Aero-engine performance model correction method, deployment method and correction system

Abstract

The invention discloses an aero-engine performance model correction method, deployment method and aero-engine performance model correction system, and belongs to the technical field of aero-engines. And correcting the constructed engine performance model based on a PSO (Particle Swarm Optimization) algorithm, wherein the inertia weight of particles in the PSO algorithm is a function of the number of iterations, the particle fitness value and the population fitness value. The inertia weight in the PSO algorithm is defined as a function which is dynamically adjusted along with the number of iterations and the fitness of the algorithm, so that the inertia weight value is adaptively adjusted in the iteration process, the solving performance of the PSO algorithm is improved, and the adaptability of the algorithm to a new data set is enhanced.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of aero-engines, and particularly relates to an aero-engine performance model correction method, a deployment method and a correction system. BACKGROUND

[0002] Modern aero-engines are strong nonlinear, complex, multivariable and time-varying systems. Due to the harsh working environment of the engine, the safety and reliability of the engine are required to be high. The actual engine performance will change in the entire life cycle due to the combined influence of external use environment differences, engine performance degradation during use, manufacturing and installation tolerances and other factors, and there are performance differences between different engine individuals. At this time, the engine performance reflected by the engine mathematical model constructed according to the design state of the component characteristics deviates from the actual characteristics of the real engine. Therefore, it is of important engineering application value to establish an engine performance model with high adaptive ability to adapt to various actual running conditions of the engine, so as to realize the advanced health management state monitoring and fault diagnosis technology of the aero-engine.

[0003] There are mainly two types of existing engine model correction methods. One is the most commonly used in engineering, which uses professional engine model design tools such as Gasturb, relies on the ability and engineering experience of the design personnel, and manually adjusts the engine characteristic data to meet the design and test data requirements. The other is that research institutions such as colleges define a group of component characteristic correction coefficients, directly or indirectly take the error sum of the model simulation output and the test data as the objective function, and calculate the component characteristic correction coefficients of the engine model through optimization algorithms, machine learning and Kalman filtering algorithms. For example, CN113569319A discloses an aero-engine model self-adaptive corrector design method based on PSO_DE intelligent algorithm. However, the inertia weight in the PSO algorithm is a fixed value, which needs to be adjusted multiple times, and the generalization ability is poor when applied to different data sets.

[0004] APPLICATION CONTENT The purpose of the present application is to provide an aero-engine performance model correction method to solve the problems raised in the background art.

[0005] To achieve the above purpose, the present application provides the following technical solution: an aero-engine performance model correction method, comprising: constructing an engine performance model; correcting the constructed engine performance model based on a PSO optimization algorithm, wherein the inertia weight of the particle in the PSO optimization algorithm is a function of the iteration number, the particle fitness value and the population fitness value.

[0006] Further, the inertia weight of the particle in the PSO algorithm decreases with the increase of the iteration number.

[0007] Further, the inertia weight function of the particle in the PSO optimization algorithm is as follows:

[0008] In the formula, represents the inertia weight value of the i-th particle in the j-th iteration, and are the maximum and minimum values of the inertia weight, is the total number of iterations set by the algorithm, and represent the maximum and minimum values of the fitness value of the population particle in the j-th iteration, represents the fitness value of the i-th particle in the j-th iteration.

[0009] Further, the correction of the engine performance model based on the PSO optimization algorithm includes: initializing the initial position and speed of the particle of the PSO optimization algorithm; evaluating the particle; judging whether the optimization target is reached based on the evaluation result of the particle, outputting the optimized particle position when the optimization target is reached, and correcting the engine performance model based on the optimized particle position.

[0010] Further, the evaluation of the particle includes: obtaining the engine characteristic correction coefficient based on the particle position, and substituting the engine characteristic correction coefficient into the engine performance model; calculating the deviation between the simulation result of the engine performance model and the actual output of the real engine, and evaluating the particle through the fitness function and the penalty function.

[0011] Further, the correction of the engine performance model based on the PSO optimization algorithm further includes: updating the particle position and speed when the evaluation result of the particle does not reach the optimization target; judging whether the maximum number of iterations is reached, outputting the optimized particle position when the maximum number of iterations is reached, and correcting the engine performance model based on the optimized particle position.

