructing the simulation model, it is necessary to consider the dimensions of the heat exchanger's pipes, the density of the liquid inside the heat exchanger, and the number of internal baffles to obtain a simulation model that more closely approximates the actual heat exchanger.

[0076] In some embodiments, the velocity field data and temperature field data are U = [u1, u2, ..., u n There are n velocity field data and temperature field data in total, u i For the i-th velocity field data and temperature field data;

[0077] in x m Let m be the location of the m-th monitoring point in the i-th velocity field data or temperature field data. The velocity or temperature at the m-th monitoring point in the i-th velocity field data or temperature field data.

[0078] The obtained temperature and velocity field data are presented as a data matrix. The specific temperature and velocity field data include the coordinates of the monitoring points and the flow velocity or temperature at those points. Combining these data points with the liquid flow velocity and temperature will lead to more accurate calculations in subsequent calculations.

[0079] In some embodiments, the flow velocity field data and temperature field data, which are randomly collected in proportion to a first preset ratio, include:

[0080] Based on the flow velocity and temperature field data collected by Latin hypercube sampling with a data volume ratio of a first preset ratio, and with the flow velocity and temperature field data having a data volume ratio of a first preset ratio as the first weight w i ;

[0081] Based on the velocity field data and temperature field data, after randomly collecting velocity field data and temperature field data at a first preset proportion, the following is also included:

[0082] Determine the rate of change of the velocity field data and the temperature field data. The rate of change is the difference between the i-th data and the (i-1)-th data. The total number of velocity field data and temperature field data is n, 1 < i ≤ n, where i and n are both integers.

[0083] Based on the data change rate, velocity field data and temperature field data are collected in proportion to a second preset ratio, from high to low. The velocity field data and temperature field data with the second preset ratio are used as the second weight w. p ;

[0084] The third weight w is obtained by adding the velocity field data and temperature field data with a data volume ratio of the first preset ratio and the weights of the velocity field data and temperature field data with a data volume ratio of the second preset ratio. m =w i +w p ;

[0085] The flow field data and temperature field data are collected in a third preset ratio according to the third weight, from high to low.

[0086] To cover the entire parameter domain and obtain coarse information about the entire sampling space, a first weight w is set. i Initial sampling was performed using Latin hypercube sampling (LHS). Features were extracted based on the sample data in the target parameter domain. A second weight w was set. p Furthermore, an adaptive sampling strategy based on the rate of data change was developed to map the data changes of the original samples to an adaptive sampling space, thereby constructing a sample feature extraction model. A dynamic weight term w was constructed based on the region of interest in the parameter domain. m =w i +w p The weight term of each sampling point is obtained by weighting and summing the physical information weight of each sampling point with the initial sampling weight. The optimal order reduction performance is obtained by obtaining the sampling point with the larger weight term.

[0087] To improve the accuracy of the order reduction model and ensure the consistency of the error distribution without increasing computational cost, a data change rate is introduced to quickly extract feature information from the original sample set as input to the adaptive sampling model. The data change rate is the difference between the i-th data point and the (i-1)-th data point. A dynamic weight term is constructed based on the region of interest in the parameter domain. The weight term for each sampling point is obtained by weighted summing the physical information weights of each sampling point with the initial sampling weights. Optimal order reduction performance is achieved by selecting sampling points with larger weight terms.

[0088] Specifically, in the first sampling process, taking a total of 100 data points, a first preset ratio of 60%, and a second preset ratio of 50% as an example, the weight of the 60 data points after the first sampling is the first weight w. i The second sampling sorts the data by rate of change from high to low, resulting in 50 data points with weights designated as the second weight w. pThe aforementioned 50 and 60 data points may or may not overlap. Therefore, combining the data from both samplings will result in a maximum of 110 data points and a minimum of 60 data points. The weights of the data from the two samplings will be added together, while the weights of the data from the first sampling will remain unchanged. The data will then be sorted from highest to lowest weight, and the proportion of collected data will be the third preset ratio.

