Sparse sensor layout optimization method for turbine disk temperature field inversion

Abstract

A sparse sensor layout optimization method for turbine disk temperature field inversion, which obtains snapshot data of turbine disk temperature field evolution over time by finite element simulation at equal time intervals, and constructs it as structured tensor data, then constructs a reduced order model through high order tensor decomposition, subsequently obtains an optimized sensor layout through tensor fiber-pivoted orthogonal triangular decomposition, and finally realizes accurate reconstruction of turbine disk temperature field based on tensor compressive sensing method. The invention solves the problem of difficulty in realizing accurate turbine disk temperature field inversion using limited sparse sensor layout by capturing the physical information of turbine disk temperature field and constructing an accurate reduced order model, and has broad application prospects in the field of surface temperature field measurement of large rotating thermal mechanical hot end components.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of turbine disk temperature measurement system optimization, and particularly relates to a sparse sensor layout optimization method for turbine disk temperature field inversion. BACKGROUND

[0002] The turbine disk functions to convert part of thermal energy and potential energy in high-temperature fluid into mechanical energy and drive the operation of the remaining working components in the system. The turbine disk works in extremely harsh conditions and bears complex force and thermal load during work. The stress and temperature borne by each part are different, and particularly, the turbine disk requires as high fatigue, endurance and good anti-creep ability as possible within the allowable temperature range. In order to better guarantee the safe operation of the system, the turbine disk residual service life prediction method based on accurate turbine disk temperature field measurement has become one of the key technologies for realizing predictive maintenance from visual maintenance for complex mechanical systems.

[0003] At present, the turbine disk mainly realizes non-contact measurement of the turbine disk surface measurement point temperature by using a specially-made high-temperature glue to paste a silicon carbide micro temperature measurement crystal on the turbine disk surface, wherein the installation position of the temperature measurement crystal depends on manual experience test selection. In the SGT-800 gas turbine developed by the German Siemens Energy Company, as many as 90 temperature measurement crystals are installed on the surface of each blade to realize accurate measurement of the blade surface temperature field.

[0004] In the existing turbine disk surface temperature field field measurement technology, there are problems such as that the temperature measurement crystal layout depends on expert experience, and a large number of redundant temperature measurement points are required for high-precision inversion of the temperature field. Limited by the engineering conditions such as measurement cost and sensor weight, it is difficult to meet the high-precision requirements of the turbine disk surface temperature field field measurement of the complex system, and cannot be applied to the residual service life prediction technology of the turbine disk.

[0005] The above information disclosed in the background section is only intended to enhance the understanding of the background of the present application, and therefore can include information that is not prior art known to those of ordinary skill in the art. SUMMARY

[0006] In view of the problems in the prior art, the present application provides a sparse sensor layout optimization method for turbine disk temperature field inversion, which is a new means and method for installing temperature measurement crystals on the turbine disk in an optimized layout based on high-order tensor decomposition and tensor compressed sensing and realizing accurate inversion of the turbine disk temperature field under different system operating conditions.

[0007] The present application is implemented by the following technical solutions:

[0008] A sparse sensor layout optimization method for turbine disk temperature field inversion comprises the following steps:

[0009] Step A: Simulate and calculate the trend of the thermal parameters of the turbine disk evolving with the operating time under different operating states, sample the temperature field information on the surface of the turbine disk at equal time intervals dt to obtain temperature field time snapshot data, and save it in a binary data serialization format containing discrete unstructured grid point coordinates and corresponding temperature numerical information;

[0010] Step B: Process the temperature field information of each working condition into tensor data in the tensor data format;

[0011] Step C: Sequentially perform high-order singular value decomposition on all the tensor data, and decompose to obtain the core tensor and the modal matrix;

[0012] Step D: Construct a reduced-order model of the dynamic system of the temperature field of the turbine disk evolving with time based on the obtained multiple core tensors and modal matrices, and obtain the dictionary tensor A for reconstructing the temperature field of the turbine disk;

[0013] Step E: Perform tensor train-pivot orthogonal triangular decomposition on the dictionary tensor A to obtain the permutation matrix P and the optimized sensor layout;

[0014] Step F: Obtain the measured values of the temperature on the surface of the turbine disk at the corresponding installation positions by installing according to the optimized sensor layout through a test run or setting virtual sensors by finite element simulation, and replace the values at the corresponding index positions in the permutation matrix P to obtain the measurement matrix And ||Y||0 = N;

[0015] Step G: Reconstruct the temperature field on the surface of the turbine disk using the tensor compressive sensing algorithm based on the dictionary tensor A and the measurement matrix Y.

