Temperature sensing method of piezoelectric active vibration suppression system in low-temperature wind tunnel test

Abstract

The invention discloses a temperature sensing method of a piezoelectric active vibration suppression system in a low-temperature wind tunnel test. According to the method, a potential actual measurement point layout method is formulated according to simulation results under different low-temperature wind tunnel test working conditions in combination with actual working conditions, and intensive layout of measurement points is realized on the premise of ensuring key temperature acquisition; in order to realize acquisition of a large amount of discrete point temperature information, a simulation process is approximated through an agent model, a nonlinear mapping model between an actual measurement characteristic matrix and simulation node temperature is constructed through a simulation node layout method considering interpolation precision and rate research system discretization, and a deep learning method is used for training and solving the agent model, so that an actual measurement characteristic matrix is obtained. A large amount of discrete point temperature information of the system is rapidly obtained instead of finite element calculation; in order to realize accurate interpolation reconstruction of full-field temperature, a local large-gradient temperature interpolation method and a global smooth temperature interpolation method are researched, a full-field temperature interpolation weighted fusion model is constructed, a regional three-dimensional temperature field interpolation method is formed, and high-precision interpolation of a whole temperature field of a system is ensured.

Description

Technical Field

[0001] The invention belongs to the field of low-temperature wind tunnel tests and relates to a real-time temperature sensing method for an active vibration suppression system based on a proxy model combined with spatial interpolation. Background Art

[0002] Low-temperature wind tunnel tests can simulate full-scale Reynolds numbers and are the main means of obtaining key aerodynamic parameters of aircraft. To ensure measurement accuracy, the piezoelectric active vibration suppression system suppresses the vibration of the aircraft model during the wind tunnel test. Its core uses the piezoelectric effect to output reverse torque to reduce model vibration. As the temperature decreases, the strain capacity of controlling the output force and displacement of the piezoelectric actuator decreases significantly. Therefore, in order to effectively control the working temperature of the piezoelectric actuator and ensure the real-time response capability of the vibration suppression system, it is necessary to clarify its cooling law and instantaneous temperature information in the low-temperature wind tunnel environment to guide the active temperature control system. Considering the heat transfer characteristics, working environment and insulation structure limitations of the piezoelectric actuator itself, the lowest temperature is often distributed on the end face. However, the piezoelectric actuator is embedded in the support structure and is severely blocked, and there is axial extrusion, which makes it impossible to directly set the measurement point on the end face, making it difficult to monitor the lowest temperature of the piezoelectric actuator in real time, and maintaining the optimal output performance of the active vibration suppression system is challenging. Therefore, in order to improve the performance of the vibration suppression system in a low-temperature environment and ensure the safety and reliability of the aircraft design, it is of great significance to study the real-time temperature perception method of the piezoelectric active vibration suppression system.

[0003] In the paper "Transition detection by temperature sensitive paint at cryogenic temperatures in the European Transonic Wind tunnel (ETW)" published by Fey et al. of the German Aerospace Center at the 20th International Congress on Instrumentation in Aerospace Simulation Facilities, aiming at the problem that the surface temperature of the aircraft model cannot be monitored due to the influence of low temperature in non-contact measurement in the cryogenic transonic wind tunnel, which further leads to the difficulty in identifying boundary layer transition, a temperature measurement system based on temperature sensitive paint (TSP) technology was proposed. This temperature measurement system realizes high-precision measurement of the wing surface temperature of the three-dimensional model and high-precision acquisition of transition images in the temperature range of 115K < T < 180K. However, for the wing, the active vibration suppression system is embedded in the support structure and the end face is blocked, making it impossible to achieve full-field temperature perception of the system. Shen Xing et al. of Nanjing University of Aeronautics and Astronautics in the paper "Design and Analysis of a Thermally Insulating and Heating Scheme for Piezoelectric Stack Actuators Used in the Cryogenic Environment" aimed at the need to study the heating method of piezoelectric actuators in the low-temperature environment, constructed a two-dimensional theoretical heat transfer model to obtain the temperature distribution of piezoelectric actuators, and verified the effectiveness of the method through finite element analysis. This method can quickly evaluate the temperature distribution of piezoelectric actuators. However, due to the deviation between the model assumption and the actual situation, there are certain errors in the calculation results, and the boundary (end face) temperature needs to be applied for correction, which conflicts with the actual measurement requirements.

