Method for evaluating typical gas path component characteristics of aero-engine based on flight data

Abstract

The invention relates to the technical field of aero-engine maintenance, in particular to an aero-engine typical gas circuit component characteristic evaluation method based on flight data, and the method comprises the steps that firstly, two sets of engine simulation models need to be established, secondly, establishing a design point general component characteristic model based on engine design point working data; engine performance parameters which are not measured in flight are complemented by using a general digital twin model; and inputting the complemented gas path parameters into the engine design point general characteristic model, adding correction factors by taking approximation of the complemented engine gas path parameters as a target, and correcting the part characteristic diagram to obtain an engine personalized part characteristic model. The evaluation method can be used for calculating the characteristics of the typical parts of the engine, answering the performance deviation degree of the parts of the engine and the reasons causing the performance degradation of the engine, and providing technical support for the condition-based maintenance work of engines in an overhaul factory and the replacement of airline engine units.

Description

Technical Field

[0001] This invention relates to the field of aircraft engine maintenance technology, and more specifically, to a method for evaluating the characteristics of typical air path components of an aircraft engine based on flight data. Background Technology

[0002] The following four factors can lead to a decline in engine performance during engine operation: (1) Different pollutants (dust, ash, oil droplets, water mist, hydrocarbons and industrial chemicals, etc.) adhere to the surface of the air passage components, resulting in increased roughness of the engine blade surface; (2) Harsh operating environments such as sand and dust continuously erode the engine blades; (3) Irreversible performance deterioration of the engine caused by the oxidation reaction or chemical action of air pollutants (sodium and potassium salts, inorganic acids and other chemical reaction elements) and combustion process pollutants (such as sulfur oxides); (4) Foreign objects (birds, runway gravel, etc.) are sucked into the engine airflow passage or the internal passage components are damaged by the peeling of coatings and other substances; at the same time, as the engine is used for longer, the engine blade tip clearance also increases irreversibly, causing a decrease in the efficiency of engine components and thus a decline in engine performance. Compared to ground-based test benches, which can increase the number of sensors at key engine sections to measure airflow parameters (such as the corresponding outlet pressure and temperature of the fan outlet, the pressure after the high-pressure compressor, and the exhaust pressure after the low-pressure turbine), the test airflow performance data of line engines installed on aircraft is scarce (parameters such as fan speed N1, compressor speed N2, fan guide vane angle a1, compressor guide vane angle a2, low-pressure turbine exhaust temperature T6, nozzle angle D8, and throttle lever angle Pla). Therefore, the existing flight data cannot be used to establish simulation models of engine component characteristics to further evaluate the degree of deviation in engine component performance and the reasons for performance degradation.

[0003] Currently, in the civil aviation field, fingerprint diagrams provided in maintenance manuals can be used in conjunction with nonlinear encoders constructed using 1DCNN and MLP to decouple the deviation values ​​hidden in the air path parameters and to preliminarily assess the most severely degraded components of the engine's air path performance (Multi-component performance evaluation method for aero-engine air path (CN115408924A)). However, some types of aero-engines lack relevant technical data such as fingerprint diagrams, so the above technology cannot be referenced and applied to some types of aero-engines.

[0004] Currently, neural network models can be built based on the engine's physical structure to answer questions about the overall performance degradation of the engine (a method for constructing digital engineering models for aero-engines (CN113987686A)). However, the main problem with this method is that it cannot explain the component-level causes of changes in the overall engine performance parameters (engine thrust, exhaust temperature, etc.). Summary of the Invention

[0005] The purpose of this invention is to provide an evaluation method for the characteristics of typical airflow components of an aero-engine based on flight data. This method can utilize existing flight data from the route to analyze and quantify the degree of performance deviation of engine components and explain the reasons for performance degradation, providing technical support for condition-based engine maintenance work in overhaul factories and replacement of engine units on the route.