[0016] In the sparse sensor layout optimization method for the temperature field inversion of the turbine disk, the trend of the thermal parameters of the turbine disk evolving with the operating time is simulated and calculated by means of finite element simulation or numerical simulation, where the different operating states include start-up, variable operating conditions and rated operating conditions.

[0017] In the sparse sensor layout optimization method for the temperature field inversion of the turbine disk, the turbine disk is the pre-pressure turbine pump turbine disk of a liquid rocket engine.

[0018] In the sparse sensor layout optimization method for the temperature field inversion of the turbine disk, Step B includes:

[0019] Step B1: Read in all the temperature field time snapshot data under the s-th (0 < s ≤ n) working condition in chronological order;

[0020] Step B2: Perform quadratic linear interpolation and resampling on the coordinates of discrete unstructured grid points in the single-time snapshot data according to a given sampling rate to obtain structured two-dimensional matrix data;

[0021] Step B3: Stack the two-dimensional matrix data with time sequence as the third dimension, and finally convert all temperature field time snapshot data under a single working condition into tensor data;

[0022] Step B4: Perform steps B1 to B3 sequentially on the data for all operating conditions to obtain multiple tensor data converted from the time snapshot data of the temperature field under each operating condition. Where i and j are the spatial dimensions of the temperature field, and k represents the product of the number of operating conditions n and the number of time snapshots of the temperature field under each operating condition.

[0023] In the sparse sensor layout optimization method for turbine disk temperature field inversion, step C includes:

[0024] Step C1: Set the residual tolerance value ε, and the cutoff size r is automatically determined by the value of the residual tolerance value ε;

[0025] Step C2: For tensor data The core tensor is obtained by performing higher-order singular value decomposition. and mode matrix and Where p, q, and r are the dimensions of the core tensor in the spacetime dimension;

[0026] Step C3: Perform steps C1 to C2 sequentially on all operating condition data to obtain multiple core tensor vectors obtained from the decomposition of all tensor data. And the modal matrix vector [A1,A2,…,A] n ]、[B1,B2,…,B n ] and [C1,C2,…,C n ].

[0027] In the sparse sensor layout optimization method for turbine disk temperature field inversion, step D includes:

[0028] Step D1: Utilizing the core tensor and mode matrix A s and B s Calculate the reduced-order model for the s-th working condition. Where r is the cutoff size of the reduced-order model;

[0029] Step D2: For s, with a step size of 1, execute step D1 sequentially within the range of 1 to n to obtain the reduced-order model tensor corresponding to each of the operating conditions.

[0030] Step D3: Stack all the reduced-order model tensors according to the actual operating conditions during engine operation in the order of the tube fiber direction of the tensors for all the reduced-order models corresponding to the operating conditions to obtain a dictionary tensor for reconstructing the temperature field of the turbine disk

[0031] In the sparse sensor layout optimization method for turbine disk temperature field inversion, step D1 includes:

[0032] Step D11: The h-th (0 < h ≤ r) front slice of the reduced-order model is:

[0033] Step D12: For h, with a step size of 1, in the range of values from 1 to r, loop and execute step D11 to obtain the reduced-order model

[0034] In the sparse sensor layout optimization method for turbine disk temperature field inversion, step E includes:

[0035] Step E1: Set the number of sensors to be selected as N, and set the truncation size of the TFQR decomposition algorithm to N;

[0036] Step E2: Define the positions where temperature sensors are restricted to be installed in the turbine disk and the positions corresponding to the absence of physical entities in the structured matrix as the rejection regions, and the remaining positions except the rejection regions as the acceptance regions

[0037] Step E3: Execute the TFQR decomposition algorithm on the dictionary tensor to obtain a permutation matrix where ||P||0 = N;

[0038] Step E4: Obtain the position indices of all N non-zero values in the permutation matrix P, and the position indices are the optimized sensor layout positions.

