Visual traceability method and system for vibration energy of supporting stator mechanical structure

Abstract

The invention belongs to the technical field of engines, and discloses a support stator mechanical structure vibration energy visual traceability method, which is used for aero-engine support system vibration response transfer law research. Testing, analyzing and researching a vibration response signal transmission rule of the aero-engine supporting system; aero-engine supporting system vibration response signal prediction based on deep learning; a vibration signal traceability technology of a certain type of aero-engine supporting system is researched; an aero-engine vibration energy visualization technology is realized; testing, analyzing and researching a vibration response signal transmission rule of the aero-engine supporting system; aero-engine supporting system vibration response signal prediction based on deep learning; the invention relates to a vibration signal traceability technology research of an aero-engine supporting system. The method can be transplanted to fault diagnosis and health management of turbofan engines under research in China, provides technical support for design of new turbofan engines, can also provide theoretical basis for parameter optimization design of aero-engine supporting systems, and has great military and economic values.

Description

Technical Field

[0001] The present invention belongs to the technical field of engines, and in particular relates to a method and system for visually tracing the vibration energy of a supporting stator mechanical structure. Background Art

[0002] Rotating machinery is widely used in aerospace, automobile, shipbuilding and other related equipment. Since the working conditions of rotating machinery are very complex, and the vibration signal is easily distorted and submerged during the transmission process of complex structures and accessory systems, cross-structure system failure is one of the most easily overlooked and difficult to eliminate vibration failures. Aircraft engines are very complex thermal rotating machinery. Many parts work under high temperature, high pressure, high-speed rotation, strong vibration and complex and changeable environmental conditions. They are often subjected to high loads and thermal shocks. The working environment is extremely harsh, so it is easy to fail, and it has significant characteristics such as multiple failure modes and multi-mode composite failure. How to accurately describe the transmission law of vibration signals in aircraft engine systems and the contribution of each transmission path is a difficulty in dealing with certain primary fault diagnosis. Realizing rapid vibration tracing can effectively guide fault diagnosis and management work. In recent years, the study of vibration transmission mechanism has become one of the research hotspots in the field.

[0003] The vibration response transmission law refers to the transmission relationship between the response of the bearing seat and the response of the outer side of the casing. This relationship is a many-to-many relationship, that is, each bearing seat excitation will be transmitted to all places on the outside of the engine casing, and at the same time, any point on the outside of the casing will receive the response of each bearing seat excitation. The superposition of these responses determines the characteristics of the vibration response signal. Therefore, the transmission law here is not a simple point-to-point transmission, but the joint action of multiple sources of excitation at this point. How to use intuitive visualization methods to characterize the general laws of the above mechanical phenomena, and what are the characteristics of the above laws for aircraft engine products? It is urgent to summarize and refine scientific problems and conduct research.

[0004] In addition, there is currently no software in China that can achieve automatic power flow visualization. The main reason is that the calculation and visualization of vibration energy requires a lot of computing resources. The vibration state at each moment requires complex mathematical calculations, including dynamic analysis, frequency analysis, etc. Although technically feasible, most industrial simulation software currently does not support real-time vibration energy visualization due to cost, performance and practical needs. However, vibration energy visualization technology provides an effective tool for the health management of aircraft engines, which can warn of potential problems in advance, optimize performance, and ensure flight safety. The application of these technologies is crucial to the reliability and efficiency of the modern aviation industry. Summary of the invention

[0005] In view of the problems existing in the prior art, the present invention provides a method for visually tracing the vibration energy of a supporting stator mechanical structure.

[0006] The present invention is implemented as follows: a method for visually tracing the vibration energy of a supporting stator mechanical structure, the technology comprising:

[0007] S1: Research on the vibration response transmission law of aircraft engine support system;

[0008] S2: Experimental and analytical study on the transmission law of vibration response signal of aircraft engine support system;

[0009] S3: Prediction of vibration response signals of aircraft engine support system based on deep learning;

[0010] S4: Research on the vibration signal tracing technology of a certain type of aircraft engine support system;

[0011] S5: Realize the technology of visualizing the vibration energy of aircraft engines;

[0012] S6: Experimental and analytical study on the transmission law of vibration response signal of aircraft engine support system;

[0013] S7: Prediction of vibration response signals of aircraft engine support system based on deep learning;

[0014] S8: Research on vibration signal tracing technology of aircraft engine support system.

[0015] Further, Step 0: Program starts;

[0016] Step 1: Input and define material properties and finite element types through the material and element interface;

[0017] Step 2: Input the geometric parameters of the research object through the structural model command interface and create a geometric model;

[0018] Step 3: Input the material properties, unit types and geometric models generated in Steps 1 and 2 into the ANSYS pre-processing module ( / PREP) to create a finite element model of the research object;

[0019] Step 4: Input the load attributes and action position through the load condition command interface to apply the excitation load to the finite element model created in Step 3;

[0020] Step 5: Input the constraint type and action position through the boundary condition command interface to apply constraints to the finite element model created in Step 3;

[0021] Step 6: Enter the ANSYS solution module ( / SOLU). Enter the dynamic solution type and its related setting parameters through the solver setting command interface;

[0022] Step 7: Solve;

[0023] Step 8: Enter the ANSYS post-processing module ( / POST). Enter the analysis time, data extraction location, data type, data output storage location, etc. through the post-processing setting command interface;

[0024] Step 9: Extract the internal force, velocity and other parameters in the finite element model of the research object and save them to the file system;

[0025] Step 10 The following sub-steps are also executed in the self-compiled calculation program:

[0026] Sub step 1: Read the ANSYS output calculation result file and unit node relationship file from the file system, and organize them into a matrix and store them in the workspace;

[0027] Sub step 2: MATLAB reads the unit coordinate system and node coordinates of the finite element model

[0028] system and result coordinate system to convert physical space and computational space;

[0029] Sub step 3: Construct the calculation domain of the structural sound intensity vector field;

[0030] Sub step 4: According to the relevant calculation formula, solve the structural sound intensity vector of each unit in the calculation domain, and write it into the result matrix according to the unit node relationship;

[0031] Sub step 5: Output the structural sound intensity calculation results and save them to the file system;

[0032] Step 11: For the transient structural sound intensity vector field, the communication mechanism continues to control the calculation process, looping from Step 8 to Step 10, outputting the structural sound intensity calculation result files at different times and saving them to a file.

[0033] The software system is operated until all analysis moments are processed;

[0034] Step 12: The post-processing software TECPLOT reads the constructed computational domain;

[0035] Step 13: TECPLOT software reads the macro file, reads the calculated structural sound intensity calculation result file from the file system, and draws the structural sound intensity vector diagram;

[0036] Step 14: TECPLOT software reads the macro file and saves the structure-borne sound intensity vector diagram to the specified file directory;

[0037] Step 15: For the transient structural sound intensity vector field, repeat Step 12 to Step 14 until the structural sound intensity vector field at all analysis moments is drawn and saved;

[0038] Step 16: The program ends and exits S.

[0039] Further, the S1 specifically includes:

[0040] According to the simulation structure of the aircraft engine support system, an improved blind source separation algorithm is proposed based on information theory and maximum entropy algorithm to decompose the vibration response signals of each measuring point of the casing and determine the characterization relationship between the signals. Combined with the power flow frequency response function, the transmission law of the transient / steady-state total vibration energy and vibration energy components between the bearing seat and the casing is studied.

[0041] (1) Basic research on structure-borne sound intensity method / vibration wave theory;

[0042] (2) Compile the trace visualization code to basically realize the energy trace visualization.

[0043] Further, the S2 specifically includes:

[0044] Focusing on the vibration transmission problem of aircraft engine support system, the obtained casing vibration response signal is used to study the excitation-response mapping relationship, deeply explore the mutual relationship between the vibration signals of each measuring point of the casing under different excitations, and obtain the transmission law of vibration response.

[0045] (1) Establish a simplified finite element model and meshing of the support system;

[0046] (2) Select the response measurement point to obtain the amplitude-frequency characteristic curve of the external casing under applied excitation;

[0047] (3) Analyze the transient vibration signals of the response measuring points under single-point excitation (supports 1, 2, and 3) and multi-point excitation (1, 2 / 2, 3 / 1, 3 / 1, 2, and 3) at fixed frequencies (50 Hz / 100 Hz / 150 Hz).

[0048] Further, the S3 specifically includes:

[0049] Based on the dynamic model of the aircraft engine support system and the obtained casing vibration response signal, the LSTM model is used to model and predict the time series data, analyze the prediction ability of the casing measurement point vibration signal, and realize the prediction of the single structure and cross-structure vibration signal of the aircraft engine for the specific needs of aircraft engine health management;

[0050] (1) Obtaining vibration signals at three sampling points on the outer casing of the dual-rotor system;

[0051] (2) The CNNLSTM model is trained using vibration signals from three sampling points to accurately predict the time-domain vibration signal at the third point when only two vibration signals are input;

[0052] (3) Fast Fourier transform of time domain signals to accurately predict spectral characteristics.

[0053] Further, the S4 specifically includes:

[0054] Aiming at the problem of vibration tracing of stator casing, based on the established model and the study of the vibration response transmission law of the support system, the vibration signal tracing technology of a certain type of aircraft engine support system under single / multi-point excitation conditions was studied. In addition, the abnormal vibration diagnosis capability of the proposed single excitation vibration signal tracing technology was tested to verify the effectiveness and accuracy of the proposed tracing technology.

[0055] (1) Realize the establishment of the first generation vibration traceability response dictionary and signal traceability verification (based on the four-point acceleration response vibration signal of the outer casing under single-point excitation).

[0056] (2) Complete the exploration of the method for establishing the vibration traceability response dictionary for the Nth generation response.