[0089] In some embodiments, after collecting flow field data and temperature field data in a third preset ratio according to a third weight from high to low, the method further includes:

[0090] Novel Equations for One-Way Fluid-Structure Coupled Systems This represents the flow velocity field data and temperature field data;

[0091] in, Represents all N i ×N i The set of real matrices, A 11 For velocity field data, A 22 For temperature field data, Characterize the coupling relationship between velocity field data and temperature field data. It is the solution of two subsystems. It is the input vector and x i Related matrices i = 1, 2, It is the set of natural numbers;

[0092] Determine A 11 and A 22 low-order approximation and Low-order approximations satisfy Low-order approximations are data with lower accuracy and smaller data volume than velocity field data and temperature field data. This is an approximate solution for the lower-order approximation, and the lower-order approximation relation is: and

[0093] According to the POD-RBF (Proper Orthogonal Decomposition-Radial Basis Function, a radial basis model based on eigenorthogonal decomposition) order reduction algorithm, A 11 low-order approximation Represented as A 22 low-order approximation Represented as

[0094] Among them, P 1 Representative composition A11 Solving for the input parameter vector P that depends on 1 It consists of p elements, p 1 =(p1,...,p i ,...p p ), p i The internal fouling thickness of the pipes in the heat exchanger corresponding to the simulation model, the inlet temperature of the pipes in the heat exchanger corresponding to the simulation model, and the inlet flow velocity are all represented by P. 2 Representative composition A 22 Solving for the input parameter vector P that depends on 2 It consists of p+M elements. The first element is P 1 , Characterizing the velocity field data, where j = 1, 2, ..., M, B 11 Let f be the coefficient matrix of the velocity field data. 11 B is the radial basis function for the velocity field data. 22 f is the coefficient matrix of the temperature field data. 22 is the radial basis function for the temperature field data.

[0095] In some embodiments, A is reduced to its original order according to the POD-RBF reduction algorithm. 11 low-order approximation Represented as A 22 low-order approximation Represented as Following that, it also includes:

[0096] Setting an autoencoder will input vector Mapped to the intermediate vector F = σ(Wp) 2 +b);

[0097] in, P 1 +1+M>n, It is a weight matrix. σ is the bias vector, and σ is the activation function of the coding layer.

[0098] The intermediate vector F = σ(Wp) 2 +b) is mapped to the output vector

[0099] in, P 1 +1+M>n, It is a weight matrix. It is a bias vector. It is the activation function of the decoding layer;

[0100] Determine the reconstructed vector and the original input vector P2 Reconstruction error between the two E represents the reconstruction error;

[0101] When the reconstruction error is minimized, F is used instead. P in 2 ;

[0102] Based on the lower-order approximation after substitution and The scale thickness of the heat exchanger is determined by the correspondence between the scale thickness and temperature field data.

[0103] For A 11 Construct a set U1 of N1 sampled values ​​u of its physical field. By continuously changing the influence of A... 11 The input parameters (flow rate and scale thickness) are used to obtain M u1 values. j The eigenvalues ​​and eigenvectors are obtained directly by introducing the singular value decomposition (SVD) method, avoiding unnecessary computational overhead.

[0104] Suppose A is an m×n matrix with rank k, and there exists an orthogonal basis: V = (v1, ..., v1) / (v2, ..., v3) / (v4, ..., v5) / (v6, ..., v7) / (v8, ..., v9) / (v1, ..., v1) / (v1, ..., v1) / (v2 ... k v1, ..., v k In a k-dimensional space, k linearly independent vectors form a basis for that space. The matrix is ​​denoted as U = (Av1, ..., Av) so that it remains an orthogonal basis after transformation. k );

[0105] By the definition of an orthogonal basis, any (Av) i ) T (Av j =0, expanding this expression gives v i T A T Av j =0. When v i It is A T When the eigenvectors of A are given, we have (A T A)v j =λv i That is, λv i T v j =0. From this, we can deduce the eigenvalues ​​and eigenvectors of matrix U: Its matrix identifier is AV=UΣ.

[0106] Wherein, orthogonal matrix and eigenvalue diagonal matrix 0 ≤ p = rank(A) ≤ min(n, m), and σ1 ≥ ... ≥ σ p ≥0, the transformation yields A=UΣV T Substituting into U1, we can obtain the positive definite covariance matrix C, eigenvector V, and POD basis Φ of U1. 11 =U 11 V.