[0012] The application also discloses an engine performance model deployment method, comprising: constructing an engine performance model; correcting the constructed engine performance model based on the PSO optimization algorithm, the inertia weight of the particle in the PSO optimization algorithm being a function of the number of iterations, the fitness value of the particle, and the fitness value of the population; and ​​​​​The modified engine performance model is cross-compiled to generate a dynamic link library file, and then deployed to the airborne platform.

[0013] Furthermore, the method also includes extracting engine characteristic data into a characteristic configuration file.

[0014] Furthermore, the method also includes revising the engine performance model based on the PSO optimization algorithm, uploading the revised engine characteristic data to the airborne platform, and replacing the characteristic configuration file.

[0015] This application also discloses an aero-engine performance model correction system, including: A building module configured to build an engine performance model; The correction module is configured to correct the constructed launch performance model based on the PSO optimization algorithm, in which the particle inertia weight is a function of the iteration number, particle fitness value, and population fitness value.

[0016] Compared with the prior art, the beneficial effects of this application are: This application defines the inertia weight in the PSO algorithm as a function that is dynamically adjusted with the number of algorithm iterations and fitness, so as to adaptively adjust the inertia weight value during the iteration process, thereby improving the solution performance of the PSO optimization algorithm and enhancing the algorithm's adaptability to new datasets. Attached Figure Description

[0017] Figure 1 This is a flowchart of the method in this application; Figure 2 Revise the flowchart for the model; Figure 3 This is a schematic diagram illustrating model correction and deployment. Figure 4 A schematic diagram for constructing an engine performance model; Figure 5 The curve of the penalty term expression. Detailed Implementation

[0018] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0019] A method for correcting an aero-engine model, referring to Figure 1 ,include: S100: Constructing an engine performance model; S200: The constructed launch performance model is modified based on the PSO optimization algorithm. In the PSO optimization algorithm, the particle inertia weight is a function of the iteration number, particle fitness value, and population fitness value. Specifically, in step S100, an engine performance model (e.g., an engine component-level whole-machine performance model) can be established based on the engine design structure and component design characteristics. In particular, the principles and methods for modeling existing engine performance models are relatively mature; for example, referring to… Figure 4 The main approach is based on the principles of aerodynamics and thermodynamics, which studies the characteristics of each component, establishes its mathematical model, and then, based on the configuration characteristics and the matching and coordination constraints between the components, constructs a whole-machine-level mathematical model. Finally, appropriate mathematical methods are used to solve the problem, which will not be elaborated here. Specifically, in step S200, after the engine performance model is established, the engine performance model is corrected based on the improved PSO optimization algorithm so that the simulation error results of the engine model conform to the real characteristics of the individual engine, thus forming the engine's unique characteristic model. Specifically, referring to... Figure 2 The modification of the engine performance model includes the following steps: S201: Initialization: Randomly generate the initial position and flight speed of particles with a population size of M1 within the range given by the control parameters, and set the particle dimension to M2 and the generation to n; S202: Evaluate the particles. Specifically, the engine components of the engine model are corrected as follows: (1) (2) in This represents the flow correction factor for engine components. This indicates the value before flow correction for engine components. This indicates the corrected flow rate value for engine components. This represents the efficiency correction factor for engine components. This indicates the values ​​before engine component efficiency correction. This indicates the adjusted values ​​for engine component efficiency.

[0020] The above correction coefficient and Determined by the following formula: (3) (4) Since the higher-order terms in the above formula have relatively small numerical impact, and considering that the optimization parameters should be minimized, only the quadratic terms are retained to obtain the correction coefficients. and The calculation formula is as follows: (5) (6) In the formula , , , , and For undetermined coefficients, and These are represented as non-design point speed and design point speed, respectively. This indicates the position of the speed line in the characteristic diagram.

[0021] In addition, the combustion chamber efficiency and total pressure recovery coefficient can be corrected.

[0022] (7) (8) , , and For undetermined coefficients, and These are represented as non-design point speed and design point speed, respectively.