[0057] Further, the S5 specifically includes:

[0058] The structural sound intensity method is extended into a matrix form and applied to the field of aeroengines. Through a computing system composed of finite element tools and self-compiled programs, the transient and steady-state total structural sound intensity field of the casing under the excitation of the unbalanced force of the high and low pressure rotors and the structural sound intensity field of longitudinal waves, shear waves, torsional waves and bending waves are solved and visualized, and the transmission characteristics and distribution laws of the transient and steady-state vibration energy carried by these four vibration waves on the casing are analyzed. In addition, the intrinsic physical relationship between structural sound intensity and structural vibration characteristics is derived and analyzed through the motion equation.

[0059] Further, the S6 specifically includes:

[0060] (1) Harmonic response analysis of aircraft engine support system model;

[0061] (2) Transient response analysis of aircraft engine support system model.

[0062] Furthermore, S7 specifically includes: studying how to effectively use the LSTM model to predict aircraft engine sensor data containing missing signals, and enhancing the expression and application of these predictions through vibration energy visualization technology.

[0063] Further, the S8 specifically includes:

[0064] Aiming at the energy dispersion caused by discontinuity of supporting structure, starting from the geometric transformation of signal, the signal feature enhancement method based on geometric peak transformation is extended, and a new vibration signal separation and recognition method is proposed; according to the similarities and differences of excitation signals in different parts, the rank and sparsity characteristics of vibration signals are explored, and a new separation and recognition framework is proposed, and finally the mapping relationship between the excitation source and the response of the casing measurement point is obtained;

[0065] For a certain type of engine support system, the transient / steady-state total vibration energy and energy components of the support system are obtained by the Lagrangian method using the excitation signal and the vibration response model of the casing. Combined with the power flow frequency response function, the energy transfer path between multiple bearing seats and the casing is studied. Subsequently, the proposed synchronous oscillation source tracing method is further expanded, taking into full consideration the scenario design for models and actual engineering. For scenarios where there are multiple inducing conditions for subsynchronous oscillations in actual systems, a support system source tracing model based on a multi-source domain adaptive algorithm is established, and a subsynchronous oscillation source tracing method that is simultaneously effective for multiple inducing conditions is proposed.

[0066] Finally, based on the open-loop mode resonance theory, an excitation source inversion method in the equivalent system is proposed. Combined with the mapping relationship data set between the excitation signal and the response signal, a synchronous oscillation tracing model based on a multi-source domain adaptive network is established, and a casing vibration tracing method under single / multi-point excitation conditions is developed.

[0067] Another object of the present invention is to provide a computer device, which includes a memory and a processor, wherein the memory stores a computer program, and when the computer program is executed by the processor, the processor executes the steps of the method for visualizing the vibration energy of the supporting stator mechanical structure.

[0068] Another object of the present invention is to provide a system for visually tracing the vibration energy of an aircraft engine support system, the system comprising:

[0069] A vibration signal decomposition module is used to decompose the vibration response signals of each measuring point of the casing and determine the characterization relationship between the signals based on the simulated structure of the aircraft engine support system and the blind source separation algorithm developed by information theory and maximum entropy algorithm;

[0070] A vibration energy transfer analysis module, which is used to study the transient and steady-state total vibration energy and the transmission law of vibration energy components between the bearing seat and the casing in combination with the power flow frequency response function;

[0071] A trace visualization module, used to compile trace visualization code and realize the visualization expression of energy traces;

[0072] A data storage module is used to store the processed vibration response signal and energy trace data.

[0073] Another object of the present invention is to provide a system for studying vibration transmission of an aircraft engine support system, the system comprising:

[0074] A finite element model building module, which is used to build and mesh a simplified finite element model of an aircraft engine support system to solve the vibration transmission problem of the support system;

[0075] An excitation response analysis module is used to use the obtained casing vibration response signal to deeply analyze the excitation response mapping relationship and reveal the transmission law of vibration response;

[0076] An experimental simulation module for analyzing transient vibration signals of response measurement points under single-point excitation and multi-point excitation at a fixed frequency;

[0077] A data output module is used to output the analysis results of vibration transmission laws.

[0078] Another object of the present invention is to provide a system for predicting vibration signals and analyzing spectrum characteristics of an aircraft engine support system, the system comprising:

[0079] A time series modeling module, which is used to perform time series modeling and prediction of casing vibration response signals using the LSTM model based on the dynamic model of the aircraft engine support system;

[0080] A vibration signal prediction module is used to accurately predict the time domain vibration signal by acquiring and training the vibration signals of three sampling points of the outer casing of the dual-rotor system;

[0081] A spectrum feature analysis module, used to perform fast Fourier transform on the predicted time domain signal to obtain accurate prediction of the spectrum features;

[0082] A health management support module for applying prediction and analysis results to the health management of aircraft engines.

[0083] Another object of the present invention is to provide a system for tracing the vibration source of an aircraft engine support system, the system comprising:

[0084] A signal feature enhancement module is used to address the energy dispersion problem caused by the discontinuity of the supporting structure. A signal feature enhancement method based on geometric peak transformation is proposed to improve the accuracy of vibration signal separation and recognition;

[0085] A subsynchronous oscillation source tracing module is used to establish a support system source tracing model based on a multi-source domain adaptive algorithm for subsynchronous oscillation problems under multi-source induced conditions, and propose an effective subsynchronous oscillation source tracing method;

[0086] An excitation source inversion module is used to perform excitation source inversion in an equivalent system based on the open-loop mode resonance theory, and to build a synchronous oscillation tracing model of a multi-source domain adaptive network by combining the mapping relationship dataset between the excitation signal and the response signal;

[0087] A vibration tracing implementation module is used to realize casing vibration tracing under single / multi-point excitation conditions.

[0088] In combination with the above technical solutions and the technical problems solved, the advantages and positive effects of the technical solutions to be protected by the present invention are as follows:

[0089] First, the structural sound intensity method is a method that calculates the sound intensity by measuring the velocity and pressure on the surface of the structure, thereby analyzing the transfer path and distribution of energy in the structure. In the study of aircraft engine vibration energy, this method is used to describe the flow of vibration energy in detail. By combining force and velocity measurements, the structural sound intensity method can not only help engineers identify the source and dissipation location of energy, but also clearly show the flow path of energy in various engine components through the visualization of vector fields.

[0090] The application of this technology enables engineers to perform vibration analysis and control more accurately. For example, during the engine design and testing phase, the structure-borne sound intensity method can be used to evaluate the effects of different design schemes on vibration control, optimize structural design, reduce unnecessary energy loss, and extend the service life of the engine. In addition, this method can also help maintenance personnel identify fault locations during routine maintenance and perform targeted maintenance and repairs.

[0091] With the development of artificial intelligence technology, deep learning models, especially long short-term memory networks (LSTM) and convolutional LSTM networks (CNNLSTM), have been widely used in the prediction and analysis of time series data. In the monitoring of aircraft engine vibration energy, these models are used to model and predict vibration signals, realizing real-time monitoring and early warning of engine vibration status.

[0092] By training the deep learning model, vibration patterns can be learned from historical vibration data and future vibration trends can be predicted. This prediction not only helps to detect potential abnormal vibrations, but also allows timely intervention before engine failure occurs to avoid serious mechanical failures. In addition, the application of this technology can also monitor the vibration status in real time during engine testing and operation, optimize the engine's operating parameters, and improve the engine's operating efficiency and safety.

[0093] Vibration signal tracing technology refers to the technology of tracing the source and propagation path of vibration through in-depth analysis of vibration signals. This research direction is particularly important in the visualization of aviation engine vibration energy. By establishing a vibration transmission model, it is possible to analyze how vibration propagates from the source to other parts of the engine, which is of great significance for diagnosing faults and optimizing vibration control strategies.

[0094] The application of this technology can not only help engineers more accurately identify the specific source of vibration, but also analyze the specific mechanism of vibration propagation, providing a scientific basis for engine design and maintenance. For example, in the engine design stage, by simulating the propagation path of vibration, the structural design can be predicted and optimized to reduce unnecessary vibration. In the maintenance stage, this technology can help maintenance personnel accurately find the source of the fault and improve maintenance efficiency and quality.

[0095] The development of aviation engine vibration energy visualization technology, especially the application of structural sound intensity method, the integration of deep learning models and the research of vibration signal tracing technology, has greatly improved the monitoring and analysis capabilities of aviation engine vibration status. The combined application of these technologies can not only improve the operating efficiency and safety of the engine, but also play an important role in the engine design, testing and maintenance process, and is an important support for aviation engine health management and fault diagnosis.

[0096] Second, the present invention applies the rotor mechanical vibration energy visualization technology to practical engineering problems, realizes the implementation of vibration transmission analysis and tracing in the aircraft engine support system, expands the research ideas of aircraft engine vibration and fault diagnosis, and helps related companies to intuitively understand the law of rotor mechanical vibration energy distribution. It is also urgent to assist in the research on the vibration transmission law and tracing of the aircraft engine support system under multi-source excitation conditions and develop the tracing identification technology of the vibration signal of a certain type of engine casing under multi-source excitation. The research results will have a high guiding significance for the in-depth understanding of the vibration transmission and nonlinear dynamic behavior of the aircraft engine support system, and have important application value for improving the structural integrity and reliability of a certain turbofan engine. It can be used for batch maintenance of a certain engine and greatly reduce the maintenance cost. This technical method can be transplanted to the fault diagnosis and health management of turbofan engines under development in my country, and provide technical support for the design of new turbofan engines. At the same time, it can also provide a theoretical basis for the parameter optimization design of the aircraft engine support system, which has great military and economic value.