[0004] At present, in cryogenic wind tunnel tests, the regional temperature perception mostly adopts non-contact measurement methods such as TSP and theoretical calculation methods. However, the existing real-time temperature monitoring methods cannot achieve high-precision measurement of the full-field temperature under the constraints of blocked areas / contact surfaces. Therefore, there is an urgent need for a real-time and accurate full-field temperature perception method that meets the actual test requirements, does not affect the output displacement of the actuator, and is not restricted by severe structural blockages. Summary of the Invention

[0005] The present invention overcomes the shortcomings of existing methods and proposes a full-field temperature perception method for an active vibration suppression system under low-temperature wind tunnel conditions. The method first formulates a potential measurement point layout method based on the simulation results under different low-temperature wind tunnel test conditions and actual working conditions, proposes a characteristic measurement point optimization method based on statistical analysis, and forms an intensive layout of temperature sensors; secondly, the simulation node layout method of the system discretization is studied considering the interpolation accuracy and rate, and the temperature state of the measurement point is characterized by the measured characteristic matrix, and a nonlinear mapping model between the measured characteristic matrix and the simulation node temperature is constructed to achieve rapid acquisition of a large number of discrete point temperature information of the system instead of finite element calculation; finally, considering the uneven temperature distribution of the system, the local large gradient temperature interpolation method and the global smooth temperature interpolation method are studied, and a full-field temperature interpolation weighted fusion model is constructed to form a regional three-dimensional temperature field interpolation method. This method can realize real-time and accurate perception of the full-field dynamic temperature of the active vibration suppression system, laying a foundation for the subsequent formulation of a coordinated temperature control strategy for piezoelectric actuators in different directions throughout the field.

[0006] The technical solution of the present invention:

[0007] A temperature sensing method for a piezoelectric active vibration suppression system in a low-temperature wind tunnel test, the steps are as follows:

[0008] 1) Carry out finite element simulation analysis based on a typical active vibration suppression structure model, and obtain the temperature distribution law and sensitive areas through finite element simulation under different low-temperature wind tunnel test conditions; select the most representative measuring points by comprehensively considering the simulation analysis results, actual operation limitations and other factors, and determine the number and distribution of potential temperature sampling points; select potential measuring points through statistical analysis methods to achieve a small number of actual measuring points to represent the changes in the entire temperature field to the greatest extent, and determine the number and distribution of actual measuring points to form a method for optimizing measuring points; finally, conduct temperature measurement tests under low-temperature conditions to achieve the reproduction of the low-temperature wind tunnel test environment and the collection of actual temperature data. The details are as follows:

[0009] 1.1) Establish an active vibration suppression structure model, meet the actual work needs of low-temperature wind tunnel tests, change the test conditions such as the angle of attack and the incoming flow velocity, carry out finite element simulation under different low-temperature wind tunnel test conditions, and obtain the temperature distribution law and sensitive areas;

[0010] 1.2) According to the simulation results, determine the area with large temperature gradient variation, and select the most representative measuring points as potential temperature measurement points based on the actual operation feasibility, structural geometric characteristics, including the factors affecting the measurement accuracy of the sensor size and the force position;

[0011] 1.3) Based on the simulation results of 1.1, obtain the temperature time series data of potential measurement points under different test conditions, and calculate the standardized matrix X by principal component analysis (PCA) method. std The covariance matrix C of is:

[0012]

[0013] Among them, X is the potential measurement point coordinate matrix, μ is the mean value, and σ is the standard deviation. Then perform eigendecomposition Cv i =λ i v i , where λ i is the ith eigenvalue, v i is the same as λ i Select the component that has the greatest impact on the temperature change of the system, determine the number and distribution of actual measurement points, and complete the optimization of the measurement point location:

[0014] Y=X std V k (3)

[0015] Importance i =||Yi|| (4)

[0016] In the above process, V k is the matrix composed of the first k eigenvectors, Y is the projected data matrix, Y i is the projection vector of the i-th measurement point in the principal component space, ||Yi|| is its Euclidean norm;

[0017] 1.4) Build a small low-temperature wind tunnel test platform to reproduce the low-temperature wind tunnel test environment, arrange temperature sensors at the selected locations, conduct temperature measurement tests under different working conditions, and complete the collection of temperature time series data at the measuring points.