[0107] Let P 1 Representative composition A 11 Solving for the input parameter vector P that depends on 1 It consists of P elements, p 1 =(p1,...,p i ,...p p ), p i These parameters are used to describe the thickness of scale, the inlet and outlet temperatures of cold and hot flows, and so on.

[0108] Using the POD-RBF order reduction algorithm, A can be reduced to a lower order. 11 low-dimensional model Represented as: Among them B 11 Let f be the coefficient matrix of the velocity field data. 11 The radial basis function is used for the velocity field data, and the interpolation function used is the Inverse multiquadric function.

[0109] Let P 2 Representative composition A 22 Solving for the input parameter vector on which the solution depends, at this point, P 2 It consists of p+M elements, that is The first element represents the composition of A. 11 The solution depends on the input parameter vector, and the remaining M elements are used to describe A. 11 The input parameters are used to obtain N1 items. Where j = 1, 2, ..., M, then similarly, A can be... 22 low-dimensional model Represented as Among them B 22 f is the coefficient matrix of the temperature field data. 22 The radial basis function for the temperature field data is used, and the interpolation function is the Inverse multiquadric function.

[0110] A method for dimensionality reduction of heat exchanger subsystem parameters based on autoencoders for A 22 Parameter P 2Dimensionality reduction is achieved by using backpropagation and optimization methods (such as gradient descent) to guide the neural network to learn a mapping relationship, thereby obtaining a dimensionality-reconstructed output. This maximizes the accuracy of the solution while meeting computational requirements.

[0111] During the encoding phase, the autoencoder uses the following formula: The function shown will take the input vector Mapped to intermediate vector And P1+1+M>n, F=σ(Wp) 2 +b);

[0112] in, It is a weight matrix. σ is the bias vector, and σ is the activation function of the coding layer, such as ReLU or sigmoid.

[0113] During the decoding phase, the autoencoder passes the intermediate vector... Mapped to output vector And P 1 +1+M>n, therefore...

[0114] in It is a weight matrix. It is a bias vector. If it is the activation function of the decoding layer, then the reconstructed vector and the original input vector P 2 The reconstruction error between the two is as follows

[0115] The goal of an autoencoder is to minimize the reconstruction error. This is equivalent to learning the appropriate weight matrix W through gradient descent and backpropagation. and the bias vector b and Make Minimize. The intermediate vector F is the one that can approximate the input-output mapping relationship. By using F to replace... P in 2 This solves the dimensionality explosion problem. At this point, a coupled reduced-order model can be constructed.

[0116] In some embodiments, determining the correspondence between scale thickness and temperature field data based on collected flow field data and temperature field data at a first preset ratio includes:

[0117] The temperature field data and the internal scale thickness of the heat exchanger pipes corresponding to the temperature field data simulation model are input into the BP neural network to obtain the correspondence between the scale thickness output by the BP neural network and the temperature field data.

[0118] The scale thickness of the heat exchanger pipes is determined based on the correspondence between the inlet and outlet temperatures, scale thickness, and temperature field data. This includes:

[0119] The inlet and outlet temperatures of the heat exchanger pipes are input into a BP neural network to obtain the scale thickness of the heat exchanger pipes.

[0120] Figure 2 A schematic diagram of a device digital twin modeling and intelligent sensing system provided by the present invention includes:

[0121] Model building unit 21 is used to build a simulation model based on the actual operating data of the heat exchanger, including the dimensions of the heat exchanger's pipes.

[0122] The data determination unit 22 is used to obtain the flow velocity field data and temperature field data output by the simulation model based on the inlet temperature and inlet flow velocity of the liquid in the pipe of the heat exchanger corresponding to the simulation model and the scale thickness inside the pipe of the heat exchanger corresponding to the simulation model. The flow velocity field data includes the liquid flow velocity at each monitoring point in the simulation model, and the temperature field data includes the temperature at each monitoring point in the simulation model. The monitoring point represents a preset position inside the pipe of the heat exchanger corresponding to the simulation model.