[0097] Third, as auxiliary evidence of the inventiveness of the claims of the present invention, it is also reflected in the following important aspects:

[0098] (1) The expected benefits and commercial value of the technical solution of the present invention after transformation are:

[0099] ①Extend engine life

[0100] By monitoring and analyzing vibration energy, the health of the stator structure can be assessed more accurately, maintenance and servicing strategies can be optimized, and unnecessary shutdowns and repairs can be reduced, thereby extending the service life of the engine and reducing operating costs.

[0101] ② Improve operational efficiency

[0102] Vibration energy visualization tracing technology can help airlines and engine manufacturers perform fault diagnosis and maintenance more efficiently, improve engine operating efficiency, reduce downtime, and increase aircraft utilization and flight punctuality.

[0103] ③ Reduce maintenance costs

[0104] By early detection and prediction of failures, the occurrence of sudden failures can be reduced, and the cost of emergency repairs and replacement of parts can be reduced. At the same time, optimized maintenance plans can also reduce the waste of manpower and material resources and improve the efficiency of maintenance work.

[0105] ④Technical barriers

[0106] The present invention has a high technical barrier and can form a significant competitive advantage. Mastering this technology can take a leading position in the market, fill the market gap, and gain more market share.

[0107] ⑤Additional services

[0108] In addition to the technology itself, we can also provide related consulting, training, maintenance and other additional services to form a complete solution and increase revenue sources.

[0109] (2) The technical solution of the present invention fills the technical gap in the industry at home and abroad:

[0110] ①There is currently no commercial simulation analysis software in China that can perform visual analysis of vibration energy.

[0111] ② International commercial simulation analysis software has made slow progress in the development of this technology BRIEF DESCRIPTION OF THE DRAWINGS

[0112] Figure 1 This is a technical route for visualizing the vibration energy of an aircraft engine provided by an embodiment of the present invention;

[0113] Figure 2 It is a diagram of various forces and moments of a shell unit provided by an embodiment of the present invention;

[0114] Figure 3 is the bending wave structure sound intensity field provided by the embodiment of the present invention;

[0115] Figure 4 The shear wave structure sound intensity field provided by the embodiment of the present invention is

[0116] Figure 5 The torsional wave structure sound intensity field provided by the embodiment of the present invention is

[0117] Figure 6 The longitudinal wave structure sound intensity field provided by the embodiment of the present invention is

[0118] Figure 7 The three-dimensional model of the support system provided by the embodiment of the present invention is

[0119] Figure 8 The outer casing measuring point selection position provided by the embodiment of the present invention is

[0120] Fig. 9 is the first-order mode provided by an embodiment of the present invention;

[0121] Fig.10 is the second-order mode provided by the embodiment of the present invention;

[0122] Fig.11 is the excitation force application position provided by the embodiment of the present invention;

[0123] Fig.12 is a response curve of applying excitation to bearing seat No. 1 provided in an embodiment of the present invention;

[0124] Fig.13 is a response curve of applying excitation to bearing seat No. 2 provided in an embodiment of the present invention;

[0125] Fig.14 is a response curve of applying excitation to bearing seat No. 3 provided in an embodiment of the present invention;

[0126] Fig.15 is a time-frequency diagram of applying excitation to bearing seat No. 1 provided in an embodiment of the present invention;

[0127] Fig.16 is a time-frequency diagram of applying excitation to bearing seat No. 2 provided in an embodiment of the present invention;

[0128] Fig.17 is a time-frequency diagram of applying excitation to bearing seat No. 3 provided by an embodiment of the present invention;

[0129] Fig.18 It is a time-frequency diagram of applying excitation to bearing seats No. 1 and No. 2 provided in an embodiment of the present invention;

[0130] Fig.19 It is a time-frequency diagram of applying excitation to bearing seats No. 1 and No. 3 provided by an embodiment of the present invention;

[0131] Fig. 20 It is a time-frequency diagram of applying excitation to bearing seats No. 2 and No. 3 provided by an embodiment of the present invention;

[0132] Fig.21 It is a time-frequency diagram of applying excitation to bearing seats No. 1, 2, and 3 provided by an embodiment of the present invention;

[0133] Fig. 22 It is a vibration energy streamline provided by an embodiment of the present invention;

[0134] Fig.23 is the RNN structure provided by the embodiment of the present invention;

[0135] Fig.24 It is the LSTM structure provided by the embodiment of the present invention;

[0136] Fig.25 is a comparison chart of data prediction and original data provided by an embodiment of the present invention;

[0137] Fig.26 is a frequency domain diagram of original data provided by an embodiment of the present invention;

[0138] Fig. 27 is a frequency domain diagram of predicted data provided by an embodiment of the present invention;

[0139] Fig.28 is a signal tracing model provided by an embodiment of the present invention;

[0140] Fig.29 is the amplitude response ratio of each point under the same excitation provided by the embodiment of the present invention;

[0141] Fig.30 It is the amplitude response ratio of each point under different excitations provided by the embodiment of the present invention. DETAILED DESCRIPTION

[0142] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0143] Aiming at the shortcomings of existing aircraft engine vibration monitoring technology, this paper proposes a vibration energy tracing method based on deep learning and visualization technology. The existing technology has problems such as insufficient accuracy, poor real-time performance and weak visualization ability in the study of vibration signal transmission law and vibration signal tracing, which makes it difficult to meet the needs of modern aircraft engine maintenance and fault diagnosis.

[0144] First, the present invention establishes a high-precision vibration model by deeply studying the vibration response transmission law of the aircraft engine support system, combining theoretical analysis and numerical simulation. Compared with the traditional simplified model, this method can more accurately describe the vibration transmission law between different components and nodes, and improve the accuracy and reliability of vibration monitoring.

[0145] Secondly, the present invention introduces deep learning technology to predict the vibration response signal of the aircraft engine support system. By constructing convolutional neural network (CNN) and recurrent neural network (RNN) models, and using a large amount of experimental data for training and optimization, high-precision prediction of vibration signals is achieved. This method overcomes the problems of weak data processing capability and low prediction accuracy in traditional vibration signal prediction methods, and improves the intelligence level and prediction ability of the system.

[0146] Thirdly, the present invention proposes a vibration signal tracing technology, which combines the vibration transmission law model and deep learning prediction results, and uses signal reverse analysis technology to accurately locate the source and propagation path of the vibration signal. This method solves the problem of insufficient positioning accuracy and difficulty in accurately identifying the vibration source in traditional tracing technology, and provides strong technical support for fault diagnosis.

[0147] Finally, the present invention realizes the visualization of the vibration energy of aircraft engines. By using advanced visualization technology, the transmission and distribution of vibration energy can be intuitively displayed in the form of three-dimensional heat maps, dynamic images, etc. Compared with traditional text and two-dimensional chart display methods, this visualization technology greatly enhances the intuitiveness and effectiveness of information transmission, helps engineers understand and analyze vibration conditions more quickly, and improves the efficiency of fault diagnosis and maintenance.

[0148] In summary, the present invention significantly improves the ability of aircraft engine vibration monitoring and fault diagnosis through a number of technological innovations, solves a number of key problems in the prior art, and has high application value and broad prospects.

[0149] The working principle of the method for visualizing the source of vibration energy of a supporting stator mechanical structure provided by the embodiment of the present invention can be summarized in detail as follows:

[0150] This method aims to achieve visual tracing of the vibration energy of the aircraft engine support system through a series of steps. This includes the study of the vibration response transmission law, the test and analysis of the vibration response signal, the signal prediction based on deep learning, the research on the tracing technology of the vibration signal, and finally the visualization of the vibration energy.

[0151] 1. Study on the vibration response transmission law of aircraft engine support system (S1)

[0152] Based on the improved blind source separation algorithm: According to the simulation structure of the aircraft engine support system, the blind source separation algorithm based on information theory and maximum entropy algorithm is used to decompose the vibration response signals of each measuring point of the casing to determine the characterization relationship between the signals.

[0153] Power flow frequency response function analysis: Combined with the power flow frequency response function, the transmission law of transient / steady-state total vibration energy and vibration energy components between the bearing seat and the casing is studied.

[0154] 2. Experimental and analytical study on the transmission law of vibration response signal (S2 / S6)

[0155] Through experimental means, the vibration response signals of the aircraft engine support system are collected and analyzed to verify and supplement the conclusions of theoretical research.

[0156] 3. Vibration response signal prediction based on deep learning (S3 / S7)

[0157] Deep learning techniques, such as neural network models, are used to predict the vibration response signals of aircraft engine support systems, providing data support for fault diagnosis and prediction.

[0158] 4. Research on vibration signal tracing technology (S4 / S8)

[0159] Study the source and propagation path of vibration signals, identify key features in vibration signals through signal processing techniques such as spectrum analysis and modal analysis, and provide a basis for the visualization of vibration energy.

[0160] 5. Realize the visualization technology of aviation engine vibration energy (S5)

[0161] Structural sound intensity method / basic research on vibration wave theory: Study the propagation characteristics of structural sound intensity and vibration waves in aircraft engine support systems to provide theoretical support for the visualization of vibration energy.

[0162] Compile trace visualization code: Develop or compile specific visualization code to plot and display vibration energy traces in aircraft engine support systems to achieve energy visualization.

[0163] 6.ANSYS finite element analysis and data processing (Step 110)

[0164] The finite element model of the aircraft engine support system was established using ANSYS software, and excitation loads and constraints were applied.

[0165] The dynamics solution is performed through the solver to extract parameters such as internal force and velocity in the finite element model.

[0166] The extracted data are processed and analyzed using data processing tools such as MATLAB, and parameters such as structure-borne sound intensity are calculated.

[0167] 7. Post-processing and visualization (Step 1115)

[0168] Use post-processing software such as TECPLOT to read the calculated parameters such as structural sound intensity and draw visual results such as structural sound intensity vector diagrams.

[0169] Save the visualization results to the specified file directory for subsequent analysis and application.