[0039] In the sparse sensor layout optimization method for turbine disk temperature field inversion, step G includes:

[0040] Step G1: Initialize and set the parameters of the TCS algorithm, which include the maximum number of iterations t, the algorithm tolerance ε tol and the regularization coefficient λ;

[0041] Step G2: Use the TCS algorithm to solve to obtain the sparse modal vector

[0042] Step G3: Calculate the reconstructed turbine disk surface temperature field structured matrix data Y hat , given

[0043] In the sparse sensor layout optimization method for turbine disk temperature field inversion, the turbine disk is located in a liquid rocket engine turbopump, an aero gas turbine engine, or a gas turbine.

[0044] Compared with the prior art, the present invention has the following advantages:

[0045] This invention employs a tensor-based reduced-order model construction method to model and characterize the dynamic system characteristics of the turbine disk temperature field evolution over time. This accurately captures the physical characteristics of the temperature field evolution, effectively reducing the problem dimensionality during temperature field reconstruction, lowering computational resource requirements, and improving time efficiency. By fully utilizing machine learning methods based on the temperature field reconstruction error minimization algorithm, it effectively guides the sensor layout in the design of the turbine disk temperature field measurement system, effectively approximating the optimal sparse sensor layout. This effectively reduces the number of sensors while ensuring accurate temperature field reconstruction. It also meets the requirements of lightweight design for hot-end accessories in large rotating thermodynamic machinery, while effectively reducing the time and economic cost of building a precise physical quantity measurement system. This invention is not limited to the specific application scenario of turbine disk temperature field measurement optimization; it can be used for measurement optimization scenarios in various complex mechanical equipment systems where multiple components evolve over time and exhibit univariate correlations under given operating conditions. Attached Figure Description

[0046] Various other advantages and benefits of the present invention will become apparent to those skilled in the art upon reading the detailed description of the preferred embodiments below. The accompanying drawings are for illustrative purposes only and are not intended to limit the invention. It is obvious that the drawings described below are merely some embodiments of the invention, and those skilled in the art can obtain other drawings based on these drawings without any inventive effort. Furthermore, the same reference numerals denote the same parts throughout the drawings.

[0047] In the attached diagram:

[0048] Figure 1 This is a flowchart of the sparse sensor layout optimization method for turbine disk temperature field inversion described in this invention.

[0049] Figure 2 Here is a flowchart of the higher-order singular value decomposition algorithm;

[0050] Figure 3 Here is a flowchart of the tensor fiber-pivot orthogonal triangulation algorithm;

[0051] Figure 4 Here is a flowchart of the tensor compressed sensing algorithm;

[0052] Figure 5 Sparse sensor layout for turbine disk temperature field and temperature field inversion effect based on optimized sensor layout.

[0053] The present invention will be further explained below with reference to the accompanying drawings and embodiments. Detailed Implementation

[0054] Specific embodiments of the invention will now be described in more detail with reference to the accompanying drawings. While specific embodiments of the invention are shown in the drawings, it should be understood that the invention may be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that this invention will be thorough and complete, and will fully convey the scope of the invention to those skilled in the art.

[0055] It should be noted that certain terms are used in the specification and claims to refer to specific components. Those skilled in the art will understand that different terms may be used to refer to the same component. This specification and claims do not distinguish components based on differences in terminology, but rather on differences in function. The terms "comprising" or "including" used throughout the specification and claims are open-ended and should be interpreted as "comprising but not limited to." The following descriptions are preferred embodiments for carrying out the invention; however, these descriptions are for the purpose of understanding the general principles of the specification and are not intended to limit the scope of the invention. The scope of protection of this invention is determined by the appended claims.

[0056] To facilitate understanding of the embodiments of the present invention, further explanations and descriptions will be provided below with reference to the accompanying drawings and specific embodiments. The accompanying drawings do not constitute a limitation on the embodiments of the present invention.