[0123] Acquisition unit 23 is used to randomly acquire flow field data and temperature field data at a first preset ratio.

[0124] The correspondence determination unit 24 is used to determine the correspondence between the scale thickness and the temperature field data based on the collected flow field data and temperature field data of the first preset ratio.

[0125] The scale thickness determination unit 25 is used to determine the scale thickness of the heat exchanger pipes based on the correspondence between the inlet temperature, outlet temperature, scale thickness, and temperature field data of the heat exchanger pipes.

[0126] Based on the above embodiments:

[0127] Actual operating data also includes the density of the liquid inside the heat exchanger and the number of baffles inside the heat exchanger;

[0128] Model building unit 21 is specifically used to build a simulation model based on the size of the heat exchanger's pipes, the density of the liquid inside the heat exchanger, and the number of baffles inside the heat exchanger.

[0129] The velocity field data and temperature field data are U=[u1,u2,...,u n There are n velocity field data and temperature field data in total, u i For the i-th velocity field data and temperature field data;

[0130] in x m Let m be the location of the m-th monitoring point in the i-th velocity field data or temperature field data. The velocity or temperature at the m-th monitoring point in the i-th velocity field data or temperature field data.

[0131] Acquisition unit 23 is specifically used to acquire flow field data and temperature field data with a data volume ratio of a first preset ratio according to Latin hypercube sampling, wherein the flow field data and temperature field data with a data volume ratio of the first preset ratio are used as the first weight w. i ;

[0132] Also includes:

[0133] The rate of change determination unit is used to determine the rate of change of the velocity field data and the temperature field data. The rate of change is the difference between the i-th data and the (i-1)-th data. The total number of velocity field data and temperature field data is n, 1 < i ≤ n, where i and n are both integers.

[0134] The second acquisition unit is used to acquire flow field data and temperature field data in a second preset ratio, arranged from high to low according to the data change rate. The flow field data and temperature field data in the second preset ratio are used as the second weight w. p ;

[0135] The weighting unit is used to add the velocity field data and temperature field data with a data volume ratio of a first preset ratio, and to add the weights of the velocity field data and temperature field data with a data volume ratio of a second preset ratio to obtain a third weight w. m =w i +w p ;

[0136] The third acquisition unit is used to acquire flow field data and temperature field data in a third preset ratio according to the third weight, from high to low.

[0137] Equation representation unit for novel equations used in unidirectional fluid-structure interaction systems This represents the flow velocity field data and temperature field data;

[0138] in, Represents all N i ×N i The set of real matrices, A 11 For velocity field data, A 22 For temperature field data, Characterize the coupling relationship between velocity field data and temperature field data. It is the solution of two subsystems. It is the input vector and x i Related matrices i = 1, 2, It is the set of natural numbers;

[0139] Determine the approximate value unit, used to determine A 11 and A 22 low-order approximation and Low-order approximations satisfy Low-order approximations are data with lower accuracy and smaller data volume than velocity field data and temperature field data. This is an approximate solution for the lower-order approximation, and the lower-order approximation relation is: and

[0140] The low-order approximation determination unit, according to the POD-RBF order reduction algorithm, determines A. 11 low-order approximation Represented as A 22 low-order approximation Represented as

[0141] Among them, P 1 Representative composition A 11 Solving for the input parameter vector P that depends on 1 It consists of p elements, p 1 =(p1,...,p i ,...p p ), p i The internal fouling thickness of the pipes in the heat exchanger corresponding to the simulation model, the inlet temperature of the pipes in the heat exchanger corresponding to the simulation model, and the inlet flow velocity are all represented by P. 2 Representative composition A 22 Solving for the input parameter vector P that depends on 2 It consists of p+M elements. The first element is P 1 , Characterize the velocity field data, where j = 1, 2, ..., M, B 11 Let f be the coefficient matrix of the velocity field data. 11 B is the radial basis function for the velocity field data. 22 f is the coefficient matrix of the temperature field data. 22 is the radial basis function for the temperature field data.