[0006] The embodiments of the present invention are implemented as follows: A method for evaluating the characteristics of typical aero-engine gas path components based on flight data, comprising: S1. Build a general digital twin model, using parameters that are measured in both test runs and actual flights as input features, and using air path performance parameters that are measured only in test runs and not in actual flights as target features; S2. Input the flight route data into the general digital twin model to complete the missing air path performance parameters; S3. Establish a general component characteristic model based on engine design point data, and add correction factors to the parameters of each component using the general component characteristic model; S4. Input the completed flight route data into the general component characteristic model and output the deviation values ​​of each component parameter relative to the performance baseline.

[0007] The beneficial effects of the embodiments of the present invention are: This invention provides a method for evaluating the characteristics of typical aero-engine gas path components based on flight data. The method first requires establishing two engine simulation models: a general digital twin model based on ground test bench data and a design-point general component characteristic model based on engine design-point operating data. The general digital twin model is used to complete performance parameters not measured during flight. The completed gas path parameters are then input into the engine design-point general characteristic model, and correction factors are added to approximate the engine test performance parameters, thus correcting the component characteristic diagram and obtaining a personalized engine component characteristic model. This evaluation method can calculate the characteristics of typical engine components, answer the degree of deviation in engine component performance and the causes of engine performance degradation, providing technical support for condition-based engine maintenance in overhaul plants and replacement of engine units on the line. Attached Figure Description

[0008] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for use in the embodiments. It should be understood that the following drawings only illustrate certain embodiments of the present invention and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other relevant drawings can be obtained based on these drawings without paying any creative work.

[0009] Figure 1A flowchart illustrating a method for evaluating the characteristics of typical aero-engine gas path components based on flight data, provided in an embodiment of the present invention; Figure 2 This is a schematic diagram illustrating the feature selection of a method for evaluating the characteristics of typical aero-engine gas path components based on flight data, provided in an embodiment of the present invention. Figure 3 A flowchart illustrating the modification of a general characteristic model for an evaluation method of typical aero-engine gas path components based on flight data, provided in an embodiment of the present invention. Figure 4 The sensitivity analysis results of a method for evaluating the characteristics of typical aero-engine gas path components based on flight data provided in an embodiment of the present invention are shown below. The comparison relationships are as follows: Out_fan: external fan, In_fan: internal fan, Com: compressor, Hpt: high-pressure turbine, Lpt: low-pressure turbine, π: compression ratio, w: flow rate, η: efficiency.

[0010] Figure 5 This is a comparison chart showing the effect of correcting the objective function of an evaluation method for the characteristics of typical aero-engine gas path components based on flight data, provided in an embodiment of the present invention. Detailed Implementation

[0011] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention are described clearly and completely. Obviously, the described embodiments are only a part of the embodiments of the present invention, not all of them. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without inventive effort are within the scope of protection of the present invention. Therefore, the following detailed description of the embodiments of the present invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely to illustrate selected embodiments of the invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without inventive effort are within the scope of protection of the present invention.

[0012] The following provides a detailed description of the evaluation method for typical aero-engine gas path components based on flight data, as provided by this invention. A complete flowchart can be found below. Figure 1 shown.

[0013] This invention provides a method for evaluating the characteristics of typical aero-engine gas path components based on flight data, comprising: S1. Build a general digital twin model, using parameters that are measured in both test runs and actual flights as input features, and using air path performance parameters that are measured only in test runs and not in actual flights as target features; S2. Input the flight route data into the general digital twin model to complete the missing air path performance parameters; S3. Establish a general component characteristic model based on engine design point data, and add correction factors to the parameters of each component using the general component characteristic model; S4. Input the completed flight route data into the general component characteristic model and output the deviation values ​​of each component parameter relative to the performance baseline.

[0014] In the existing technology, some parameters of aero engines are only measured during ground bench tests, and are not measured during actual flight, as detailed in Table 1.

[0015] Table 1. Comparison of Engine Ground Test Bench Test and Flight Test Parameters

[0016] Among them, "√" indicates that the parameter has been measured, "×" indicates that the parameter has not been measured, and "—" indicates that it is uncertain whether the parameter has been measured. This is mainly due to the differences in engine models, some models measure the parameter while others do not.