[0057] like Figures 1 to 5 As shown, the sparse sensor layout optimization method for turbine disk temperature field inversion includes the following steps:

[0058] Step A: Simulate and calculate the trend of the evolution of the turbine disk thermodynamic parameters with operating time under different operating conditions. Sample the temperature field information on the surface of the turbine disk at equal time intervals dt to obtain temperature field time snapshot data, and save it as a binary data serialization format containing discrete unstructured grid point coordinates and corresponding temperature values.

[0059] Step B: Process the temperature field time snapshot data for each working condition into tensor data in tensor data format;

[0060] Step C: Perform high-order singular value decomposition on all the tensor data in sequence, and decompose to obtain a core tensor and modal matrices;

[0061] Step D: Construct a reduced-order model of the dynamic system for the evolution of the temperature field of the turbine disk over time based on the obtained multiple core tensors and modal matrices, and obtain a dictionary tensor A for reconstructing the temperature field of the turbine disk;

[0062] Step E: Perform tensor fiber-pivot orthogonal triangular decomposition on the dictionary tensor A to obtain a permutation matrix P and an optimized sensor layout;

[0063] Step F: Obtain the measured values of the surface temperature of the turbine disk at the corresponding installation positions by installing according to the optimized sensor layout through a test run or setting virtual sensors by finite element simulation, and replace the values at the corresponding index positions in the permutation matrix P to obtain a measurement matrix And ||Y||0 = N;

[0064] Step G: Reconstruct the surface temperature field of the turbine disk using the tensor compressive sensing algorithm based on the dictionary tensor A and the measurement matrix Y.

[0065] In a preferred embodiment of the sparse sensor layout optimization method for turbine disk temperature field inversion, the trend of the thermal parameters of the turbine disk evolving over the operating time is simulated and calculated by means of finite element simulation or numerical simulation, where different operating states include start-up, off-design conditions, and rated conditions.

[0066] In a preferred embodiment of the sparse sensor layout optimization method for turbine disk temperature field inversion, the turbine disk is the pre-pressure turbine pump turbine disk of a liquid rocket engine.

[0067] In a preferred embodiment of the sparse sensor layout optimization method for turbine disk temperature field inversion, Step B includes:

[0068] Step B1: Read in all the temperature field time snapshot data under the s-th (0 < s ≤ n) condition in chronological order;

[0069] Step B2: Perform quadratic linear interpolation and resampling on the discrete unstructured grid point coordinates of a single time snapshot data according to a given sampling rate to obtain structured two-dimensional matrix data;

[0070] Step B3: Stack the two-dimensional matrix data with time sequence as the third dimension, and finally convert all the temperature field time snapshot data under a single condition into tensor data;

[0071] Step B4: Sequentially perform Steps B1 to B3 on the data of all conditions to obtain multiple tensor data converted from all the temperature field time snapshot data under each condition in all conditions Where i and j are the spatial dimensions of the temperature field, and k represents the product of the number of operating conditions n and the number of time snapshots of the temperature field under each operating condition.

[0072] In a preferred embodiment of the sparse sensor layout optimization method for turbine disk temperature field inversion, step C includes:

[0073] Step C1: Set the residual tolerance value ε, and the cutoff size r is automatically determined by the value of the residual tolerance value ε;

[0074] Step C2: For tensor data The core tensor is obtained by performing higher-order singular value decomposition. and mode matrix and Where p, q, and r are the dimensions of the core tensor in the spacetime dimension;

[0075] Step C3: Perform steps C1 to C2 sequentially on all operating condition data to obtain multiple core tensor vectors obtained from the decomposition of all tensor data. And the modal matrix vector [A1,A2,…,A] n ]、[B1,B2,…,B n ] and [C1,C2,…,C n ].

[0076] In a preferred embodiment of the sparse sensor layout optimization method for turbine disk temperature field inversion, step D includes:

[0077] Step D1: Utilizing the core tensor and mode matrix A s and B s Calculate the reduced-order model for the s-th working condition. Where r is the cutoff size of the reduced-order model;

[0078] Step D2: For s, with a step size of 1, execute step D1 sequentially within the range of 1 to n to obtain the reduced-order model tensor corresponding to each of the operating conditions.