[0018] 2) For the task of online reconstruction of the three-dimensional temperature field, there needs to be enough temperature measurement points to cover the entire three-dimensional space and provide prediction results in a short time. A proxy model is proposed to approximate the simulation process and reduce the computational cost. Taking into account the interpolation fitting accuracy and calculation rate, a proxy model simulation node layout optimization method is designed to ensure that the optimized interpolation result achieves the best balance between speed and accuracy; the time domain characteristics (first-order and second-order derivatives) and frequency domain characteristics (amplitude, phase) of the measured point time series data are combined to form an input feature matrix, and the simulation node temperature information at the next moment is used as the output to construct a large data set; a deep learning method is used to construct a nonlinear mapping model from the measured point to the simulation node temperature, so as to realize the rapid and accurate acquisition of a large amount of discrete temperature information of the active vibration suppression system. The specific steps are as follows:

[0019] 2.1) Considering that accurate interpolation of the three-dimensional temperature field requires a large amount of data, the simulation results are approximated by the proxy model, a large number of discrete simulation nodes are selected, and the interpolation speed and accuracy are taken as the optimization goals. The number and layout of simulation discrete nodes are determined to form a simulation discrete node layout optimization method based on the ADAM (Adaptive Moment Estimation) method:

[0020]

[0021]

[0022] Formula (5) is the defined mean square error (MSE) loss function, which represents the interpolation accuracy. The interpolation speed is proportional to the number of nodes N. In the formula, M is the number of verification points, T j is the true temperature of the jth verification point, is the interpolated temperature of the jth verification point, and calculates the gradient of the loss function with respect to the node position Combined with the ADAM algorithm of formula (6) and formula (7), the first-order moment m is updated t , second-order moment v t , correct deviations, update node positions and continuously iterate and optimize, optimize the layout of simulation nodes, and ensure that the interpolation results achieve the best balance between speed and accuracy;

[0023] 2.2) Collect the temperature data of N nodes at time t+1 in the simulation, convert the time series data of the measured points into the frequency domain, extract the amplitude and phase of the main frequency components, combine the derivatives of the time series data to form the feature matrix of the measured points, construct the data set and split it (training set: validation set: test set = 6:2:2);

[0024] 2.3) Taking the measured point feature matrix as input and the instantaneous temperature of the simulation node as output, a nonlinear mapping relationship between the two is constructed based on the recurrent neural network (RNN):

[0025] h t =σ(W in x t +W h h t-1 +b h ) (8)

[0026] y t =softmax(W out h t +b out ) (9)

[0027] Where the time step is t and the input vector is x t ,h t is the hidden state vector, and the output vector is y t , W in is the input weight matrix, W h is the hidden state weight matrix, W out is the output weight matrix, b h , b outare the bias vectors corresponding to the hidden layer and the output layer, respectively. Here, the activation function is σ = tanh. Use the training set data to train the model, the validation set to adjust the model parameters, and the test set to evaluate the model accuracy;

[0028] 3) Divide the interpolation area based on the temperature gradient. The area with obvious temperature changes is the local area, and the rest is the global area; construct a large gradient local temperature interpolation model and a global smooth transition temperature interpolation model, and determine the model parameters through boundary conditions; study the combination method of multi-region interpolation, and achieve smooth transition of interpolation results in different regions through weighted averaging, so as to realize real-time and accurate reconstruction of the full-field temperature of the system through discrete point temperature information. The specific steps are as follows:

[0029] 3.1) Based on the temperature gradient, the interpolation area is divided into the area with obvious temperature changes as the local area, and the rest as the global area. For the discrete simulation nodes in the local area, the Gaussian function is selected as the basis function to construct the RBF interpolation model:

[0030]

[0031] Among them, the basis function φ(r) = e -∈r2 ,and ε is the shape parameter, m is the number of discrete points in the region, and w i is the weight, p(x,y,z) is the polynomial term, and the weight w is determined by minimizing the interpolation error i And the polynomial p(x,y,z):

[0032]

[0033] Use local radial basis functions (RBF) to achieve high-precision interpolation in areas with large temperature changes;

[0034] 3.2) For the three-dimensional interpolation of the global area, firstly, based on the cubic spline interpolation method, a single-dimensional cubic spline interpolation is performed, and a one-dimensional cubic spline interpolation is performed in the x direction (y and z are fixed), and the cubic spline interpolation function S in the x direction is obtained. x (x,y j ,z k ):