[0142] Also includes:

[0143] The encoding unit is used to set the autoencoder to process the input vector. Mapped to the intermediate vector F = σ(Wp) 2 +b);

[0144] in, P 1 +1+M>n, It is a weight matrix. σ is the bias vector, and σ is the activation function of the coding layer.

[0145] The decoding unit is used to decode the intermediate vector F = σ(Wp) 2 +b) is mapped to the output vector

[0146] in, P 1 +1+M>n, It is a weight matrix. It is a bias vector. It is the activation function of the decoding layer;

[0147] The reconstruction error determination unit is used to determine the reconstruction vector. and the original input vector P 2 Reconstruction error between the two E represents the reconstruction error;

[0148] The substitution cell is used to replace F when the reconstruction error is minimized. P in 2 ;

[0149] The scale thickness determination unit 25 is specifically used to determine the scale thickness based on the replaced low-order approximation. and The scale thickness of the heat exchanger is determined by the correspondence between the scale thickness and temperature field data.

[0150] The correspondence determination unit 24 is used to input the temperature field data and the internal scale thickness of the pipe of the heat exchanger corresponding to the simulation model of the temperature field data into the BP neural network to obtain the correspondence between the scale thickness output by the BP neural network and the temperature field data.

[0151] The scale thickness determination unit 25 is used to input the inlet temperature and outlet temperature of the heat exchanger pipes into the BP neural network to obtain the scale thickness of the heat exchanger pipes.

[0152] Figure 3 A schematic diagram of the structure of a digital twin modeling and intelligent sensing device for equipment provided by the present invention includes:

[0153] Memory 31 is used to store computer programs;

[0154] The processor 32 is used to implement the steps of the above-mentioned digital twin modeling and intelligent sensing method for the device when executing a computer program.

[0155] Please refer to the above embodiments for a description of the digital twin modeling and intelligent sensing device provided in this application, which will not be repeated here.

[0156] The present invention also provides a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the steps of the above-described digital twin modeling and intelligent sensing method for the device.

[0157] The description of the computer-readable storage medium provided in this application is given in the above embodiments and will not be repeated here.

[0158] It should also be noted that, in this specification, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0159] Those skilled in the art will further recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of both. To clearly illustrate the interchangeability of hardware and software, the components and steps of the various examples have been generally described in terms of functionality in the foregoing description. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this invention.