[0079] Step D3: Apply the reduced-order model tensors to all operating conditions according to the actual working conditions of the engine, following the fiber orientation of the tensors, to the reduced-order models corresponding to all operating conditions. Stacking the tensors yields a dictionary tensor for reconstructing the turbine disk temperature field.

[0080] In a preferred embodiment of the sparse sensor layout optimization method for turbine disk temperature field inversion, step D1 includes:

[0081] Step D11: Reduced-order model The h-th (0 < h ≤ r) front slice of

[0082] is:

[0083] In a preferred embodiment of the sparse sensor layout optimization method for turbine disk temperature field inversion, step E includes:

[0084] Step E1: Set the number of sensors selected as N, and set the truncation size of the TFQR decomposition algorithm as N;

[0085] Step E2: Define the positions where temperature sensors in the turbine disk are restricted from being installed and the positions corresponding to the absence of physical entities in the structured matrix as the rejection region, and the remaining positions except the rejection region as the acceptance region

[0086] Step E3: Perform the TFQR decomposition algorithm on the dictionary tensor to obtain the permutation matrix where ||P||0 = N;

[0087] Step E4: Obtain the position indices of all N non-zero values in the permutation matrix P, and the position indices are the optimized sensor layout positions.

[0088] In a preferred embodiment of the sparse sensor layout optimization method for turbine disk temperature field inversion, step G includes:

[0089] Step G1: Initialize and set the TCS algorithm parameters, which include the maximum number of iterations t, the algorithm tolerance ε tol , and the regularization coefficient λ;

[0090] Step G2: Use the TCS algorithm to solve to obtain the sparse modal vector

[0091] Step G3: Calculate the reconstructed structured matrix data Y of the turbine disk surface temperature field hat , given by

[0092] In a preferred embodiment of the sparse sensor layout optimization method for turbine disk temperature field inversion, the turbine disk is provided in a liquid rocket engine turbine pump, an aero gas turbine engine or a gas turbine.

[0093] In one embodiment, taking the inversion of the temperature field on the surface of a turbine disk by a liquid rocket engine as an example, the specific implementation includes the following steps:

[0094] Step A: Use finite element simulation or numerical simulation to perform high-fidelity simulation calculations on the trend of the thermal parameters of the turbine disk evolving with the operating time under different operating conditions (start-up, variable conditions, rated conditions) of the liquid rocket engine. Sample the temperature field information on the surface of the turbine disk at equal time intervals dt, and save it in a binary data serialization format containing discrete unstructured grid point coordinates and corresponding temperature numerical information;

[0095] Step B: Process the temperature field information of each condition into the tensor data format used in the method of the present invention. The specific steps of Step B are as follows:

[0096] Step B1: Read in all the temperature field time snapshot data under the s-th (0 < i ≤ n) condition in chronological order;

[0097] Step B2: Perform bilinear interpolation and resampling on the discrete unstructured grid point coordinates for a single time snapshot data at a given sampling rate to obtain structured two-dimensional matrix data;

[0098] Step B3: Stack the two-dimensional matrix with the third dimension being the time sequence. Finally, convert all the temperature field time snapshot data under a single condition into tensor data;

[0099] Step B4: Sequentially perform Steps B1 - B3 on all condition data to obtain multiple tensor data converted from all the temperature field time snapshot data under each condition in all conditions where the number of tensor data is equal to the number of conditions n;

[0100] Step C: Sequentially perform the high-order tensor decomposition algorithm on all the tensor data constructed in Step B to decompose and obtain the core tensor and modal matrices. The detailed process of the HOSVD algorithm is as Figure 2 described; the specific steps of Step C are as follows:

[0101] Step C1: Set the residual tolerance value ε, and at the same time, the truncation size r is automatically determined by the value of ε;

[0102] Step C2: Perform high-order singular value decomposition on the tensor to obtain the core tensor and the modal matrices and

[0103] Step C3: Sequentially perform Steps C1 - C2 on all condition data to obtain multiple core tensors decomposed from all the tensor data and modal matrices [A1, A2, …, A n , [B1, B2, …, B n , and [C1, C2, …, C n ;

