Aero-engine warm-up process on-board data correlation analysis and visualization method

Abstract

The present application belongs to the technical field of airborne data analysis of aero-engine, and particularly relates to a method for analyzing and visualizing the correlation of airborne data in the warming-up process of an aero-engine, comprising: data acquisition, data slicing, data classification and reconstruction, sample matrix construction, correlation calculation, time sequence relationship screening, significance screening, and correlation analysis result visualization. The present application slices and reconstructs the airborne data based on the working state, can comprehensively analyze the correlation between various parameters in the warming-up process of an aero-engine, and provides more complete information compared with the traditional method, which is helpful for deeply understanding the vibration response characteristics of the warming-up process and provides support for the optimized design of the warming-up program.

Description

Technical Field

[0001] This invention belongs to the field of airborne data analysis technology for aero-engines, specifically relating to a method for correlation analysis and visualization of airborne data during the warm-up process of an aero-engine. Background Technology

[0002] In the aviation field, engine warm-up is a crucial pre-flight preparation procedure. This process involves running the engine at a high speed for a period of time before takeoff, raising the temperature throughout the engine, bringing its stress and deformation conditions close to operational levels, and optimizing the rotor-stator clearance to improve engine performance and reduce internal damage. For military engines, the warm-up process is particularly time-sensitive. Excessive warm-up time slows down the time to reach the designated airspace, impacting operational efficiency and responsiveness. Conversely, insufficient warm-up time may fail to achieve the desired preheating effect, reducing engine thrust, increasing takeoff distance, and even causing serious malfunctions due to uneven thermal deformation or inadequate lubrication.

[0003] By analyzing engine parameters recorded by airborne sensors, it is possible to conduct in-depth research on the impact of various parameters on the warm-up effect. This process is of great significance for designing more efficient warm-up programs to balance warm-up time and effectiveness, thereby enhancing the combat readiness response capability of military engines.

[0004] Correlation analysis is a commonly used data analysis method in existing technologies, used to study the influence relationships between variables and infer their causal connections. While correlation analysis can help identify key factors affecting warm-up performance, its application in warm-up process analysis has limitations. Specifically, most existing methods can only analyze the correlation between the maximum engine vibration value and warm-up time, failing to comprehensively assess the combined impact of numerous parameters during the aero-engine warm-up process. Furthermore, when multiple variables are involved, the complexity of correlation analysis increases dramatically, leading to difficulties in data interpretation and information redundancy. Summary of the Invention

[0005] To address the aforementioned technical problems, this invention provides a method for correlation analysis and visualization of airborne data during the warm-up process of an aero-engine. The purpose is to comprehensively analyze the correlation relationships of airborne data, automatically filter and present effective information, and provide a reference for the optimized design of the warm-up program.

[0006] This invention is implemented by providing a method for correlation analysis and visualization of airborne data during the warm-up process of an aero-engine, comprising the following steps: Step S0: Data acquisition, collecting airborne data and performing data preprocessing; Step S1: Data slicing. Based on the engine operating status parameters, the airborne data of each test run is divided into time periods, and airborne data segments during the warm-up process are extracted. Step S2: Data classification and reconstruction. The airborne data fragments during the warm-up process are divided into independent variable parameters and dependent variable parameters, and features are extracted to construct independent variable tensors and dependent variable tensors. Step S3: Sample matrix construction. Flatten the independent and dependent variable tensors according to the number of trials to obtain the independent variable parameter matrix. and dependent variable parameter matrix ; Step S4: Correlation calculation, calculate the independent variable parameter matrix. and dependent variable parameter matrix Correlation coefficient matrix ; Step S5: Time series relationship filtering. Based on the time series relationships, the correlation coefficient matrix is ​​filtered. The system is divided into a valid region and an invalid region, and the coefficients in the invalid region are set to 0. Step S6: Significance screening, setting a significance threshold The correlation coefficient matrix Less than the significance threshold The correlation coefficient was considered a non-significant correlation coefficient, so the correlation coefficient matrix was deleted. The non-significant correlation coefficient in; Step S7: Visualize the correlation analysis results by drawing a correlation heatmap, using colors to represent the degree of correlation between the independent and dependent variable parameters.

[0007] Preferably, in step S0, the airborne data is aircraft-engine parameter data, including casing vibration, throttle lever angle, main fuel setpoint, afterburner fuel setpoint, inlet air total temperature, low-pressure rotor relative speed, high-pressure rotor relative speed, low-pressure turbine afterburner total gas temperature, low-pressure turbine afterburner exhaust pressure, and aircraft axial overload. Data preprocessing includes noise filtering and time alignment.