[0017] A general digital twin model can be used to complete the missing parameters. A specific general digital twin model can be found in Chinese Patent CN13987686A, which describes a hierarchical structure of a neural network model built based on physical information; however, this is not the focus of this invention and will not be elaborated upon here.

[0018] There are similarities and differences between the test parameters and flight measurement parameters of aero-engines. The missing engine cross-section air path parameters in flight are strongly correlated with the performance degradation caused by the manufacturing and installation tolerances and use of aero-engines. Therefore, the prediction of missing parameters in flight requires the selection and filtering of the existing data collected by the engine. The following two requirements must be ensured: (1) The input features are parameters that are measured in both test and actual flight, while the target features are only test parameters; (2) The input features contain information on engine performance degradation, and the physical relationship between the input features and the target features is not affected by engine performance degradation.

[0019] Engine ground test bench data and flight data can be categorized into environmental parameters, external input parameters (flight altitude, Mach number, etc.), parameters measured during both test runs and flights, and parameters measured during test runs but not during flights. All measured parameters refer to engine performance parameters. Environmental parameters and external input parameters, as "external" parameters, are merely the conditions driving the aero-engine's operation and do not have a direct relationship with engine performance. The impact of engine component performance degradation is reflected in the measured performance parameters. Under specific operating conditions, the values ​​of these performance parameters represent the current state of engine performance degradation. Therefore, the relationship between "external" parameters and component performance parameters contains information about performance degradation. The mapping relationship between engine performance parameters represents the thermodynamic relationship of engine components; this implicit physical relationship is unaffected by engine performance degradation. For example... Figure 2 As shown, the relationship between the parameters in the left block diagram is affected by the engine performance degradation, while the relationship between the parameters in the left and right block diagrams is unaffected. Through this classification, the parameters are filtered to the clustering stage, using the parameters in the left block diagram as the input features of this general digital twin model, and the parameters in the right block diagram as the target features of this general digital twin model.

[0020] Furthermore, in other preferred embodiments of the present invention, step S1 further includes: The general digital twin model is trained using test data from multiple engines, ensuring that the absolute value of the self-tested steady-state error of the general digital twin model is less than 1%. Optionally, the number of engines should be no less than 10.

[0021] Furthermore, in other preferred embodiments of the present invention, step S1 further includes: The weights of the trained general digital twin model are corrected using factory test data of a single engine, so that the steady-state error of the general digital twin model in testing a single engine is less than 1%, thus completing personalized training.

[0022] Given the individual differences in engines, this invention uses ground bench test data of a single engine corresponding to the new engine manufacturing acceptance or overhaul qualified engine to adaptively modify the neural network nodes and weight values ​​of the general digital twin model, and verifies it using ground bench test data, so that the results are closer to the specific engine model.

[0023] Furthermore, in other preferred embodiments of the present invention, in step S3, the specific method for establishing a general component characteristic model based on engine design point data is to obtain the engine design point operating parameters and component characteristic diagrams through engine design documents, and then determine the operating point of each component on the characteristic diagram based on traditional engineering thermodynamic calculations and common engine operating conditions, thereby establishing a general component characteristic model for the engine.

[0024] Optionally, when adding correction factors to the parameters of each component, the following important factors need to be considered: (a) The selected component parameters to be corrected must be strongly related to the target performance parameters; otherwise, there will not be enough search space during the correction process, and a better correction result cannot be found.

[0025] (b) There is no limit to the number of correction factors, but different combinations of correction factors will result in different correction effects. Moreover, the more correction factors selected, the larger the search space, and the closer the correction result is to the target parameter.

[0026] (c) The upper and lower limits of the correction factor also determine the range of the search space. Therefore, in the adaptive correction process, it is very important to choose appropriate upper and lower limits of the parameters, which may require multiple trials to ensure that the optimal value is included in the search space.

[0027] Taking a dual-shaft hybrid exhaust turbofan engine as an example, based on the above three principles, 15 alternative component parameters that can be modified by adding correction factors can be introduced for the four rotating components of the engine (fan, compressor, high-pressure turbine, and low-pressure turbine), as shown in Table 2.