[0170] 8. End of the program (Step 16)

[0171] After all calculations and analyses are completed, the program ends and exits.

[0172] Through the above working principle, this method can realize the visual tracing of the vibration energy of the aircraft engine support system, providing strong support for the design, fault diagnosis and performance optimization of the aircraft engine.

[0173] An embodiment of the present invention provides a method for visually tracing the vibration energy of a supporting stator mechanical structure, the technology comprising:

[0174] S1: Research on the vibration response transmission law of aircraft engine support system;

[0175] S2: Experimental and analytical study on the transmission law of vibration response signal of aircraft engine support system;

[0176] S3: Prediction of vibration response signals of aircraft engine support system based on deep learning;

[0177] S4: Research on the vibration signal tracing technology of a certain type of aircraft engine support system;

[0178] S5: Realize the technology of visualizing the vibration energy of aircraft engines;

[0179] S6: Experimental and analytical study on the transmission law of vibration response signal of aircraft engine support system;

[0180] S7: Prediction of vibration response signals of aircraft engine support system based on deep learning;

[0181] S8: Research on vibration signal tracing technology of aircraft engine support system.

[0182] Working principle of the method for visualizing the vibration energy of supporting stator mechanical structures.

[0183] 1. Overview

[0184] The present invention relates to a visual tracing method for the vibration energy of an aero-engine, which is mainly achieved through the study of vibration response transmission laws, deep learning prediction, vibration signal tracing technology and vibration energy visualization technology.

[0185] 2. Specific steps and working principles

[0186] S1: Study on the vibration response transmission law of aircraft engine support system

[0187] Objective: To study the vibration response transmission law of aircraft engine support system to understand how vibration propagates in the support system.

[0188] Method: Through theoretical analysis and numerical simulation, the vibration model of the support system is established, and the transmission law of vibration response between different components and nodes is analyzed.

[0189] S2: Experimental and analytical study on the transmission law of vibration response signal of aircraft engine support system

[0190] Objective: To study the transmission rules of vibration signals in the support system through experimental verification and analysis.

[0191] Methods: Sensors were arranged on the actual aircraft engine support system to measure the vibration signals. The accuracy of the vibration transmission law model was verified through data analysis and spectrum analysis.

[0192] S3: Prediction of vibration response signals of aircraft engine support system based on deep learning

[0193] Objective: Use deep learning algorithms to predict the vibration response signals of a support system.

[0194] Method: Collect a large amount of vibration signal data, build a deep learning model (such as convolutional neural network, recursive neural network, etc.), train and optimize it, and realize accurate prediction of vibration response signals.

[0195] S4: Research on the Vibration Signal Tracing Technology of a Certain Aircraft Engine Support System

[0196] Objective: To study how to trace the vibration signal of a certain type of aircraft engine support system and find out the vibration source and propagation path.

[0197] Method: Combining the vibration transmission law model and deep learning prediction results, the source and propagation path of the vibration signal are traced through signal reverse analysis technology.

[0198] S5: Visualization of aircraft engine vibration energy

[0199] Objective: To visualize the transfer and distribution of vibration energy.

[0200] Method: Based on the vibration transmission law and traceability results, visualization technology (such as heat maps, three-dimensional images, etc.) is used to display the distribution and transmission path of vibration energy to help engineers intuitively understand the vibration situation.

[0201] S6: Experimental and analytical study on the transmission law of vibration response signal of aircraft engine support system

[0202] Objective: To deepen the understanding of the laws of vibration signal transmission through further experiments and analysis.

[0203] Methods: Repeat and expand previous experiments, collect more data for analysis, and verify and improve the vibration transmission law model.

[0204] S7: Prediction of vibration response signals of aircraft engine support system based on deep learning

[0205] Objective: To further optimize and apply deep learning models to improve the accuracy and robustness of vibration signal prediction.

[0206] Methods: New experimental data and optimized deep learning algorithms were used to improve the model's training and prediction capabilities to ensure prediction accuracy under different operating conditions.

[0207] S8: Research on Vibration Signal Tracing Technology of Aircraft Engine Support System

[0208] Objective: To further study and apply vibration signal tracing technology to improve the accuracy and applicability of tracing.

[0209] Method: Combining new experimental data with optimized tracing algorithm, the positioning accuracy of vibration signal source and propagation path is improved to ensure the application effect of tracing technology in different types of engines.

[0210] This invention achieves accurate prediction and intuitive display of the vibration energy of the aircraft engine support system through the comprehensive application of theoretical research, experimental verification, deep learning prediction, signal tracing and visualization technology. Through in-depth research on the laws of vibration signal transmission, combined with advanced deep learning algorithms and visualization technology, it solves the problem that traditional methods are difficult to accurately predict and trace vibration energy, and provides an effective technical means for vibration monitoring and fault diagnosis of aircraft engines.

[0211] In order to study the law of vibration energy transfer, the structural sound intensity method can be applied to the transient / steady-state vibration analysis of the entire aircraft engine. The structural sound intensity method combines the force and velocity at any point in the elastic structure to characterize the energy flow in the vibrating structure. Figure 1 As shown:

[0212] This method was proposed by scholars such as Noiseux and expanded into a vector acoustic field method of energy flow in the field of structural acoustic transmission. Since the structural acoustic intensity field is a vector field, the magnitude and direction of any point in the field can predict and quantify the magnitude and direction of the vibration energy transfer at that point. Therefore, through the visualization of the vector field, the structural acoustic intensity method can be used to describe the main transmission path and distribution characteristics of the vibration energy in the structure, and its divergence field can be used to indicate the source and sink of the vibration energy in the structure for direct analysis and judgment of the main source and main dissipation position of the vibration energy in the structure, and provide vibration reduction measures from the three levels of vibration source, propagation path and receptor. Based on this, this chapter expands the structural acoustic intensity method into a matrix form and applies it to the field of aeroengines. Through a computing system composed of finite element tools and self-compiled programs, the total structural acoustic intensity field of the casing in transient and steady states under the excitation of the unbalanced force of the high and low pressure rotors, as well as the structural acoustic intensity field of the longitudinal wave, shear wave, torsional wave and bending wave are solved and visualized, and the transmission characteristics and distribution laws of the transient and steady-state vibration energy carried by these four vibration waves on the casing are analyzed. In addition, the intrinsic physical relationship between structure-borne sound intensity and structural vibration characteristics was derived and analyzed through the equation of motion.

[0213] (1) Structural sound intensity method theory

[0214] The propagation of vibration in the structure is essentially the transfer of vibration energy. The vibration load generated by the unbalanced force of the rotor is essentially transmitted to the casing through the support plate in the form of vibration energy, and is transmitted on the casing and coupled and influenced by the vibration of other components.

[0215] As a power flow analysis method, the structural sound intensity method considers the magnitude and phase relationship of force and velocity response to characterize the magnitude and direction of vibration energy transfer. The general tensor expression of transient structural sound intensity is

[0216] I i (t)=-∑ j σ ij (t)v j (t)i,j=1,2,3 (1)

[0217] Where: ij (t) and v j (t) respectively represent the stress tensor component and velocity vector component of the structure at time t. It can be seen from formula (1) that the structural sound intensity method characterizes the vibration energy flow of the structure per unit time and per unit cross-sectional area. Based on formula (1), the stress tensor component and velocity vector component can be extracted through a self-compiled program, the vibration energy change can be visualized, and the main transmission path of the vibration energy in the structure and the vibration source, vibration sink and other information can be quantitatively analyzed. Combining the finite element method and the structural sound intensity method, the abnormal vibration and noise of the engine components can be analyzed to help identify the fault location and type, which is of great significance for maintenance and fault prevention.

[0218] Vibration propagates in the structure in the form of various waves, mainly including four types: longitudinal waves, shear waves, torsional waves, and bending waves. Vibration energy is carried by these vibration waves and transmitted in the structure. On any vibration power flow streamline, the tangent direction of a point is consistent with the power flow vector direction of that point. Since the wall thickness of the aircraft engine casing structure is much smaller than its radius, shell elements (L type) can be used for simulation. For details on the forces, moments, and displacements of typical shell elements, see Fig. 27 shown.

[0219] Respectively represent the translational displacement and rotational displacement of the shell element. Substituting the above force, moment and displacement into equation (1), and through Fourier transform, the frequency domain expression of the vibration power flow per unit width of the shell element can be obtained. Since the shell element is a 2D element, its vibration power flow can be decomposed into two components in the x and y directions perpendicular to the normal z in the element coordinate system: and As shown in formula (2).

[0220]

[0221] The total vibration power flow is I tol It can be expressed as

[0222]

[0223] Where N11 and N22 are in-plane axial forces, N12 and N21 are in-plane shear forces (N12 = N21), Q13 and Q23 are transverse shear forces, M11 and M22 are bending moments, M12 and M21 are torques (M12 = M21), ux, uy and uz are translational displacements, θx and θy are unit rotational displacements. Im represents the imaginary part of the complex number.

[0224] To characterize the ratio of the vibration energy flux transmitted through a certain section in a certain direction to the total vibration energy passing through the section. Definition of vibration power flux ratio

[0225]

[0226] In the formula, is the vibration energy vector transmitted through a certain section to a certain direction, n is the statistical number of vibration energies transmitted in the specific direction on the section, and A is the area of ​​the section. Among them, the numerator term represents the vibration energy flux through the section in a certain transmission direction, and the denominator term represents the total vibration energy transmitted through the section in this direction.

[0227] (2) Self-compiled code and environment deployment

[0228] This section provides a method for post-processing the model harmonic response analysis results using self-compiled APDL command flow code under the Windows system, importing the data results into MATLAB software for analysis, converting the file format through the Windows system, configuring environment variables, and generating energy streamlines through Tecplot. The code framework is as follows.