[0035]

[0036] Among them, the coefficient It is determined by the boundary condition that the first-order derivative and the second-order derivative are equal at the connection, and then a one-dimensional cubic spline interpolation is performed in the y direction (z is fixed), and S is obtained y (x,y,z k ):

[0037]

[0038] in, is the coefficient in the interpolation process. Finally, one-dimensional cubic spline interpolation is performed in the z direction to obtain the final three-dimensional cubic spline interpolation function S(x,y,z):

[0039]

[0040] in, is the coefficient. Through the above interpolation process, a smooth and accurate temperature interpolation result of the global area is obtained;

[0041] 3.3) The smooth transition between local and global interpolation results is achieved through distance weighted averaging. For the query point T(x q ,y q ,z q ), the temperature interpolation result is expressed as:

[0042]

[0043] The beneficial effect of the present invention is to provide a temperature sensing method for a piezoelectric active vibration suppression system in a low-temperature wind tunnel test for the task of real-time sensing of the full-field temperature in consideration of the narrow closed test space, severe structural obstruction, and unclear extreme temperature distribution of the active vibration suppression system. Compared with the traditional method, which has problems such as the inability to measure the obstructed area and poor theoretical modeling accuracy, the method proposed in the present invention can form an optimized layout of actual measurement points for the narrow closed space in the active vibration suppression system of the low-temperature wind tunnel, construct an agent model to simulate and obtain a large amount of discrete temperature information, accurately interpolate and fit the three-dimensional temperature field by region, and solve and predict the full-field temperature information of the system in real time. In the temperature assurance process of the active vibration suppression system in the low-temperature wind tunnel test, the measurement requirements are low, the area is large, the response is fast, and the accuracy is high. It can ensure real-time and non-destructive monitoring of the full-field temperature, and provide data support for the precise control of the full-field temperature of the vibration suppression section. This method is applicable to many types of active vibration suppression systems. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] Figure 1 Schematic diagram of the piezoelectric actuator output performance changing with temperature.

[0045] Figure 2 It is a flowchart of the temperature sensing method of the low-temperature wind tunnel piezoelectric active vibration suppression system of the present invention.

[0046] Figure 3 It is a low-temperature wind tunnel environment temperature measurement test platform.

[0047] Figure 4 This is a schematic diagram of the simulation agent model construction and training process of the present invention.

[0048] Figure 5 This is a schematic diagram of the temperature field reconstruction principle by spatial interpolation of the method of the present invention. DETAILED DESCRIPTION

[0049] The specific implementation of the present invention is further described below in conjunction with the accompanying drawings and technical solutions.

[0050] (1) This method is based on a piezoelectric active vibration suppression system built by the research team. It includes a tail support rod support system and circumferentially evenly distributed piezoelectric actuators. In order to improve the test efficiency, a key section of the active vibration suppression system is established. The key section model can accurately reflect the thermal behavior of all areas of interest in the system. Based on the laboratory environment, a low-temperature wind tunnel environmental temperature measurement test platform is built, including small wind tunnels (such as Figure 3 As shown in the figure, it includes the test section, compression section and steady flow section), liquid nitrogen tank, key section of active vibration suppression system, measurement platform based on LabVIEW, thermocouple temperature sensor and host computer. Before the low-temperature environment temperature measurement test begins, it is necessary to design the layout of the actual measurement points to ensure the temperature collection at the key positions and reduce the number of measurement points as much as possible. In order to obtain the key temperature measurement area, finite element simulations are carried out under various working conditions to determine the cooling law and temperature sensitive areas, and the layout of potential actual measurement points is determined in combination with the feasibility of implementation. Then, the principal component analysis method is used to select the components with the greatest impact on the temperature change of the system among the potential measurement points, and the final layout of the actual measurement points is determined. Finally, the thermocouple sensor of this method is statically and dynamically calibrated using accurately calibrated temperature measurement equipment to improve the accuracy of temperature measurement, and the liquid nitrogen flow rate is controlled by valve pressure, and the angle of attack is adjusted by the scimitar device to collect the temperature time series signals of the actual measurement points under different test conditions (corresponding to the simulation).