[0028] Table 2. Parameters of the component to be corrected

[0029] Specifically, the correction formulas for the parameters of each component are as follows:

[0030] In the formula, Com p ,Com w ,Com e These are the engine component pressure ratio correction factor, airflow correction factor, and efficiency correction factor; pressure ratio π i Traffic W i Efficiency η i These are the pressure ratio, airflow rate, and efficiency values ​​of the engine components after repair; pressure ratio π. i0 Traffic W i0 Efficiency η i0 , , respectively represent the pressure ratio, airflow rate, and efficiency value of the engine component before repair; i refers to the engine component.

[0031] In order to select the most suitable component parameters for correction, in step S3, the complete air path performance parameters of the engine are used as the target performance parameters. Through sensitivity analysis, the component parameters with the highest sensitivity to the target performance parameters are selected and correction factors are added.

[0032] Specifically, the sensitivity analysis employs a single-factor perturbation method, keeping all other correction factors equal to 1, and changing only one correction factor at a time, increasing it by 1%, to calculate the relative changes in the target performance parameters using an engine performance model. The rates of change of each target performance parameter caused by changes in the parameters of a single component to be corrected are summed and normalized to obtain the sensitivity of the performance parameters to the correction factors. Based on the sensitivity analysis results, components are selected for correction.

[0033] Furthermore, in other preferred embodiments of the present invention, step S3 further includes: By using the variable-weight particle swarm optimization algorithm to optimize the added correction factor, the output of the general component characteristic model is aimed at approximating the completed engine air path parameters, thus completing the personalized training of the general component characteristic model.

[0034] Particle Swarm Optimization (PSO) is an intelligent optimization algorithm proposed by Drs. Kennedy and Eberhart CgJ in 1995 based on simulations of bird flock foraging behavior. It has received widespread attention in recent years. This algorithm can effectively solve multi-parameter optimization problems and is an efficient global optimization method. In PSO, each particle has a fitness function to determine its position and a velocity value to determine its direction and distance of flight. Finally, the particle swarm searches the solution space by following the current best particle. Each particle updates its velocity and position according to the following formula.

[0035]

[0036] In the formula: i,j=1,2,…,d (d-dimensional search space); t is the number of iterations; w is the inertia weight; c1 and c2 are positive learning factors; r1 and r2 are random numbers uniformly distributed between 0 and 1; p ij The optimal solution position (pbest) found for the particle itself; p gj The optimal position (gbest) found for the entire group.

[0037] The convergence of the engine model was studied by taking the concave, convex, and linearly decreasing forms of the w function. Considering both the calculation results and convergence speed, it was determined that the particle swarm optimization algorithm achieved the best convergence when using a concave inertia weight. Therefore, the inertia weight function of the variable-weight particle swarm optimization algorithm is defined as a concave function: , In the formula, T is the total number of iterations; n is the current number of iterations.

[0038] The primary basis for guiding the search direction in particle swarm optimization (PSO) is fitness; therefore, designing a suitable fitness function is crucial. Generally, the design of the fitness function should comprehensively consider the requirements of the problem itself and the range of its value variation. The sum of the absolute values ​​of the relative errors between the engine model output and the actual engine test data can be used as the standard for evaluating the model matching accuracy, i.e., the objective function. Specifically, the objective function of the variable-weight particle swarm optimization algorithm is...

[0039] In the formula, The calculated value of a certain target performance parameter. The overall test value for a certain target performance parameter. These are the weighting coefficients; The smaller the value, the smaller the difference between the calculated and predicted target performance parameters and the actual values.

[0040] The correction process using the variable-weight particle swarm optimization algorithm is as follows: Figure 3 As shown, the specific steps are as follows: Step 1: First, use the test data from the engine design point to correct (scale) the component characteristic diagram.

[0041] Step 2: Use the modified characteristics to calculate the throttling characteristics, compare the calculated values ​​with the test results of the whole machine, and analyze whether the error meets the engineering requirements.