[0008] Preferably, in step S1, the warm-up process is divided into a start-up phase, an idle inspection phase, a warm-up phase, a standby phase, and a take-off phase. By analyzing the relative speed of the high-pressure rotor and the axial overload of the aircraft, the start and end times of each phase are determined, and the time interval between the start and end times of each phase is used as a preset time slice to extract the airborne data segments within each preset time slice.

[0009] Preferably, in step S2, the dependent variable parameters include, but are not limited to, the total vibration amplitude at the measuring point location, the amplitude of each component of the spectrum, the lubricating oil temperature, and parameters used to characterize the warm-up response of the aero-engine; the working state parameters of the warm-up process other than the dependent variable parameters are regarded as independent variable parameters. By extracting the time-domain feature parameters of the independent and dependent variable parameters under different test runs and different states, the features of the independent and dependent variable parameters are reconstructed into a fourth-order tensor, wherein the time-domain feature parameters are one or more statistics selected from a preset set of statistics, which includes the mean, maximum value, variance, and other statistics used to characterize the time-domain variation characteristics of the parameters. In, elements Indicates the first The first test run, the first The data segment, the first The parameter, the first Each time-domain feature.

[0010] Preferably, in step S3, the independent variable parameter matrix Dependent variable parameter matrix The same row in the matrix corresponds to the same test run sample, and the independent variable parameter matrix Each column in the matrix corresponds to a time-domain feature of an independent variable parameter within a preset time slice; the dependent variable parameter matrix... Each column in the table corresponds to a time-domain feature of a dependent variable parameter within a preset time slice.

[0011] Preferably, in step S4, the correlation coefficient matrix elements in express The Middle Variables and The Middle The correlation coefficients of the variables.

[0012] Preferably, in step S5, the specific operation of time series relationship filtering is as follows: for the correlation coefficient matrix Each element in the matrix is ​​labeled with a time tag 1, 2, 3, 4, and 5, based on its corresponding independent and dependent variable parameters. These five time tags correspond sequentially to the start-up, idle check, warm-up, standby, and takeoff phases of the warm-up process. The correlation coefficient matrix is ​​then... Elements whose independent variable parameter time tag is greater than the dependent variable parameter time tag are classified as invalid and assigned a value of 0.

[0013] Preferably, in step S6, if the correlation coefficient matrix If an entire column contains elements with no significant correlation coefficients, delete the entire column. If an entire row contains elements with no significant correlation coefficients, delete the entire row.

[0014] Compared with the prior art, the advantages of the present invention are as follows: 1) This invention, by slicing and reconstructing airborne data based on operational status, enables a comprehensive analysis of the correlations between various parameters during the warm-up process of an aero-engine. Compared to traditional methods, this invention provides more complete information, facilitating a deeper understanding of the vibration response characteristics of the warm-up process and supporting the optimized design of the warm-up program; 2) By filtering based on time series relationships and significance, this invention effectively eliminates irrelevant or insignificant correlation information, significantly reducing data redundancy and avoiding information overload in multi-parameter analysis. This efficient information processing method makes the correlation analysis results clearer, and the degree of correlation between variables is intuitively displayed through heatmaps, which helps to quickly identify key parameters and their mutual influences, providing an intuitive basis for optimizing the warm-up program. After removing useless information, the visualization results are more focused and effective, significantly improving the practicality and operability of the analysis. Attached Figure Description

[0015] Figure 1 This is a schematic diagram of the relative speed signal of the high-pressure rotor of the engine during the warm-up process in an embodiment of the present invention; Figure 2 This is a schematic diagram of the relative speed signal of the high-voltage rotor during the start-up phase of the warm-up process in an embodiment of the present invention; Figure 3 This is a schematic diagram of the relative rotational speed signal of the high-voltage rotor in the warm-up section, as shown in an embodiment of the present invention. Figure 4 This is a schematic diagram of the axial overload of the takeoff section and the relative speed signal of the high-pressure rotor in an embodiment of the present invention; Figure 5 This is a schematic diagram of a data slice of the warm-up process in an embodiment of the present invention; Figure 6 This is a schematic diagram of the effective and invalid regions of the correlation coefficient matrix in an embodiment of the present invention; Figure 7 This is a schematic diagram of the correlation heatmap in an embodiment of the present invention. Detailed Implementation