[0051] (2) After obtaining the measured data of some sampling points, in order to realize the three-dimensional reconstruction of the temperature field, a large amount of data needs to be prepared to ensure the accuracy of the reconstruction results. A large amount of discrete temperature information can be obtained through the simulation process of the proxy model and the calculation time can be reduced. Figure 4 As shown in the figure. First, a large number of finite element simulations are carried out based on the key segment model of the vibration suppression system to cover most of the test conditions. A large number of discrete points are selected from them to ensure that the subsequent interpolation reconstruction has sufficient data support, and the number and position of simulation nodes are adjusted according to the interpolation speed and reconstruction accuracy to form a simulation node layout optimization method; then, a proxy model is constructed to characterize the nonlinear mapping relationship between the measured point temperature and the simulation node temperature, and the proxy model is solved by a recurrent neural network. The first and second order derivatives (time domain characteristics) and amplitude and phase (frequency domain characteristics) of the measured temperature time series data are extracted to form an input feature matrix, and the simulation node temperature information at the next moment is obtained as output. The input feature matrix and output vector under each working condition constitute a large data set; finally, the large data set is divided according to the ratio of 6:2:2 (training set: validation set: test set), the training set data is used to train the model, the validation set is used to adjust the model parameters, and the test set is used to evaluate the model accuracy to achieve the acquisition of discrete temperature information of the active vibration suppression system. Ensure that the temperature information of a large number of discrete points in the system can be quickly obtained and predicted instead of finite element calculation.

[0052] (3) After obtaining a large amount of discrete temperature information under different test conditions, the interpolation area is divided based on the temperature gradient change. The principle is as follows: Figure 5 As shown in the figure. First, for the local area with large temperature gradient, the RBF interpolation model is constructed, and the model parameters are optimized by minimizing the interpolation error to achieve high-precision interpolation in the area with large temperature changes; then, for the global area with small temperature gradient, a multidimensional cubic spline interpolation model is constructed, and the model coefficients are determined by boundary conditions to obtain smooth and accurate temperature interpolation results in the global area; finally, for the whole system, a full-field temperature interpolation weighted fusion model is constructed to form a regional three-dimensional temperature field interpolation method, which realizes a smooth transition between local and global interpolation results and ensures the real-time perception and prediction of the full-field temperature information of the active vibration suppression system.