[0229] Step 1: Extract stress and displacement results and calculate vibration energy

[0230] / POST1 ! Enter post-processing mode

[0231] ! Extract stress results

[0232] *GET,NODE_COUNT,NODE,0,COUNT! Get the total number of nodes

[0233] *GET,ELEM_COUNT,ELEM,0,COUNT! Get the total number of units

[0234] ! Create an array to store stress results

[0235] DIM,STRESS_ARRAY,ARRAY,NODE_COUNT,6

[0236] ! Loop through all nodes and extract stress results

[0237] DO I=1,NODE_COUNT

[0238] *GET,NODE_I,X,NODE,I,LOC! Get node coordinates

[0239] *GET,NODE_I,SELN,NODE,I,S,X! Get node stress results

[0240] STRESS_ARRAY(I,1)=NODE_I(2)! X-direction stress

[0241] STRESS_ARRAY(I,2)=NODE_I(3)! Stress in the Y direction

[0242] STRESS_ARRAY(I,3)=NODE_I(4)! Stress in Z direction

[0243] STRESS_ARRAY(I,4)=NODE_I(5)! XY shear stress

[0244] STRESS_ARRAY(I,5)=NODE_I(6)! XZ shear stress

[0245] STRESS_ARRAY(I,6)=NODE_I(7)! YZ shear stress

[0246] ENDDO

[0247] ! Extract displacement results

[0248] *GET, U_COUNT, NODE, 0, COUNT, U, X! Get displacement results

[0249] ! Create an array to store the displacement results

[0250] DIM,DISPLACEMENT_ARRAY,ARRAY,NODE_COUNT,3

[0251] !Loop through all nodes and extract displacement results

[0252] DO I=1,NODE_COUNT

[0253] *GET,U_I,X,NODE,I,U,X! Get the displacement of the node in the X direction

[0254] *GET,U_I,Y,NODE,I,U,Y! Get the Y-direction displacement of the node

[0255] *GET,U_I,Z,NODE,I,U,Z! Get the displacement of the node in the Z direction

[0256] DISPLACEMENT_ARRAY(I,1)=U_I(2)! X-direction displacement

[0257] DISPLACEMENT_ARRAY(I,2)=U_I(3)! Displacement in the Y direction

[0258] DISPLACEMENT_ARRAY(I,3)=U_I(4)! Displacement in Z direction

[0259] ENDDO

[0260] Calculate vibration energy

[0261] DIM,VIBRATION_ENERGY,ARRAY,NODE_COUNT

[0262] DO I=1,NODE_COUNT

[0263] VIBRATION_ENERGY(I)=(STRESS_ARRAY(I,1)2+STRESS_ARRAY(I,2)2+STRESS_ARRAY(I,3)2) / 2+(DISPLACEMENT_ARRAY(I,1)2+DISPLACEMENT_ARRAY(I,2)2+DISPLACEMENT_ARRAY(I,3)2) / 2

[0264] ENDDO

[0265] Step 2: Visualize the vibration energy distribution

[0266] / POST26! Enter vibration energy post-processing mode

[0267] ! Visualize the distribution of vibration energy on the model

[0268] *VWRITE,VIBRATION_ENERGY! Write vibration energy to variable

[0269] PLNSOL,S! Draw contour lines on a plane

[0270] Step 3: Expand the 3D solid model into a plane

[0271] / PREP7! Enter preparation mode

[0272] ! Expand the 3D solid model to a plane

[0273] *GET,MIN_X,ELEM,1,AREA,XMIN! Get the minimum X coordinate of the model

[0274] *GET,MAX_X,ELEM,1,AREA,XMAX! Get the maximum X coordinate of the model

[0275] *GET,MIN_Y,ELEM,1,AREA,YMIN! Get the minimum Y coordinate of the model

[0276] *GET,MAX_Y,ELEM,1,AREA,YMAX! Get the maximum Y coordinate of the model

[0277] ! Create plane

[0278] LSEL,S,ELEM! Select all units

[0279] *GET,ELEM_COUNT,ELIST! Get the number of units

[0280] ETABLE,LST,AREA,ALL! Expand all elements to the plane

[0281] PLNSOL,ALL! Draw contour lines on a plane

[0282] Step 4: Mark the trajectory of vibration energy flow

[0283] / POST26! Enter vibration energy post-processing mode

[0284] ! Mark the trajectory of vibration energy flow on the plane and use arrows to indicate direction and magnitude

[0285] PLNSOL,VIBRATION_ENERGY! Draw contour maps on a plane

[0286] PLVECT,VIBRATION_ENERGY! Draw vector graphics on a plane

[0287] / DEVICE,PNG,SAVE,DESKTOP\flow.png! Save the plane diagram with energy flow trajectory marked to the computer desktop

[0288] To ensure the accuracy of the programmed calculation results, this chapter establishes a communication mechanism between MATLAB and ANSYSAPDL software. The model vibration energy simulation results are extracted from ANSYS and imported into MATLAB for visualization of the vibration energy transfer path.

[0289] APDL calls MATLAB command flow:

[0290]

[0291] MATLAB calls APDL scripting language:

[0292] %creat flag for ANSYS call MATLAB%

[0293] FLAG = zeros(1, 1);

[0294] FLAG(1,1)=0

[0295] Fid2=fopen('FLAG.TXT','w');

[0296] While fid2 == 1

[0297] Fid2=fopen('FLAG.TXT','w');

[0298] End

[0299] Fprintf(fid2,'%f\t',FLAG(1,1));

[0300] St2 = fclose(fid2);

[0301] While st2==1

[0302] St2 = fclose(fid2);

[0303] End

[0304] %close matlab&call for ANSYS%

[0305] Exit

[0306] (3) Calculation results and verification

[0307] This section uses the above calculation system to solve the transient and steady-state total structural sound intensity field of the casing under the excitation of the unbalanced force of the high and low pressure rotors, as well as the single-point excitation of the structural sound intensity field of longitudinal waves, shear waves, torsional waves and bending waves, and analyzes the transmission characteristics and distribution laws of the transient and steady-state vibration energy carried by these four vibration waves on the casing. In addition, the inherent physical relationship between the structural sound intensity and the structural vibration characteristics is derived and analyzed through the motion equation.

[0308] In order to more directly and clearly display the structural sound intensity field of different vibration waves on the casing, the self-compiler program unfolds the casing into a two-dimensional plane along the circumference, and the result coordinate system and the corresponding result data output vector are also changed accordingly, such as Figure 3 shown.

[0309] Expand the casing model circumferentially. Figure 3 Observe in the direction shown, refer to the relative positions of the components of the actual model, and the straight line corresponding to the end x=200 is the flange edge. Apply a sinusoidal excitation of 20 Hz and 60 N to (100,50) to obtain the vibration wave structural sound intensity field along the circumferential direction. Apply a shear wave with a direction of 45 degrees and a scale of 2 to each part model, and intercept the key frame structural sound intensity field at 20 seconds as shown below: Figure 4 A torsional wave with an angular velocity of 2 rad / s and a radius of 10 mm is applied, and the key frame structural sound intensity field at 20 s is intercepted as shown in Figure 5 As shown. A longitudinal wave with a wavelength of 0.5 mm and a speed of 1 mm / s is applied, and the key frame structural sound intensity field at 20 seconds is intercepted as follows Figure 6 shown.

[0310] Most of the vibration energy on the casing is carried and transmitted by longitudinal waves, bending waves and shear waves, while the vibration energy transmitted by torsional waves accounts for the smallest proportion. The casing flange has an obstructive effect on the transmission of longitudinal wave vibration energy, which can effectively cut off the transmission of longitudinal waves and make the vibration energy carried by longitudinal waves gather on the vibration source and the casing near it. On the contrary, the vibration energy carried by shear waves and torsional waves can freely pass through the flange without changing the transmission direction. Therefore, shear waves in the casing are more likely to cause the whole machine vibration of the aircraft engine. The vibration energy transmitted along the main path on the casing is not carried by a single type of vibration wave throughout the whole process, but different types of vibration waves will convert and carry each other during the transmission process. The bending wave vibration energy is mainly distributed near the casing connected to the support plate, which is offset from the energy distribution position of the shear wave and is the source of the main transmission path of the vibration energy on the casing. Therefore, dissipating the bending wave vibration energy at this location can effectively block the transmission of vibration energy and reduce the vibration noise during the operation of the aircraft engine. The steady-state structural sound intensity field reflects the average distribution law and transmission characteristics of vibration energy within a vibration response cycle. As the source of the vibration energy of the casing, the steady-state peak value of the bending wave vibration energy is equivalent to the transient peak value. The steady-state structural sound intensity is the complex vector product of force and displacement (velocity), so a single amplitude (velocity) distribution cannot point out and display the information such as the vibration source, vibration convergence point and main vibration propagation path of the casing. The structural sound intensity method analyzes the structural vibration problem from the perspective of energy transfer. Therefore, it has an intrinsic physical connection with the structural vibration characteristics through energy parameters such as the kinetic energy, strain energy and damping dissipation of the structure. The vibration control of the structure is essentially to control the vibration energy flow from three levels: the vibration source, the transmission path and the vibration convergence point by reducing the vibration excitation energy, blocking or changing the vibration energy transmission path and enhancing the vibration energy dissipation.