[0042] Step 3: Construct an objective function using the calculated data and test data.

[0043] Step 4: Then, following the optimization procedure, change the correction factors to obtain the corresponding objective function. Through planned searching, a set of optimal correction factors is finally obtained through optimization. Step 5: Calculate the coupling coefficient at different conversion speeds and correct the component characteristics.

[0044] Optionally, after determining the type of parameter to be optimized in the optimization algorithm, an adjustable range should be set for different parameters to be optimized. If the range is too large, the working efficiency will be reduced, and the model may even fail to converge; if the range is too small, it may be difficult to find the optimal solution. Based on actual engineering experience, the correction factor range is set to [-0.20, +0.20].

[0045] Furthermore, existing and extended flight data are input into the modified personalized component characteristic model to calculate the component characteristics under typical engine operating conditions, primarily referring to the flow rate and efficiency of typical engine components (fan, compressor, high-pressure turbine, and low-pressure turbine). The output values ​​show the deviations of each component parameter from the performance baseline, including efficiency and flow rate deviations. By analyzing these deviations, the degree of performance degradation of engine components can be determined and quantified, and the reasons for the overall engine's airflow performance decline can be explained.

[0046] The following specific examples will provide further explanation. Example

[0047] This embodiment provides a method for evaluating the characteristics of typical aero-engine gas path components based on flight data, which includes the following steps: S1. Taking a certain type of engine as an example, in order to evaluate the degree of deviation of engine component flow and efficiency from the design state using the engine physical mechanism model, and referring to the measurement parameters of ground test bench data, it was determined that five parameters were missing from the flight data (fan outlet bypass pressure P13 and temperature T13, fan outlet internal pressure P23 and temperature T23, and engine thrust Fn) and could not be measured. By establishing a neural network model embedded with physical knowledge, a mapping relationship was established between the input features (inlet total temperature and total pressure, fan speed and guide vane angle, compressor speed and guide vane angle, compressor outlet pressure, throttle lever angle, low-pressure turbine after-temperature and pressure, and exhaust nozzle angle) and the target parameters (fan outlet bypass pressure and temperature, fan outlet internal pressure and temperature, and engine thrust), thereby establishing a general digital twin model.

[0048] Using test data from no fewer than 10 ground test benches, the general digital twin model was trained. After training, the mean error (absolute value) of each target parameter of the general digital twin model during the entire self-test steady-state process was 0.39%, 0.38%, 0.32%, 0.48%, and 0.42%, respectively, all less than 1%. Therefore, the general digital twin model can be considered to have been successfully constructed.

[0049] The neural network nodes and weights of the general digital twin model were adaptively corrected using the test data of a single ground test bench corresponding to the overhauled and qualified product. The ground test bench data was used for verification. The average errors of the steady-state process tests for P13, P23, T13, T23, and Fn were 0.51%, 0.66%, 0.60%, 0.38%, and 0.60%, respectively. Since the steady-state error of the personalized digital twin model is less than 1%, the trained personalized digital twin model is qualified.

[0050] S2. Select flight data of the engine on the route and input it into the general digital twin model that has been trained in a personalized manner to obtain the extended results of the key cross-section parameters of the engine (external bypass pressure P13 and temperature T13 at the fan outlet, internal bypass pressure P23 and temperature T23 at the fan outlet, and engine thrust Fn), and complete the missing gas path performance parameters.

[0051] S3. Using the engine design documents, obtain the engine design point operating parameters and component characteristic diagrams. Then, based on traditional engineering thermodynamic calculations and common engine operating conditions, determine the operating point of each component on the characteristic diagram and establish a general engine component characteristic model. Add correction factors with the approximate and completed engine airflow parameters as the target to correct the component characteristic diagrams, ultimately obtaining a personalized engine component characteristic model, as detailed below.