[0160] The above description of the disclosed embodiments enables those skilled in the art to make or use the invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the invention. Therefore, the invention is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for digital twin modeling and intelligent sensing of equipment, characterized in that, include: A simulation model is constructed based on the actual operating data of the heat exchanger, including the dimensions of the heat exchanger's pipes. Based on the inlet temperature and inlet velocity of the liquid in the pipes of the heat exchanger corresponding to the simulation model, and the scale thickness inside the pipes of the heat exchanger corresponding to the simulation model, the flow velocity field data and temperature field data output by the simulation model are obtained. The flow velocity field data includes the liquid flow velocity at each monitoring point in the simulation model, and the temperature field data includes the temperature at each monitoring point in the simulation model. The monitoring point represents a preset position inside the pipes of the heat exchanger corresponding to the simulation model. The flow velocity field data and the temperature field data are randomly collected in proportion to a first preset ratio; The correspondence between the scale thickness and the temperature field data is determined based on the collected flow field data and temperature field data of the first preset ratio. The scale thickness of the heat exchanger pipes is determined based on the correspondence between the inlet temperature, outlet temperature, scale thickness, and temperature field data of the heat exchanger pipes. The flow velocity field data and the temperature field data, with a random data collection ratio of a first preset ratio, include: The velocity field data and temperature field data are collected according to a first preset ratio based on the Latin hypercube sampling data, and the velocity field data and temperature field data of the first preset ratio are used as the first weight. ; Based on the velocity field data and the temperature field data, after randomly collecting velocity field data and temperature field data at a first preset ratio, the method further includes: Determine the rate of change of the velocity field data and the temperature field data. The rate of change is the difference between the i-th data and the (i-1)-th data. The total number of the velocity field data and the temperature field data is n, where 1 < i ≤ n, and i and n are both integers. Based on the data change rate, velocity field data and temperature field data are collected in proportion to a second preset ratio, from high to low. The velocity field data and temperature field data with the second preset ratio are used as the second weight. ; The third weight is obtained by adding the flow velocity field data and the temperature field data with a data volume ratio of the first preset ratio and by adding the weights of the flow velocity field data and the temperature field data with a data volume ratio of the second preset ratio. ; The velocity field data and temperature field data are collected in a third preset ratio from high to low according to the third weight; After collecting flow field data and temperature field data in a third preset ratio according to the third weight from high to low, the data also includes: Novel Equations for Unidirectional Fluid-Structure Coupled Systems This refers to the flow velocity field data and the temperature field data; in, , Indicates all The set of real matrices, For the aforementioned velocity field data, For the temperature field data, Characterizes the coupling relationship between the velocity field data and the temperature field data. It is the solution to the novel equation. It is the input vector and Related matrices , , It is the set of natural numbers; Determine the and low-order approximation and The lower-order approximation satisfies The lower-order approximation is data with lower accuracy than the flow velocity field data and the temperature field data, and with a smaller data volume than the flow velocity field data and the temperature field data. This is an approximate solution to the lower-order approximation, and the lower-order approximation relation is: and ; According to the POD-RBF order reduction algorithm, low-order approximation Represented as ,Will low-order approximation Represented as ; Among them, the It consists of p elements. , The internal scale thickness of the pipes in the heat exchanger corresponding to the simulation model, the inlet temperature of the pipes in the heat exchanger corresponding to the simulation model, and the inlet flow velocity are characterized. Depend on Composed of elements, The first element is , Characterizing the velocity field data, wherein , This is the coefficient matrix of the velocity field data. For the radial basis functions of the velocity field data. This is the coefficient matrix of the temperature field data. Let M be the radial basis function of the temperature field data, where M is a positive integer. and It is POD-based.

2. The device digital twin modeling and intelligent sensing method as described in claim 1, characterized in that, The actual operating data also includes the density of the liquid inside the heat exchanger and the number of baffles inside the heat exchanger. A simulation model is constructed based on the actual operating data of the heat exchanger, including: A simulation model is constructed based on the dimensions of the heat exchanger's pipes, the density of the liquid inside the heat exchanger, and the number of baffles inside the heat exchanger.

3. The device digital twin modeling and intelligent sensing method as described in claim 1, characterized in that, The velocity field data and the temperature field data are There are a total of n velocity field data and temperature field data. For the i-th velocity field data or temperature field data; in , This refers to the position of the m-th monitoring point in the i-th velocity field data or temperature field data. The velocity or temperature at the m-th monitoring point in the i-th velocity field data or temperature field data.

4. The device digital twin modeling and intelligent sensing method as described in claim 1, characterized in that, According to the POD-RBF order reduction algorithm, low-order approximation Represented as ,Will low-order approximation Represented as Following that, it also includes: Setting an autoencoder will input vector Mapped to intermediate vector ; in, , , It is a weight matrix. It is a bias vector. It is the activation function of the coding layer; intermediate vector Mapped to output vector ; in, , , It is a weight matrix. It is a bias vector. It is the activation function of the decoding layer; Determine the reconstructed vector and the original input vector Reconstruction error between the two E is the reconstruction error; When the reconstruction error is minimized, F is used instead. In ; Based on the replaced lower-order approximation and The scale thickness of the heat exchanger is determined by the correspondence between the scale thickness and the temperature field data.

5. The device digital twin modeling and intelligent sensing method as described in any one of claims 1 to 4, characterized in that, The correspondence between scale thickness and temperature field data is determined based on the collected flow field data and temperature field data at the first preset ratio, including: The temperature field data and the internal scale thickness of the pipes of the heat exchanger corresponding to the simulation model of the temperature field data are input into the BP neural network to obtain the correspondence between the scale thickness output by the BP neural network and the temperature field data. The scale thickness of the heat exchanger pipes is determined based on the correspondence between the inlet and outlet temperatures of the heat exchanger pipes and the scale thickness and temperature field data, including: The inlet and outlet temperatures of the heat exchanger pipes are input into the BP neural network to obtain the scale thickness of the heat exchanger pipes.