[0311] Technology application scenario 1: Experimental and analytical study on the transmission law of vibration response signal of aircraft engine support system

[0312] The internal structure of the engine is complex and not easy to disassemble. The high temperature environment inside the engine is very unfavorable to the operation of the sensor. Usually, when the engine is working, the sensor vibration measurement point is selected on the external casing. Before the vibration signal reaches the casing, it will pass through multiple transmission paths and be modulated in multiple ways. These vibration responses are coupled to form a complex vibration response on the casing. Therefore, it is very important to obtain the vibration transmission law of the aircraft engine support system through the casing vibration signal. In this process, vibration energy visualization technology plays a vital role. This technology can convert complex vibration signals into intuitive graphics or images, helping engineers and technicians to more easily understand the transmission characteristics and dynamic response of vibration signals. Through visual representation, abnormal or atypical behaviors in vibration sources and transmission paths can be more clearly identified, thereby achieving more effective fault diagnosis and system maintenance. In addition, vibration energy visualization can also help optimize the layout and adjustment of sensors to ensure the most accurate and useful vibration data to improve the monitoring and maintenance efficiency of the entire aircraft engine support system.

[0313] (1) Harmonic response analysis of aircraft engine support system model

[0314] Establish a simplified three-dimensional model of the support system Figure 7 , and divide the mesh. The bearing seats (from left to right are No. 1 bearing seat, No. 2 bearing seat, and No. 3 bearing seat) use 5mm tetrahedral meshes, and the casing and support plate use 2mm hexahedral meshes, with a total of 677773 nodes and 140944 units.

[0315] The vibration measurement point location is selected as follows: Figure 8 shown.

[0316] The bottom surface of the bearing seat is set as a fixed support, and the first 10 modes of the calculated system are shown in Table 3. The modal superposition method is used for harmonic response analysis, and the first 10 modes are set for calculation. Fig. 9 and Fig.10 are the first two modes of the model.

[0317] Table 3 The first 10 modes of the support system

[0318]

[0319]

[0320] Apply sinusoidal excitation to the side of the bearing seat as follows Fig.11 As shown in the figure, the sizes are 100 N, 200 N, and 300 N. Set the solution frequency range to 0-600 Hz, the load step interval to 5 Hz, and a total of 120 steps.

[0321] The amplitude-frequency response curve of the outer casing obtained by applying excitation to the No. 1 bearing seat is as follows: Fig.12 shown.

[0322] The amplitude-frequency response curve of the outer casing obtained by applying excitation to the No. 2 bearing seat is as follows: Fig.13 shown.

[0323] The amplitude-frequency response curve of the outer casing obtained by applying excitation to the No. 3 bearing seat is as follows: Fig.14 As shown:

[0324] According to the results of modal analysis, when sinusoidal excitation is applied to the support system, the deformation occurs mainly in the outer casing when resonance occurs. From the calculated amplitude-frequency curve, it can be seen that the response has peaks near the frequencies of 90Hz, 120Hz, 240Hz and 350Hz, which is basically consistent with the natural frequency calculated by the modal calculation. In particular, the maximum peak value of the outer casing response appears at a frequency of about 350Hz. The amplitude-frequency curves of the same measuring point are roughly the same when the three bearing seats are excited respectively. Among them, the peak values ​​of the frequency response curves obtained by applying excitation to bearing seats 1 and 3 are similar, and the peak value of the frequency response curve obtained by applying excitation to bearing seat 2 is much higher than the other two. When the same bearing seat is excited, the greater the sinusoidal force applied, the greater the response peak, and the response amplitude of the casing edge measuring points (measurement points 1.3) is larger than that of the middle measuring points (measurement points 2.4).

[0325] (2) Transient response analysis of aircraft engine support system model

[0326] Transient response analysis is used to analyze the displacement and strain of a structure under time-varying loads. Through transient analysis, the response of the outer casing at different frequencies over time can be obtained. Set the solution time step to 0.01s, the solution time to 10s, and a total of 1000 steps. Apply a sine excitation of 200N to the bearing seat. Apply a sine excitation of 50HZ to different bearing seats.

[0327] Apply excitation force to bearing seat No. 1, and the time domain diagram and spectrum diagram of the outer casing measurement point are as follows: Fig.15 shown.

[0328] Apply excitation force to bearing seat No. 2, and the time domain diagram and spectrum diagram of the outer casing measurement point are as follows: Fig.16 shown.

[0329] Apply excitation force to bearing seat No. 3, and the time domain diagram and spectrum diagram of the outer casing measurement point are as follows: Fig.17 shown.

[0330] Excitation force is applied to bearing seats 1 and 2 at frequencies of 50 Hz and 100 Hz respectively. The time domain diagram and spectrum diagram of the outer casing measurement point are shown in Figure 18.

[0331] Excitation force is applied to bearing seats 1 and 3 at frequencies of 50Hz and 100Hz respectively. The time domain diagram and spectrum diagram of the outer casing measurement point are as follows: Fig.19 shown.

[0332] Excitation force is applied to bearing seats 2 and 3 at frequencies of 50Hz and 100Hz respectively. The time domain diagram and spectrum diagram of the outer casing measurement point are as follows: Fig. 20 shown.

[0333] Apply 200N excitation to the three bearing seats at frequencies of 50Hz, 100Hz, and 150Hz respectively. The time domain diagram and spectrum diagram of the outer casing measurement point are shown in the figure below. Fig.21 shown.

[0334] Through transient response analysis, it can be seen that the amplitude of the outer casing measurement point changes periodically with time, and the frequency spectrum of the outer casing response measurement point can be obtained by Fourier transform. Under single-point excitation, it can be seen that a peak appears at 49.5Hz in the frequency spectrum, which is basically consistent with the calculated 50Hz excitation. Among them, the frequency spectrum peak obtained by applying excitation to bearing seat No. 2 is the largest. Under multi-point excitation, the frequency domain graph also has peaks near the given excitation frequency.

[0335] Technology application scenario 2: Prediction of vibration response signals of aircraft engine support system based on deep learning

[0336] The LSTM model in deep learning has shown great potential in the field of aircraft engine fault diagnosis with its excellent long-term memory ability and powerful modeling ability for time series data. By using the LSTM model to model and predict time series data, the working status of the engine can be more accurately evaluated, the health status of the engine can be monitored in real time, and the faults can be predicted, so that maintenance and repair measures can be taken in time to ensure flight safety and normal flight operation. In addition, combined with vibration energy visualization technology, the prediction results of the LSTM model can be displayed in an intuitive form, so that maintenance personnel can more easily identify and analyze complex vibration signals, further improving the understanding of engine status and the accuracy of fault prediction. However, in the face of time series data with missing signals, the existing LSTM model has certain challenges. The presence of missing values ​​will interfere with the learning process of the model and reduce the prediction performance of the model. Therefore, it is of great significance to study how to effectively use the LSTM model to predict aircraft engine sensor data with missing signals and enhance the expression and application of these predictions through vibration energy visualization technology to improve the accuracy and robustness of aircraft engine fault diagnosis.

[0337] (1) Basic model

[0338] RNN stands for Recurrent Neural Networks, which is used to process sequence data. In the traditional neural network model, from the input layer to the hidden layer to the output layer, the layers are fully connected, and the nodes between each layer are disconnected. However, this ordinary neural network is powerless for many problems related to time series. For example, if you want to predict the next word in a sentence, you generally need to use the previous word, because the previous and next words in a sentence are not independent. The reason why RNN is called a recurrent neural network is that the current output of a sequence is also related to the previous output. The specific manifestation is that the network will remember the information of the previous moment and apply it to the calculation of the current output, that is, the nodes between the hidden layers are no longer disconnected but connected, and the input of the hidden layer includes not only the output of the input layer but also the output of the hidden layer at the previous moment.

[0339] In RNN, sequence data is input into the network in chronological order, and each time step corresponds to a hidden state. These hidden states constitute the intermediate representation of the network at different time steps. Hidden states play a vital role in RNN. They carry the information of sequence data by connecting neurons in each hidden layer, and transmit and update this information between different time steps. Specifically, in a simple RNN structure, the hidden state is updated at each time step to reflect the input data of the current time step and the hidden state of the previous time step. For example, in the first layer, at time step t, the neurons in the hidden layer of the RNN will obtain the hidden state ht[1], and at time step t+1, they will receive the hidden state ht[1] from the previous time step and calculate the new hidden state ht+1[1]. In this way, the transmission of hidden states between different time steps forms a continuous processing process of sequence data. In addition, from a vertical direction, different layers are also connected through hidden states. On the horizontal timeline, the hidden state of each time step will be connected to the hidden state of the same level in the next time step, and in the vertical direction, the hidden states of each level will also affect each other through specific connections. This inter-layer connection method makes the RNN model scalable in the vertical direction and can handle deeper feature representations and abstractions. In general, RNN is a structure that can be expanded both horizontally and vertically. Through the transfer and update of hidden states, it realizes the effective modeling and processing of sequence data. In practical applications, although horizontal scalability is not very common, the hierarchical connection and hidden state transfer in the vertical direction are the key mechanisms for the RNN model to process long sequence data and learn complex feature representations. The RNN network structure is as follows Fig.23 shown.

[0340] LSTM (Long Short-Term Memory) is a special type of recurrent neural network that is different from traditional feedforward neural networks. In feedforward neural networks, the input at each moment is considered to be completely unrelated to the input at the next moment, which is reasonable for some tasks (such as image classification), but for tasks involving time series (such as natural language processing or continuous data analysis), it is crucial to reasonably use past information to analyze the input at the current moment. Recurrent neural networks (RNNs) are designed to meet this need. Simply put, RNNs allow the network to reasonably use previous input information when processing the current input. Its structure allows information to be passed from one time step to the next, allowing the network to model sequence data. For example, when processing natural language text, understanding the meaning of a word requires taking into account the words that appeared before. In this case, RNN can effectively capture the contextual information in the text, not just the information of a single word. As a special form of RNN, LSTM further improves the structure of RNN. By introducing a gating mechanism, it can better handle long-term dependencies and prevent the problem of gradient disappearance or gradient explosion. This makes LSTM perform better when processing time series data, especially in tasks involving long-term memory, such as language modeling, machine translation, etc. Therefore, LSTM has become a powerful tool for processing time series data and has been widely used in natural language processing, continuous data analysis and other fields. The LSTM model is as follows Fig.24 shown.