[0104] Step D: Construct a reduced - order model of the dynamic system for the evolution of the turbine disk temperature field over time based on the multiple core tensors and modal matrices obtained in Step C, and obtain a dictionary tensor for turbine disk temperature field reconstruction. The specific steps of Step D are as follows:

[0105] Step D1: Use the core tensor and modal matrices A s and B s to calculate the reduced - order model for the sth operating condition The specific steps of Step D1 are as follows:

[0106] Step D11: The h(0 < h ≤ r) - th front slice of B s ;

[0107] Step D12: For h, with a step size of 1, in the range of values from 1 to r, loop and execute Step D11 to obtain the reduced - order model

[0108] Step D2: For s, with a step size of 1, in the range of values from 1 to n, sequentially execute Step D1 to obtain reduced - order models corresponding to all operating conditions one by one <​​​​​​​​​​​​​​​​​​​​

[0113] Step E3: For dictionary tensors Perform the TFQR decomposition algorithm to obtain the permutation matrix. Where ||P||0=N;

[0114] Step E4: Obtain the location indices of all N non-zero values ​​in P. These location indices are the optimized sensor layout locations.

[0115] Step F: Obtain the turbine disk surface temperature measurement value at the corresponding installation position by installing the sensor according to the optimized sensor layout or setting up virtual sensors through finite element simulation during test runs. Replace the value with the corresponding index position in P to obtain the measurement matrix. And ||Y||0=N;

[0116] Step G: Based on steps D and F Y, using the Tensor Compressed Sensing (TCS) algorithm, reconstructs the surface temperature field of the turbine disk. The detailed process of the TCS algorithm is as follows: Figure 4 The specific steps of step G are as follows:

[0117] Step G1: Initialize and set the TCS algorithm parameters, including the maximum number of iterations t and the algorithm tolerance ε. tol Regularization coefficient λ;

[0118] Step G2: Use the TCS algorithm to... The sparse mode vector is obtained by solving the problem.

[0119] Step G3: Calculate the reconstructed turbine disk surface temperature field structured matrix data Y hat , given

[0120] Although embodiments of the present invention have been described above in conjunction with the accompanying drawings, the present invention is not limited to the specific embodiments and application fields described above. The specific embodiments described above are merely illustrative and instructive, and not restrictive. Those skilled in the art can make many other forms based on the guidance of this specification and without departing from the scope of protection of the claims of the present invention, and all of these are within the scope of protection of the present invention.

Claims

1. A sparse sensor layout optimization method for turbine disk temperature field inversion, characterized in that, It includes the following steps: Step A: Simulate and calculate the trend of the thermal parameters of the turbine disk evolving with the operating time under different operating states, sample the temperature field information on the surface of the turbine disk at equal time intervals dt to obtain temperature field time snapshot data, and save it in a binary data serialization format containing discrete unstructured grid point coordinates and corresponding temperature numerical information; Step B: Process the temperature field information of each working condition into tensor data in the tensor data format; Step C: Sequentially perform high-order singular value decomposition on all the tensor data, and decompose to obtain the core tensor and modal matrices; Step D: Construct a reduced-order model of the dynamic system of the temperature field of the turbine disk evolving with time based on the obtained multiple core tensors and modal matrices, and obtain the dictionary tensor A for reconstructing the temperature field of the turbine disk; Step E: Perform tensor fiber-pivot orthogonal triangular decomposition on the dictionary tensor A to obtain the permutation matrix P and the optimized sensor layout; Step F: Obtain the turbine disk surface temperature measurement value at the corresponding installation position by installing the sensor according to the optimized sensor layout or setting up virtual sensors through finite element simulation during test runs. Replace the value with the corresponding index position in the permutation matrix P to obtain the measurement matrix. And ||Y||0=N; Step G: Reconstruct the temperature field on the surface of the turbine disk using the tensor compressive sensing algorithm based on the dictionary tensor A and the measurement matrix Y.

2. The sparse sensor layout optimization method for turbine disk temperature field inversion according to claim 1, characterized in that, Preferably, the trend of the thermal parameters of the turbine disk evolving with the operating time is simulated and calculated by means of finite element simulation or numerical simulation. Among them, different operating states include starting, off-design conditions, and rated conditions.