[0052] The pressure ratio, efficiency, and flow rate of each component (fan, compressor, high-pressure turbine, and low-pressure turbine) of this engine model are used as the component parameters to be corrected. The target performance parameters are the air path performance parameters measured and supplemented by the engine itself, mainly including: high-pressure compressor equivalent speed N2hs, thrust Fn, fuel consumption rate sfc, air flow rate Wa, fuel flow rate Wf, fan internal duct boost ratio πf, compressor boost ratio πc, bypass ratio B, high-pressure turbine inlet temperature Tt4, and low-pressure turbine outlet temperature Tt5.

[0053] Sensitivity analysis was performed using a single-factor perturbation method, specifically as follows: Keeping all other correction factors equal to 1, only one correction factor is changed at a time, increasing it by 1%, and engine performance model calculations are performed to obtain the relative changes in the target performance parameters. The rates of change of each target performance parameter caused by the change in the parameter of a single component to be corrected are summed and normalized to obtain the sensitivity of the target performance parameters to the correction factors. The results are as follows: Figure 4 shown.

[0054] Depend on Figure 4 It can be seen that the main factors affecting the target performance parameters are the flow rate and efficiency of the component parameters. Therefore, the efficiency and flow rate of the fan outside the bypass duct, the fan inside the bypass duct, the compressor, the high-pressure turbine, and the low-pressure turbine are selected as the component parameters to be revised. The high-efficiency global search capability of the variable weight particle swarm algorithm is used to optimize the correction factor.

[0055] The maximum number of iterations for the variable-weight particle swarm optimization algorithm is set to 50, and c1 and c2 are set to 2. In the objective function, the weighting coefficients for each component parameter are as follows: high-pressure compressor equivalent speed N2hs=12, thrust Fn=9, fuel consumption rate sfc=9, airflow rate Wa=5, fuel flow rate Wf=9, fan internal duct boost ratio πf=3, compressor boost ratio πc=3, bypass ratio B=9, high-pressure turbine inlet temperature Tt4=5, and low-pressure turbine outlet temperature Tt5=5.

[0056] The changes in the objective function before and after the correction are as follows: Figure 5 As shown, Figure 5 This study demonstrates a comparison of the objective function values ​​before and after correction at different low-pressure calculated rotational speeds, assuming an altitude H=0, flight speed Ma=0, control law of controlling low-pressure rotational speed, and a constant nozzle throat area. As the flight deviates from the design point, the difference between the general characteristic plot and the actual characteristic plot increases, causing the objective function value before correction to increase as the calculated rotational speed decreases. After correction, the objective function value is significantly reduced, especially above 50% of the calculated rotational speed, and the corrected objective function value remains at a low level.

[0057] Furthermore, with the low-pressure converted speed at 90.23%, the calculation results of the engine target performance parameters before and after correction at the low-pressure converted speed were compared with the errors of the actual test data. The comparison results are shown in Table 3.

[0058] Table 3. Comparison of errors before and after correction for low-pressure converted speed (=90.23%) (%)

[0059] It can be seen that there was a significant deviation between the calculated parameters of the model before correction and the test parameters, especially in the two important indicators of thrust Fn and fuel consumption rate sfc, where the error exceeded 10%. However, after one round of correction, the error was significantly reduced. The vast majority of indicators are now below 2%, and key performance indicators for engines, such as thrust, fuel flow rate, and fuel consumption rate, have been reduced to below 1%. This demonstrates the significant effectiveness of the correction method, ultimately resulting in a personalized engine component characteristic model.

[0060] S4. Input the completed flight route data into the personalized engine component characteristic model described above, and output the deviation values ​​of each component parameter relative to the performance baseline. Since flow rate and efficiency have the greatest impact on the target performance parameters as determined in step S3, they can be used to determine the engine's performance degradation. Therefore, when evaluating the deviation of engine component characteristics from the engine performance baseline, the main focus is on the deviation values ​​of flow rate and efficiency corresponding to typical engine components (fan, compressor, high-pressure turbine, and low-pressure turbine). The evaluation results are shown in Table 4.

[0061] Table 4. Performance evaluation results of engine components (%)

[0062] The evaluation results show that the flow rate and efficiency of the compressor and high-pressure turbine of the engine have large deviations, which indicates that the characteristics of the components have deteriorated significantly.