[0341] In LSTM, one of the key elements is the cell state, which is an important component of the LSTM network. It runs through the entire network and carries the network's memory and summary of the input data. The cell state can be simply understood as the LSTM's memory of the input data, which is updated and adjusted at each time step to reflect the network's understanding and summary of past input information. In LSTM, the update and transfer of the cell state is achieved through a series of gating mechanisms that control the flow and forgetting of information, thereby effectively processing long sequence data and preventing the problem of gradient vanishing or gradient explosion. In symbolic representation, the cell state can be represented by Ct, which represents the memory after passing through the LSTM cell at time t. This vector contains the LSTM's summary of all input information, which can capture the long-term dependencies in the sequence data, and is transferred and updated between different time steps, so that the network can better understand and predict the patterns and regularities in the sequence data. In general, the cell state is one of the cores of the LSTM network, which allows the network to effectively capture and utilize the long-term dependencies in the sequence data, thereby performing well in processing time series data, especially in tasks involving long-term memory and long sequences.

[0342] In LSTM, there are three important gating mechanisms: forget gate, input gate, and output gate, which are responsible for controlling the update and adjustment of the LSTM unit state in order to effectively process time series data and prevent the problem of gradient disappearance or gradient explosion.

[0343] Forget Gate: The function of the forget gate is to selectively forget some information in the previous unit state based on the current input and the output of the previous time step. Specifically, the forget gate first integrates the current input and the output of the previous time step into a vector, and then compresses each component of this vector to the interval (0, 1) through the sigmoid activation function. In this way, if a component is close to 0 after passing through the sigmoid function, it means that the information at the corresponding position will be forgotten; if it is close to 1, this part of the information will be retained.

[0344] Input Gate: The function of the input gate is to control how the current input information is added to the cell state. First, after integrating the current input, the tanh function is used to extract the effective information, and the sigmoid function is used to control the weighting of this information. The combination of these two realizes the function of extracting effective information from the current input and adding it to the cell state according to the weight.

[0345] Output Gate: The output gate is responsible for calculating the output value at the current moment. It maps the current cell state to the interval (1, 1) through the tanh function, then integrates the current input with the output of the previous time step, and extracts the information through the sigmoid function. This information is multiplied by the cell state after tanh processing to obtain the final output value.

[0346] (2) Experiment and data analysis

[0347] This experiment analyzes that three acceleration sensors are arranged at three positions: the low-pressure rotor end of the outer casing, the middle end of the outer casing, and the high-pressure rotor end of the outer casing. The sampling frequency is 1024Hz, the number of sampling rows is 23369, the maximum value is 0.265318g and the minimum value is 0.265318g. Column A is the collected signal of the outer casing mortgage rotor end, column B is the collected signal of the middle end of the outer casing, and column C is the collected signal of the high-pressure rotor end of the outer casing. Assuming that after running for 21s, the data of point B in the middle end of the outer casing is missing for 1s, the LSTM model is used to predict the 1168 data points. In the time domain graph, by comparing the results of time_step=5 and time_step=5000, a more consistent prediction effect is selected for display. The following figure is a time domain graph of the test set and the prediction results, where the orange curve is the prediction result and the blue curve is the actual vibration signal of point B. Fig.25The prediction results have been successfully achieved, and they are relatively consistent with the test set in the time domain graph.

[0348] In the figure below, blue represents the original data points and orange represents the predicted data points.

[0349] The predicted time domain data is transformed into a frequency domain diagram through fast Fourier transform, and compared with the frequency domain diagram of the original signal (test set) measured at point B. Fig.26 , Fig. 27 .

[0350] From the above experiments, we can see that the vibration signal prediction value under the LSTM model is more accurate and stable.

[0351] 1.4 Technology application scenario: Research on vibration signal tracing technology of a certain type of aircraft engine support system

[0352] Aiming at the energy dispersion caused by the discontinuity of the supporting structure, starting from the geometric transformation of the signal, the signal feature enhancement method based on geometric peak transformation is expanded, and a new vibration signal separation and recognition method is proposed; according to the similarities and differences of the excitation signals in different parts, the rank and sparsity characteristics of the vibration signal are explored, and a new separation and recognition framework is proposed, and finally the mapping relationship between the excitation source and the response of the casing measurement point is obtained.

[0353] For a certain type of engine support system, the Lagrangian method is used to obtain the transient / steady-state total vibration energy and energy components of the support system using the excitation signal and the vibration response model of the casing. Combined with the power flow frequency response function, the energy transfer path between multiple bearing seats and the casing is studied. Subsequently, the proposed synchronous oscillation source tracing method is further expanded, fully considering the scenario design for models and actual engineering, and establishing a support system source tracing model based on a multi-source domain adaptive algorithm for scenarios with multiple inducing conditions for subsynchronous oscillations in actual systems ( Fig.28 As shown in Figure 2, a subsynchronous oscillation tracing method with multiple inducing conditions being effective at the same time is proposed.

[0354] Finally, based on the open-loop mode resonance theory, an excitation source inversion method in the equivalent system is proposed. Combined with the mapping relationship data set between the excitation signal and the response signal, a synchronous oscillation tracing model based on a multi-source domain adaptive network is established, and a casing vibration tracing method under single / multi-point excitation conditions is developed.

[0355] Vibration energy visualization plays a central role in this series of methods. By converting complex vibration data into intuitive visual charts, technicians can more accurately track the energy transfer path and distribution. This visualization not only helps identify and analyze vibration sources and transfer paths, but also effectively monitors and adjusts signal processing methods to improve the accuracy of identification and analysis. Through intuitive graphical display, vibration energy visualization technology greatly improves engineers' understanding and control of complex system dynamics, thereby promoting more efficient and accurate fault diagnosis and system optimization.

[0356] (1) Establishment of the first generation vibration response traceability dictionary

[0357] Set the excitation F = 20000N / 100Hz, apply excitation to bearing seat No. 1 / 2, bearing seat No. 2 / 3, and bearing seat No. 1 / 3 respectively, observe the average amplitude of vibration response of four observation points A, B, C, and D under the same excitation, and calculate the amplitude response ratio of different bearing seats at each point. The calculation results are as follows: Fig.29 shown.

[0358] Calculate the proportion of the four response observation points A, B, C, and D under different excitations. The calculation results are as follows Fig.30 shown.

[0359] (2) The first generation of vibration response traceability dictionary evaluation method

[0360] The dictionary evaluates the vibration source from both horizontal and vertical aspects, effectively improving the accuracy of the judgment.

[0361] Horizontal evaluation process: when the excitation source is at bearing seat No. 3, the response amplitude of each measuring point is the largest; when the excitation source is at bearing seat No. 2, the response amplitude of each measuring point is second; when the excitation source is at bearing seat No. 1, the response amplitude of each measuring point is the smallest.

[0362] Longitudinal evaluation process: when the excitation source is at bearing seat No. 1, the response amplitude of measuring point A is the largest, and the response amplitude of measuring point C is the smallest; when the excitation source is at bearing seat No. 2, the response amplitude of measuring point D is the largest, and the response amplitude of measuring point B is the smallest; when the excitation source is at bearing seat No. 3, the response amplitude of measuring point D is the largest, and the response amplitude of measuring point C is the smallest.

[0363] (3) Prospects for the establishment of the Nth generation vibration response traceability dictionary

[0364] Due to time constraints, the team will optimize and upgrade the first-generation vibration response traceability dictionary in the future. Starting from the vibration signal time domain diagram, spectrum diagram, and envelope spectrum diagram, the spectral characteristics of the vibration signal monitoring points can be analyzed to improve the accuracy and effectiveness of the vibration tracing method.

[0365] Example 1: Vibration signal tracing and visualization analysis of aircraft engine support system

[0366] 1. Overview

[0367] This embodiment performs source tracing and visualization analysis of vibration signals for a certain type of aircraft engine support system.

[0368] 2. Specific steps

[0369] S1: Study on the vibration response transmission law of aircraft engine support system

[0370] The vibration model of the support system is established using finite element analysis software (such as ANSYS).

[0371] Simulate the vibration transmission under different working conditions and analyze the vibration transmission law between the components of the support system.

[0372] S2: Experimental and analytical study on the transmission law of vibration response signal of aircraft engine support system

[0373] Acceleration sensors are arranged at key positions of the engine support system.

[0374] Run the engine and collect vibration signals, and record the vibration data through the data acquisition system.

[0375] Analyze the vibration signal using spectrum analysis tools (such as MATLAB) to verify the accuracy of the numerical model.

[0376] S3: Prediction of vibration response signals of aircraft engine support system based on deep learning

[0377] A large amount of vibration signal data is collected and used as training data sets after preprocessing.

[0378] A convolutional neural network (CNN) is used to build a vibration signal prediction model for model training and optimization.

[0379] The prediction accuracy of the model is verified by using the test data set.

[0380] S4: Research on the Vibration Signal Tracing Technology of a Certain Aircraft Engine Support System

[0381] Combining the aforementioned vibration transmission laws and deep learning models, reverse analysis technology is used to trace the source and propagation path of the vibration signal.

[0382] Use time-frequency analysis methods to locate critical vibration sources.

[0383] S5: Visualization of aircraft engine vibration energy

[0384] Use visualization software (such as Paraview) to visualize the vibration energy.

[0385] Generate 3D heat maps to intuitively display the distribution and transfer path of vibration energy, helping engineers understand vibration conditions.

[0386] Example 2: Vibration monitoring and fault diagnosis of aircraft engine support system

[0387] 1. Overview

[0388] This embodiment performs vibration monitoring and fault diagnosis on a certain type of aircraft engine support system.