Claims

1. A temperature sensing method for a piezoelectric active vibration suppression system in a low-temperature wind tunnel test, characterized in that: The following steps are involved: 1) Based on the wind tunnel tail support active vibration suppression system, finite element simulation analysis is carried out to visualize the temperature distribution of piezoelectric actuators in all directions by drawing temperature cloud maps, revealing the temperature distribution law and sensitive areas; Taking into account the simulation analysis results and the limitations of actual operations, the most representative temperature measurement points are selected to determine the number and distribution of potential temperature measurement points. Potential temperature measurement points are optimized through statistical analysis methods to ensure that a small number of actual measurement points can reflect the changes in the entire temperature field to the greatest extent, and the number and distribution of actual measurement points are finally determined to form a method for optimizing measurement points. Finally, a low-temperature wind tunnel environmental temperature measurement test platform is built, and comprehensive calibration is carried out on the temperature sensors used in the temperature measurement test. Temperature measurement tests are carried out under low-temperature conditions to reproduce the low-temperature wind tunnel test environment and collect actual temperature data. 1.1) Establish the geometric model of the wind tunnel tail strut active vibration suppression system, including the tail strut, piezoelectric actuator, piezoelectric balance and scimitar structure; use tetrahedron and hexahedron elements for meshing, and perform local encryption in the key area, i.e. the installation position of the piezoelectric actuator; Aiming at the actual work needs of low-temperature wind tunnel tests, the angle of attack and the incoming flow velocity are changed, and finite element simulations of various working conditions under low-temperature wind tunnel environments are carried out. The temperature field distribution is calculated using the heat transfer equation as shown below: Where k is thermal conductivity, Q is heat source term, ρ is density, c p is the specific heat capacity, T is the temperature; 1.2) According to the simulation results of the above temperature field distribution, the area with drastic temperature changes is identified, and the temperature gradient of each node in the temperature field is calculated using the gradient operator: Draw a temperature cloud map, identify areas with large gradient changes, and select a large number of potential temperature measurement points based on the actual operational feasibility and structural geometric characteristics, including sensor size, sensor installation difficulty, and whether it is a stress concentration area; 1.3) Based on the partial simulation results of step 1.1), obtain the temperature time series data of potential temperature measurement points under different test conditions, organize the extracted temperature time series data into a matrix form, where each column represents the temperature time series data of a temperature measurement point, check the data and perform standardization processing on the data: Where X is the potential temperature measurement point coordinate matrix, μ is the mean value, and σ is the standard deviation; The covariance matrix C is calculated for the standardized matrix as follows: in, Yes X std The transposed matrix of , s is the number of potential temperature measurement points; Then perform the eigendecomposition: Cv i =λ i v i (5) Among them, λ i is the ith eigenvalue, v i is the same as λ i The corresponding eigenvector; The potential temperature measurement points are prioritized according to the size of the eigenvalues, and the eigencomponents whose cumulative contribution rate to the temperature change of the wind tunnel tail support active vibration suppression system reaches more than 85% are selected as the actual measurement points to complete the optimization of the actual measurement point layout: Y=X std V k (6) Importance i =||Do|| (7) Among them, V k is the matrix composed of the first k eigenvectors, Y is the projected data matrix, Y i is the projection vector of the i-th measurement point in the principal component space, ||Yi|| is its Euclidean norm; 1.4) Build a low-temperature wind tunnel environment temperature measurement test platform, including a wind tunnel, a liquid nitrogen tank, a key section of the active vibration suppression system, a measurement platform based on Labview, a thermocouple temperature sensor and a host computer. The wind tunnel includes a test section, a compression section and a steady flow section. The key section of the active vibration suppression system is placed in the test section of the wind tunnel. The liquid nitrogen tank is connected to the steady flow section through a pipeline. Thermocouple temperature sensors are arranged at the selected measuring points. The thermocouple temperature sensors are connected to the temperature acquisition board of the measurement platform based on Labview. The measurement platform based on Labview is connected to the host computer for data transmission. The temperature measurement tests under different working conditions are completed to obtain the temperature time series data of the measuring points. 2) Construct a proxy model from the measured point to the simulated discrete point temperature to approximate the simulation process; comprehensively consider the interpolation fitting accuracy and calculation rate, design a proxy model simulation node layout optimization method to achieve uniform distribution of nodes and coverage of key areas; combine the various order derivatives and frequency domain characteristics of the measured point time series data to form an input feature matrix, take the simulation node temperature information at the next moment as output, and construct a large data set; use deep learning methods to construct a nonlinear mapping model from the measured point to the simulation node temperature, and realize the acquisition of discrete temperature information of the active vibration suppression system; 2.1) In order to achieve accurate interpolation of the three-dimensional temperature field, a large number of temperature measurement points are required to cover the entire three-dimensional space; the temperature information of a large number of discrete points is obtained through finite element simulation, and then the simulation results are approximated by the simulation proxy model to quickly capture the nonlinear mapping relationship between the limited temperature measurement points and other discrete nodes; the layout of a large number of discrete nodes in the simulation affects the final temperature field reconstruction accuracy. First, a large number of discrete nodes are extracted from the finite element simulation results, and the positions of the discrete nodes are initialized to ensure that the discrete nodes