6. A digital twin modeling and intelligent sensing system for equipment, characterized in that, include: The model building unit is used to build a simulation model based on the actual operating data of the heat exchanger, including the dimensions of the heat exchanger's pipes. The data determination unit is used to obtain the flow velocity field data and temperature field data output by the simulation model based on the inlet temperature and inlet flow velocity of the liquid input to the pipe of the heat exchanger corresponding to the simulation model and the scale thickness inside the pipe of the heat exchanger corresponding to the simulation model. The flow velocity field data includes the liquid flow velocity at each monitoring point in the simulation model, and the temperature field data includes the temperature at each monitoring point in the simulation model. The monitoring point represents a preset position inside the pipe of the heat exchanger corresponding to the simulation model. The acquisition unit is used to randomly acquire the velocity field data and the temperature field data at a first preset ratio. The correspondence determination unit is used to determine the correspondence between the scale thickness and the temperature field data based on the collected flow field data and temperature field data, which are in a first preset ratio of the data volume. The scale thickness determination unit is used to determine the scale thickness of the heat exchanger pipes based on the inlet temperature, outlet temperature of the heat exchanger pipes and the correspondence between the scale thickness and temperature field data. The acquisition unit is specifically used to acquire the flow velocity field data and the temperature field data according to a first preset ratio of the data volume obtained by Latin hypercube sampling, wherein the flow velocity field data and the temperature field data of the first preset ratio are a first weight. ; Also includes: The rate of change determination unit is used to determine the rate of change of the flow velocity field data and the temperature field data. The rate of change is the difference between the i-th data and the (i-1)-th data. The total number of the flow velocity field data and the temperature field data is n, 1 < i ≤ n, and i and n are both integers. The second acquisition unit is used to acquire flow velocity field data and temperature field data in a second preset ratio, arranged from high to low according to the data change rate. The flow velocity field data and temperature field data in the second preset ratio are used as the second weight. ; The weighting determination unit is used to add the velocity field data and the temperature field data with a data volume ratio of a first preset ratio, and to add the weights of the velocity field data and the temperature field data with a data volume ratio of a second preset ratio to obtain a third weight. ; The third acquisition unit is used to acquire flow field data and temperature field data in a third preset ratio according to the third weight from high to low. Equation representation unit for novel equations used in unidirectional fluid-structure interaction systems This refers to the flow velocity field data and the temperature field data; in, , Indicates all The set of real matrices, For the aforementioned velocity field data, For the temperature field data, Characterizes the coupling relationship between the velocity field data and the temperature field data. It is the solution to the novel equation. It is the input vector and Related matrices , , It is the set of natural numbers; Determine the approximate value unit, used to determine the and low-order approximation and The lower-order approximation satisfies The lower-order approximation is data with lower accuracy than the flow velocity field data and the temperature field data, and with a smaller data volume than the flow velocity field data and the temperature field data. This is an approximate solution to the lower-order approximation, and the lower-order approximation relation is: and ; The low-order approximation determination unit is used to determine the order based on the POD-RBF reduction algorithm. low-order approximation Represented as ,Will low-order approximation Represented as ; Among them, the It consists of p elements. , The internal scale thickness of the pipes in the heat exchanger corresponding to the simulation model, the inlet temperature of the pipes in the heat exchanger corresponding to the simulation model, and the inlet flow velocity are characterized. Depend on Composed of elements, The first element is , Characterizing the velocity field data, wherein , This is the coefficient matrix of the velocity field data. For the radial basis functions of the velocity field data. This is the coefficient matrix of the temperature field data. Let M be the radial basis function of the temperature field data, where M is a positive integer. and It is POD-based.

7. A digital twin modeling and intelligent sensing device for equipment, characterized in that, include: Memory, used to store computer programs; A processor, configured to execute the computer program to implement the steps of the device digital twin modeling and intelligent sensing method as described in any one of claims 1 to 5.

8. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, implements the steps of the device digital twin modeling and intelligent sensing method as described in any one of claims 1 to 5.

Images