[0063] Since real component performance degradation requires component-level testing, but the current research process of this invention does not have the necessary testing conditions, the fault inspection data after the engine was disassembled was reviewed to verify this assessment result. It was found that the compressor casing coating of the engine was severely worn, the rotor blade height was shorter than before operation, and the radial clearance between the rotor and stator was increased. The high-pressure turbine guide vane was eroded, and most rotor blade tips were cracked. According to engine principles, these changes will lead to a deterioration in engine component performance, which is consistent with the assessment results obtained by the assessment method of this invention. This indirectly proves that the model can effectively assess the performance degradation of engine airflow components, pointing the way to locating overall engine airflow performance faults or degradation.

[0064] In summary, this invention provides a method for evaluating the characteristics of typical aero-engine gas path components based on flight data. This method first requires establishing two engine simulation models: a general digital twin model based on ground test bench data, and a design-point general component characteristic model based on engine design-point operating data. The general digital twin model is used to complete performance parameters not measured during flight. The completed gas path parameters are then input into the engine design-point general characteristic model, and correction factors are added to approximate the engine test performance parameters, thus correcting the component characteristic diagram and obtaining a personalized engine component characteristic model. This evaluation method can calculate the characteristics of typical engine components, answer the degree of deviation in engine component performance and the causes of engine performance degradation, providing technical support for condition-based engine maintenance in overhaul plants and replacement of engine units on the line.

[0065] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for evaluating the characteristics of typical aero-engine gas path components based on flight data, characterized in that, include: S1. Build a general digital twin model, using parameters that are measured in both test runs and actual flights as input features, and using air path performance parameters that are measured only in test runs and not in actual flights as target features; S2. Input the flight route data into the general digital twin model to complete the missing air path performance parameters; S3. Establish a general component characteristic model based on engine design point data, and use the general component characteristic model to add correction factors to the parameters of each component; S4. Input the completed flight route data into the general component characteristic model and output the deviation values ​​of each component parameter relative to the performance baseline.

2. The evaluation method according to claim 1, characterized in that, Step S1 also includes: The general digital twin model is trained using test data from multiple engines, so that the absolute value of the self-tested steady-state error of the general digital twin model is less than 1%.

3. The evaluation method according to claim 2, characterized in that, Step S1 also includes: The weights of the trained general digital twin model are corrected using factory test data of a single engine, so that the steady-state error of the general digital twin model in testing the single engine is less than 1%, thus completing personalized training.

4. The evaluation method according to claim 1, characterized in that, The input features contain information about engine performance degradation, but the physical relationship between the input features and the target features is not affected by engine performance.

5. The evaluation method according to claim 1, characterized in that, In step S3, the completed air path performance parameters of the engine are used as the target performance parameters. Through sensitivity analysis, the component parameters that are most sensitive to the target performance parameters are selected and correction factors are added.

6. The evaluation method according to claim 5, characterized in that, Step S3 also includes: By using the variable-weight particle swarm optimization algorithm to optimize the added correction factor, the output of the general component characteristic model is aimed at approximating the completed air path parameters of the engine, thus completing the personalized training of the general component characteristic model.

7. The evaluation method according to claim 6, characterized in that, The inertial weight function of the variable-weight particle swarm algorithm is a concave function: ， In the formula, T is the total number of iterations; n is the current number of iterations.

8. The evaluation method according to claim 7, characterized in that, The objective function of the variable-weight particle swarm algorithm is: ， In the formula, The calculated value of a certain target performance parameter. The overall test value of a certain target performance parameter is given. These are the weighting coefficients; The smaller the value, the smaller the difference between the calculated and predicted target performance parameter and the actual value.

9. The evaluation method according to claim 8, characterized in that, The optimization range of the correction factor is -0.2 to +0.

2.

10. The evaluation method according to claim 8, characterized in that, The output shows the deviation values ​​of each component's parameters relative to the performance baseline, including the efficiency deviation value and the flow rate deviation value of each component.

Images