3. The sparse sensor layout optimization method for turbine disk temperature field inversion according to claim 1, characterized in that, The turbine disk is the pre-pressure turbine pump turbine disk of a liquid rocket engine.

4. The sparse sensor layout optimization method for turbine disk temperature field inversion according to claim 1, characterized in that, Step B includes: Step B1: Read in all the temperature field time snapshot data under the s-th (0 < s ≤ n) working condition in chronological order; Step B2: Perform bilinear interpolation and resampling on the discrete unstructured grid point coordinates of a single time snapshot data at a given sampling rate to obtain structured two-dimensional matrix data; Step B3: Stack the two-dimensional matrix data with time sequence as the third dimension, and finally convert all the temperature field time snapshot data under a single working condition into tensor data; Step B4: Perform steps B1 to B3 sequentially on the data for all operating conditions to obtain multiple tensor data converted from the time snapshot data of the temperature field under each operating condition. The number of tensor data is equal to the number of operating conditions, n.

5. The sparse sensor layout optimization method for turbine disk temperature field inversion according to claim 1, characterized in that, Step C includes: Step C1: Set the residual tolerance value ε, and at the same time, the truncation size r is automatically determined by the value of the residual tolerance value ε; Step C2: For tensor data The core tensor is obtained by performing higher-order singular value decomposition. and mode matrix and Where p, q, and r are the dimensions of the core tensor in the spacetime dimension; Step C3: Perform steps C1 to C2 sequentially on all operating condition data to obtain multiple core tensor vectors obtained from the decomposition of all tensor data. And the modal matrix vector [A1,A2,…,A] n ]、[B1,B2,…,B n ] and [C1,C2,…,C n ].

6. The sparse sensor layout optimization method for turbine disk temperature field inversion according to claim 1, characterized in that, Step D includes: Step D1: Utilizing the core tensor and mode matrix A s and B s Calculate the reduced-order model for the s-th working condition. Where r is the cutoff size of the reduced-order model; Step D2: For s, with a step size of 1, execute step D1 sequentially within the range of 1 to n to obtain the reduced-order model tensor corresponding to each of the operating conditions. Step D3: Apply the reduced-order model tensors to all operating conditions according to the actual working conditions of the engine, following the fiber orientation of the tensors, to the reduced-order models corresponding to all operating conditions. Stacking the tensors yields a dictionary tensor for reconstructing the turbine disk temperature field.

7. The sparse sensor layout optimization method for turbine disk temperature field inversion according to claim 6, characterized in that, Step D1 includes: Step D11: Reduced-order model The h-th (0 < h ≤ r) front slice of Step D12: For h, with a step size of 1, repeatedly execute step D11 within the range of 1 to r to obtain the reduced-order model.

8. The sparse sensor layout optimization method for turbine disk temperature field inversion according to claim 1, characterized in that, Step E includes: Step E1: Set the number of sensors selected as N, and set the truncation size of the TFQR decomposition algorithm as N; Step E2: Define the locations in the turbine disk where the temperature sensor installation is restricted and the locations where no physical entity corresponds to them in the structured matrix as rejection regions, and the remaining locations outside the rejection regions as acceptance regions. Step E3: For dictionary tensors Perform the TFQR decomposition algorithm to obtain the permutation matrix. Where ||P||0=N; Step E4: Obtain the position indices of all N non-zero values in the permutation matrix P, and the position indices are the positions of the optimized sensor layout.

9. The sparse sensor layout optimization method for turbine disk temperature field inversion according to claim 1, characterized in that, Step G includes: Step G1: Initialize and set the TCS algorithm parameters, including the maximum number of iterations t and the algorithm tolerance ε. tol Regularization coefficient λ; Step G2: Use the TCS algorithm to... The sparse mode vector is obtained by solving the problem. Step G3: Calculate the reconstructed turbine disk surface temperature field structured matrix data Y hat , given 10. The sparse sensor layout optimization method for turbine disk temperature field inversion according to claim 1, characterized in that, The turbine disk is installed in a liquid rocket engine turbine pump, an aero gas turbine engine, or a gas turbine.

Images