[0389] 2. Specific steps

[0390] S1: Study on the vibration response transmission law of aircraft engine support system

[0391] A physical model of the support system is established, and the vibration transmission under different working conditions is simulated through a vibration test bench.

[0392] The analytical method is combined with numerical simulation to analyze the transmission path and law of vibration between different components.

[0393] S2: Experimental and analytical study on the transmission law of vibration response signal of aircraft engine support system

[0394] Acceleration sensors and displacement sensors are installed at different parts of the engine to collect vibration signals under different working conditions.

[0395] Use spectrum analysis and time domain analysis methods to study the transmission rules of vibration signals between different parts.

[0396] S3: Prediction of vibration response signals of aircraft engine support system based on deep learning

[0397] Collect a large amount of experimental data and use it as input data for deep learning models after preprocessing.

[0398] A recurrent neural network (RNN) model was constructed, trained and validated to predict vibration response signals.

[0399] S4: Research on the Vibration Signal Tracing Technology of a Certain Aircraft Engine Support System

[0400] By analyzing experimental data and the prediction results of the deep learning model, the source and propagation path of the vibration signal can be determined.

[0401] Use time-frequency analysis techniques such as wavelet transform to accurately locate the source of the fault.

[0402] S5: Visualization of aircraft engine vibration energy

[0403] Using graphics processing software, the distribution of vibration energy can be visualized in three dimensions.

[0404] Generate dynamic heat maps to display the changes in vibration energy in real time, helping engineers to detect potential faults in a timely manner.

[0405] S6: Experimental and analytical study on the transmission law of vibration response signal of aircraft engine support system

[0406] Further experimental verification and analysis will be conducted to collect more data for model optimization and improve the vibration transmission law model.

[0407] S7: Prediction of vibration response signals of aircraft engine support system based on deep learning

[0408] Use newly acquired experimental data to further train and optimize deep learning models to improve prediction accuracy.

[0409] S8: Research on Vibration Signal Tracing Technology of Aircraft Engine Support System

[0410] By integrating new experimental data and optimized tracing algorithms, the positioning accuracy of the vibration signal source and propagation path can be improved, ensuring the application effect of tracing technology in different types of engines.

[0411] The above description is only a specific implementation mode of the present invention, but the protection scope of the present invention is not limited thereto. Any modifications, equivalent substitutions and improvements made by any technician familiar with the technical field within the technical scope disclosed by the present invention and within the spirit and principle of the present invention should be covered by the protection scope of the present invention.

Claims

1. A method for visualizing the source of vibration energy of a supporting stator mechanical structure, characterized in that: The technology includes: Step 1. Input and define material properties and finite element types through the material and element interface; Step 2. Input the geometric parameters of the research object through the structural model command interface and create a geometric model; Step 3. Input the material properties, unit types and geometric models generated in steps 1 and 2 into the ANSYS pre-processing module to create a finite element model of the research object; Step 4. Input the load attributes and action position through the load condition command interface to apply the excitation load to the finite element model created in step 3; Step 5. Input the constraint type and action position through the boundary condition command interface to apply constraints to the finite element model created in step 3; Step 6. Enter the ANSYS solution module, input the dynamic solution type and its related setting parameters through the solver setting command interface, and perform the solution; Step 7. Enter the ANSYS post-processing module, enter the analysis time, data extraction location, data type, data output save location through the post-processing setting command interface, and extract the internal force and velocity parameters in the finite element model of the research object, and save the output to the file system.

2. The method for visualizing the source of vibration energy of a supporting stator mechanical structure according to claim 1, characterized in that: A method for visualizing the vibration energy of an aircraft engine support system is used, comprising the following steps: Read the calculation result files and unit node relationship files output by ANSYS from the file system, organize them into a matrix and store them in the workspace; Use MATLAB to read the unit coordinate system, node coordinate system, and result coordinate system of the finite element model to convert the physical space into the computational space; Construct the calculation domain of structural sound intensity vector field; According to the relevant calculation formula, the structural sound intensity vector of each unit is solved in the calculation domain, and the result is written into the matrix according to the unit node relationship; Output the structural sound intensity calculation results and save them to the file system; For the transient structural acoustic intensity vector field, the above steps are repeated until all analysis moments are processed; Use the post-processing software TECPLOT to read the constructed calculation domain, read the structural sound intensity calculation result file, and draw the structural sound intensity vector diagram; Save the drawn structure-borne sound intensity vector diagram to the specified file directory.

3. The method for visualizing the source of vibration energy of a supporting stator mechanical structure according to claim 1, characterized in that: The step 1 comprises: (a) Based on the simulated structure of the aircraft engine support system, an improved blind source separation algorithm is developed using information theory and maximum entropy algorithm to decompose the vibration response signals of each measuring point of the casing and determine the characterization relationship between the signals; (b) Combined with the power flow frequency response function, the transmission law of the transient and steady-state total vibration energy and the vibration energy components between the bearing seat and the casing is studied in detail; (c) Conduct basic research on structure-borne sound intensity method and vibration wave theory to provide theoretical support for the visualization of vibration energy; (d) Compile the trace visualization code to realize the visualization expression of energy traces.

4. The method for visualizing the source of vibration energy of a supporting stator mechanical structure according to claim 1, characterized in that: The S2 step includes: (a) Focusing on the vibration transmission problem of the aircraft engine support system, the excitation-response mapping relationship is deeply analyzed using the obtained casing vibration response signal; (b) Establish a simplified finite element model of the support system and perform meshing to simulate the actual structure; (c) Through single-point excitation and multi-point excitation experiments, the transient vibration signals of the response measuring points are analyzed at a fixed frequency to reveal the transmission law of the vibration response.

5. The method for visualizing the source of vibration energy of a supporting stator mechanical structure according to claim 1, characterized in that: The step 3 comprises: (a) Based on the dynamic model of the aircraft engine support system, the LSTM model is used to model and predict the time series of the casing vibration response signal; (b) Accurately predict the time domain vibration signal by acquiring and training the vibration signals of three sampling points of the outer casing of the dual-rotor system; (c) Perform fast Fourier transform on the predicted time domain signal to obtain accurate prediction of the spectral characteristics to provide support for the health management of aircraft engines.

6. The method for visualizing the source of vibration energy of a supporting stator mechanical structure according to claim 1, characterized in that: The step 7 comprises: (a) Aiming at the energy dispersion problem caused by the discontinuity of the supporting structure, a signal feature enhancement method based on geometric peak transformation is proposed to improve the accuracy of vibration signal separation and recognition; (b) Aiming at the subsynchronous oscillation problem under multi-source induced conditions, a support system tracing model based on a multi-source domain adaptive algorithm is established, and an effective subsynchronous oscillation tracing method is proposed; (c) Based on the open-loop mode resonance theory, an excitation source inversion method in the equivalent system is developed. Combined with the mapping relationship data set between the excitation signal and the response signal, a synchronous oscillation tracing model of the multi-source domain adaptive network is constructed to realize the casing vibration tracing under single / multi-point excitation conditions.

7. A system for visually tracing the vibration energy of an aircraft engine support system based on the method of claim 1, characterized in that: The system includes: A vibration signal decomposition module is used to decompose the vibration response signals of each measuring point of the casing and determine the characterization relationship between the signals based on the simulated structure of the aircraft engine support system and the blind source separation algorithm developed by information theory and maximum entropy algorithm; A vibration energy transfer analysis module, which is used to study the transient and steady-state total vibration energy and the transmission law of vibration energy components between the bearing seat and the casing in combination with the power flow frequency response function; A trace visualization module, used to compile trace visualization code and realize the visualization expression of energy traces; A data storage module is used to store the processed vibration response signal and energy trace data.

8. A system for studying the vibration transmission of an aircraft engine support system based on the method of claim 1, characterized in that: The system includes: A finite element model building module, which is used to build and mesh a simplified finite element model of an aircraft engine support system to solve the vibration transmission problem of the support system; An excitation response analysis module is used to use the obtained casing vibration response signal to deeply analyze the excitation response mapping relationship and reveal the transmission law of vibration response; An experimental simulation module for analyzing transient vibration signals of response measurement points under single-point excitation and multi-point excitation at a fixed frequency; A data output module is used to output the analysis results of vibration transmission laws.

9. A system for predicting vibration signals and analyzing spectrum characteristics of aircraft engine support systems based on the method of claim 1, characterized in that: The system includes: A time series modeling module, which is used to perform time series modeling and prediction of casing vibration response signals using an LSTM model based on the dynamics model of the aircraft engine support system; A vibration signal prediction module is used to accurately predict the time domain vibration signal by acquiring and training the vibration signals of three sampling points of the outer casing of the dual-rotor system; A spectrum feature analysis module, used to perform fast Fourier transform on the predicted time domain signal to obtain accurate prediction of the spectrum features; A health management support module for applying prediction and analysis results to the health management of aircraft engines.

10. A system for tracing the vibration source of an aircraft engine support system based on the method of claim 1, characterized in that: The system includes: A signal feature enhancement module is used to address the energy dispersion problem caused by the discontinuity of the supporting structure. A signal feature enhancement method based on geometric peak transformation is proposed to improve the accuracy of vibration signal separation and recognition; A subsynchronous oscillation source tracing module is used to establish a support system source tracing model based on a multi-source domain adaptive algorithm for subsynchronous oscillation problems under multi-source induced conditions, and propose an effective subsynchronous oscillation source tracing method; An excitation source inversion module is used to perform excitation source inversion in an equivalent system based on the open-loop mode resonance theory, and to build a synchronous oscillation tracing model of a multi-source domain adaptive network by combining the mapping relationship dataset between the excitation signal and the response signal; A vibration tracing implementation module is used to realize casing vibration tracing under single / multi-point excitation conditions.

Images