cover the entire three-dimensional space and are distributed as evenly as possible; the mean square error loss function is defined to measure the influence of the selected simulation discrete nodes on the interpolation accuracy: Where M is the number of simulation discrete nodes, T j is the true temperature of the jth simulation discrete node, is the interpolated temperature of the jth simulation discrete node, and calculates the gradient of the loss function with respect to the position of the simulation discrete node Taking interpolation accuracy and calculation speed as optimization goals, the ADAM algorithm is used to update the simulation discrete node positions: In the formula, m t is the first-order moment, v t is the second-order moment, and the deviation is corrected through multiple iterations; is the gradient of the loss function with respect to the position of the simulation discrete node; β1 and β2 are the attenuation rates; β1 controls the smoothness of the first-order moment estimation, and β2 controls the smoothness of the second-order moment estimation; the positions of the simulation discrete nodes are updated, and the layout of the simulation discrete nodes is optimized until the loss function reaches the minimum value; during the optimization process, the accuracy and speed of the interpolation results are regularly verified, and the optimization parameters are adjusted according to the verification results to form a simulation discrete node layout optimization method; 2.2) Collect the temperature data of the next moment of n simulation discrete nodes in the simulation, convert the time series data T(t) of the measured point into the frequency domain through Fourier transform, obtain the frequency domain representation T(f), extract the amplitude and phase of the main frequency components, use the amplitude and phase of the extracted main frequency components as the frequency domain features of the measured temperature, and calculate the derivatives of each order of the measured temperature data to form the time domain features. The frequency domain features and time domain features together constitute the measured feature matrix. Take the measured feature matrix as input and the temperature of the simulation node at the next moment as the output vector to construct a large data set, which is divided into a training set, a validation set and a test set in a ratio of 6:2:2; 2.3) Taking the measured feature matrix as input and the instantaneous temperature of the simulation node as output, a simulation proxy model is constructed based on a recurrent neural network to capture the temporal dependency between the measured data and the simulation node information and characterize the complex nonlinear relationship between the two. The simulation proxy model is shown in the following formula: h t =σ(W in x t +W h h t-1 +b h ) (11) y t =softmax(W out h t +b out ) (12) Where t is the time step, x t is the input vector, h t is the hidden state vector, y t is the output vector, W in is the input weight matrix, W h is the hidden state weight matrix, W out is the output weight matrix, b h 、b out are the bias vectors corresponding to the hidden layer and the output layer respectively, where the activation function is σ=tanh; use the training set in step 2.2) to train the simulation proxy model, the validation set to adjust the simulation proxy model parameters, and the test set to evaluate the accuracy of the simulation proxy model; 3) Taking into account the uneven temperature distribution of the system, the interpolation area is divided based on the temperature gradient, and the local large gradient temperature interpolation method and the global smooth temperature interpolation method are studied. The combination method of multi-region interpolation is studied, and a weighted fusion model of full-field temperature interpolation is constructed to form a regional three-dimensional temperature field interpolation method, so as to realize real-time and accurate reconstruction of the full-field temperature of the system through discrete point temperature information; 3.1) According to the temperature field simulation results, the areas with obvious temperature changes are identified. These areas have high temperature gradients and require high accuracy of temperature interpolation. They are set as local areas. The areas with relatively gentle temperature changes are divided into global areas. For the discrete simulation nodes in the local area, the Gaussian function is selected as the basis function of the RBF interpolation model. The RBF interpolation model is as follows: Among them, the basis function φ(r) = e -∈r2 ,and ε is the shape parameter, m is the number of discrete points in the region, and w i is the weight, p(x,y,z) is the polynomial term, and the weight w is determined by minimizing the interpolation error i And the polynomial p(x,y,z): Combine the basis function φ(r) and the polynomial p(x,y,z) to construct the augmented matrix A and augmented vector b: Among them, T i is the temperature value of the ith discrete node. The weight w is determined by solving the linear equations numerically. i And the polynomial p(x,y,z) is shown as follows: Among them, W is the weight coefficient vector, C is the polynomial coefficient vector, and the coefficients are brought into the RBF interpolation model to form a temperature interpolation method for the local area; 3.2) For the three-dimensional interpolation of the global area, firstly, based on the cubic spline interpolation method, a single-dimensional cubic spline interpolation is performed, and a one-dimensional cubic spline interpolation is performed in the x direction, with the y and z directions fixed, to obtain the cubic spline interpolation function S in the x direction: x (x,y j ,z k ): Among them, the coefficient It is determined by the boundary condition that the first-order derivative and the second-order derivative are equal at the connection, and then a one-dimensional cubic spline interpolation is performed in the y direction (z is fixed) to obtain S y (x,y,z k ): in, is the coefficient in the interpolation process; finally, one-dimensional cubic spline interpolation is performed in the z direction to obtain the final three-dimensional cubic spline interpolation function S(x,y,z): in, is the coefficient. Through the above interpolation process, a smooth and accurate temperature interpolation result of the global area is obtained; 3.3) The smooth transition between local and global interpolation results is achieved through distance weighted averaging. For the query point T(x q ,y q ,z q ), the temperature interpolation result is expressed as: Among them, T RBF Output temperature result for local area RBF interpolation model, T Spline The temperature results are output for the global regional cubic spline interpolation model, where α is the proportion of local results. A regional three-dimensional temperature field interpolation method is formed to achieve real-time and accurate perception of the full-field dynamic temperature of the active vibration suppression system.

Images