[0016] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0017] This invention provides a method for correlation analysis and visualization of airborne data during the warm-up process of an aero-engine, the specific steps of which are as follows: Step S0: Data acquisition. Collect airborne data, which consists of aircraft-engine parameter data. Perform preprocessing on the data. Aircraft-engine parameter data includes casing vibration, throttle lever angle, main fuel input, afterburner fuel input, total inlet air temperature, relative speed of low-pressure rotor, relative speed of high-pressure rotor, total exhaust gas temperature after low-pressure turbine, exhaust pressure after low-pressure turbine, and aircraft axial overload. Data preprocessing includes noise filtering and time alignment. Step S1: Data Slicing. Based on the engine operating status parameters, the airborne data of each test run is divided into time periods, and airborne data segments related to the warm-up process are extracted. The warm-up process includes: start-up phase, idle check phase, warm-up phase, standby phase, and take-off phase. First, differential calculations are performed on the relative speeds of the high-voltage rotor, and a stabilization threshold is set. The difference is greater than Then it is considered an "increase", greater than - and less than Then it is considered "stationary", less than - This is considered a "decline". The signal is divided according to this rule, and the result is as follows: Figure 1 As shown.

[0018] By taking the starting point of the first rising region in the speed signal as the beginning of the starting segment and the starting point of the first stable region after the start of the starting segment as the end of the starting segment, the identification of the starting segment can be completed. Figure 2 As shown.

[0019] Based on the preset warm-up speed, the first stable region where the relative speed of the high-voltage rotor is greater than 85% is extracted. Then, adjacent rising and falling regions are grouped into the warm-up section. The starting point of the rising region is the starting point of the warm-up section, and the ending point of the falling region is the ending point of the warm-up section. Figure 3 As shown.

[0020] An axial overload threshold gZ is set. After the time point when the axial overload signal first exceeds this threshold is detected, the time point when the high-pressure rotor speed signal last falls below 85% is traced back is determined as the start point of the takeoff segment. The end point of the takeoff segment is artificially defined as 60 seconds after the end point. Figure 4 As shown.

[0021] After determining the start and end times of the "start-up phase," "warm-up phase," and "takeoff phase," the time ranges for other phases can be further determined. The end point of the "start-up phase" is the start point of the "idle inspection phase," and the start point of the "warm-up phase" is its end point. Similarly, the end point of the "warm-up phase" and the start point of the "takeoff phase" are determined as the start and end times of the "standby phase," respectively. Figure 5 As shown.

[0022] In this way, the start and end times of the start-up phase, idle inspection phase, warm-up phase, standby phase, and takeoff phase in the warm-up sequence were determined, and then the parameter data for each time period were extracted.

[0023] Step S2: Data Classification and Reconstruction. In this embodiment, the casing vibration is analyzed, and the casing vibration is classified as the dependent variable parameter, while the other nine parameters are classified as independent variables. For each test run, each data segment, and each parameter, four statistics are extracted: mean, maximum value, variance, and fourth-order central moment. The features of the independent and dependent variables from the 80 test run data are reconstructed into a fourth-order tensor, with the shape of the independent variable parameter feature tensor being... The shape of the dependent variable parameter characteristic tensor is .

[0024] Step S3: Sample matrix construction. Flatten the feature tensor according to the number of trials to obtain the independent variable parameter matrix. , shape is The dependent variable parameter matrix is ​​obtained. , shape is .

[0025] Step S4: Correlation calculation, calculate the independent variable parameter matrix. and dependent variable parameter matrix Correlation coefficient matrix Correlation coefficient matrix elements in express The Middle Variables and The Middle The correlation coefficients of the variables. Correlation coefficient matrix. It has 180 rows and 20 columns, with a total of 3600 elements.

[0026] Step S5: Time Series Relationship Filtering: Based on the time series relationships, add time labels 1, 2, 3, 4, and 5 to the independent and dependent variable parameters of the starting segment, idle inspection segment, warm-up segment, standby segment, and takeoff segment, respectively; for the correlation coefficient matrix... coefficients in ,like If the time label of the corresponding independent variable parameter is less than or equal to the time label of the dependent variable parameter, then... The location is designated as the valid area, if If the time label of the corresponding independent variable parameter is greater than the time label of the dependent variable parameter, then... The location is designated as the invalid region; the valid region represents correlation coefficients that conform to the time sequence relationship of the warm-up process, while the invalid region represents correlation coefficients that do not conform to the time sequence relationship of the warm-up process, such as... Figure 6 As shown; assigning 0 to the elements in the invalid region will allow the correlation coefficient matrix to be reconstructed. It is transformed into an upper triangular block array.

[0027] Step S6, Significance Filtering: Set a significance threshold Delete the correlation coefficient matrix Less than the significance threshold The non-significant correlation coefficients, the correlation coefficient matrix All values ​​less than the significance threshold Delete entire rows and columns.

[0028] Step S7, Visualization of Correlation Analysis Results: A correlation heatmap is plotted, using colors to represent the degree of correlation between independent and dependent variable parameters, such as... Figure 7 As shown.

[0029] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for correlation analysis and visualization of airborne data during the warm-up process of an aero-engine, characterized in that, Includes the following steps: Step S0: Data acquisition, collecting airborne data and performing data preprocessing; Step S1: Data slicing, dividing the airborne data of each test run into time periods and extracting airborne data segments during the warm-up process; Step S2: Data classification and reconstruction. The airborne data fragments during the warm-up process are divided into independent variable parameters and dependent variable parameters, and features are extracted to construct independent variable tensors and dependent variable tensors. Step S3: Sample matrix construction. Flatten the independent and dependent variable tensors according to the number of trials to obtain the independent variable parameter matrix. and dependent variable parameter matrix ; Step S4: Correlation calculation, calculate the independent variable parameter matrix. and dependent variable parameter matrix Correlation coefficient matrix ; Step S5: Time series relationship filtering. Based on the time series relationships, the correlation coefficient matrix is ​​filtered. The system is divided into a valid region and an invalid region, and the coefficients in the invalid region are set to 0. Step S6: Significance screening, setting a significance threshold The correlation coefficient matrix Less than the significance threshold The correlation coefficient was considered a non-significant correlation coefficient, so the correlation coefficient matrix was deleted. The non-significant correlation coefficient in; Step S7: Visualize the correlation analysis results by drawing a correlation heatmap, using colors to represent the degree of correlation between the independent and dependent variable parameters.

2. The method for correlation analysis and visualization of airborne data during the warm-up process of an aero-engine according to claim 1, characterized in that, In step S0, the airborne data is aircraft-engine parameter data, including casing vibration, throttle lever angle, main fuel setpoint, afterburner fuel setpoint, inlet air total temperature, low-pressure rotor relative speed, high-pressure rotor relative speed, low-pressure turbine afterburner total gas temperature, low-pressure turbine afterburner exhaust pressure, and aircraft axial overload. Data preprocessing includes noise filtering and time alignment.

3. The method for correlation analysis and visualization of airborne data during the warm-up process of an aero-engine according to claim 1, characterized in that, In step S1, the warm-up process is divided into a start-up phase, an idle inspection phase, a warm-up phase, a standby phase, and a takeoff phase. By analyzing the relative speed of the high-pressure rotor and the axial overload of the aircraft, the start and end times of each phase are determined, and the time interval between the start and end times of each phase is used as a preset time slice. Airborne data segments within each preset time slice are then extracted.

4. The method for correlation analysis and visualization of airborne data during the warm-up process of an aero-engine according to claim 1, characterized in that, In step S2, the dependent variable parameters include, but are not limited to, the total vibration amplitude at the measurement point location, the amplitude of each component of the spectrum, the lubricating oil temperature, and parameters used to characterize the warm-up response of the aero-engine; the working state parameters of the warm-up process other than the dependent variable parameters are regarded as independent variable parameters. By extracting the time-domain feature parameters of the independent variable parameters and the dependent variable parameters under different test runs and different states, the features of the independent variable parameters and the dependent variable parameters are reconstructed into a fourth-order tensor respectively.

5. The method for correlation analysis and visualization of airborne data during the warm-up process of an aero-engine according to claim 3, characterized in that, In step S3, the independent variable parameter matrix Dependent variable parameter matrix The same row in the matrix corresponds to the same test run sample, and the independent variable parameter matrix Each column in the matrix corresponds to a time-domain feature of an independent variable parameter within a preset time slice; the dependent variable parameter matrix... Each column in the table corresponds to a time-domain feature of a dependent variable parameter within a preset time slice.

6. The method for correlation analysis and visualization of airborne data during the warm-up process of an aero-engine according to claim 1, characterized in that, In step S4, the correlation coefficient matrix elements in express The Middle Variables and The Middle The correlation coefficients of the variables.

7. The method for correlation analysis and visualization of airborne data during the warm-up process of an aero-engine according to claim 3, characterized in that, In step S5, the specific operation of time series relationship filtering is as follows: for the correlation coefficient matrix Each element in the matrix is ​​labeled with a time tag 1, 2, 3, 4, and 5, based on its corresponding independent and dependent variable parameters. These five time tags correspond sequentially to the start-up, idle check, warm-up, standby, and takeoff phases of the warm-up process. The correlation coefficient matrix is ​​then... Elements whose independent variable parameter time tag is greater than the dependent variable parameter time tag are classified as invalid and assigned a value of 0.

8. The method for correlation analysis and visualization of airborne data during the warm-up process of an aero-engine according to claim 1, characterized in that, In step S6, if the correlation coefficient matrix If an entire column contains elements with no significant correlation coefficients, delete the entire column. If an entire row contains elements with no significant correlation coefficients, delete the entire row.

Images