[0023] As can be seen from the above description, the process of determining the correction parameters for the characteristics (e.g., efficiency and flow rate) of engine components (e.g., compressors, gas turbines, and power turbines) can be converted into determining undetermined coefficients. , , , , and as well as , , and The determination process, correspondingly, involves the PSO algorithm's dimensions being determined by the number of engine components and the undetermined coefficients. , , , , and as well as , , and Determined, that is, the undetermined coefficients of correction factors for multiple engine components. , , , , and as well as , , and These together constitute the parameters to be optimized (i.e., variables) in the PSO algorithm. For example, taking the compressor pressure ratio correction and flow rate correction as an example, the variables in the PSO algorithm include the compressor flow rate correction coefficient and the principal coefficient (i.e.,...). ), compressor flow correction coefficient sub-coefficient one (i.e. ), Compressor flow correction coefficient sub-coefficient two ( ), compressor pressure ratio correction factor (main factor) ), compressor pressure ratio correction factor sub-coefficient one ( ) and compressor pressure ratio correction factor sub-factor two ( ).

[0024] Based on the above steps, the undetermined coefficients of each engine characteristic correction coefficient can be determined. Combining equations (5) to (8), the characteristic correction coefficients of each engine can be obtained. Substituting the engine characteristic correction coefficients into the engine performance model established above, under the same flight conditions and input parameters, the deviation between the simulation results of the engine performance model and the actual output of the real engine is calculated. The quality of the evaluation particles is calculated through the fitness function and the penalty function. The fitness function is the calculation function used to evaluate the quality of particles in the PSO optimization algorithm. Its selection directly affects the convergence speed and optimization effect of the algorithm. This application establishes the fitness function through the error between the simulation data and test data of the engine model measurement parameters. The specific steps are as follows: Define engine model number 1 The error of each measurement parameter is (9) In the formula, Represented as target parameter, subscript Represented as model simulation results, subscript Represented as baseline data, N is the number of measurement parameters.

[0025] This application uses the maximum fitness value as the optimization objective. The higher the fitness of a particle, the closer it is to the expected target; conversely, the lower the fitness of a particle, the closer it is to the expected target. All particles will move closer to the position of the particle with higher fitness. A fitness function is defined. as follows: (10) According to the fitness function, the greater the fitness of a particle, the smaller the relative deviation between all target parameters of the engine model and the baseline value, and the higher the model accuracy.

[0026] For model correction, the main technical objective is to ensure that the simulated value of any target parameter has a small error compared to the actual value. However, model correction methods based on optimization algorithms often have a problem: after optimization, the simulation results of most target parameters of the engine model are very close to the actual values, but one or a few target parameters have a large simulation error. Although this achieves an overall reduction in error, such optimization results do not meet the actual engineering requirements. Therefore, it is preferable to sacrifice some simulation accuracy of individual target parameters to ensure that all target parameters are within a small error range. To address this issue, this application adds a penalty constraint to the basic PSO optimization algorithm to reduce the likelihood of this problem occurring.

[0027] Specifically, this application establishes a fitness function for an improved PSO optimization algorithm based on the deviation between simulation data and test data of N measurement parameters. Furthermore, to ensure that all measurement parameters remain within the accuracy range, this application introduces a penalty term into the fitness function. : (10) (11) In the formula, the penalty term The expression curve is as follows Figure 5 As shown: Depend on Figure 5 It can be seen that the penalty item The larger the simulation error, the larger the penalty term, approaching 2; conversely, the smaller the simulation error, the smaller the penalty term, approaching 1. This penalty function helps prevent situations where individual target parameters have large simulation errors.

[0028] While evaluating the particles, the best individual particle of the current generation is recorded and compared with the best particle in history. If the best particle in the current generation is better than the best particle in history, the best particle in history is replaced with the best particle in the current generation.

[0029] S203: Determine whether the optimization goal has been achieved. Compare the current best particle with the set evaluation index. If the index is met, jump to step S206; otherwise, jump to step S204. Subsequently, the inertia weight function is set. Specifically, the inertia weight in this application is a function of the number of iterations, the particle fitness value, and the population fitness value. The specific particle inertia weight function is as follows: (1) In the formula, Representing the In the nth iteration The inertial weight value of each particle. and These are the maximum and minimum values ​​of the inertia weight, which can be taken as 0.4 and 0.9 according to general engineering experience. This is the total number of algorithm iterations set. and Representing the The maximum and minimum values ​​of the fitness of the population particles in each iteration. Representing the In the nth iteration The fitness value of each particle. As shown in the formula, the inertia weight decreases with increasing iteration count. Simultaneously, individual particles are also affected by their own fitness and the population fitness value. The closer the individual particle's fitness is to the optimal fitness of the population particles, the smaller the inertia weight, thus enhancing the particle's ability to continue searching more precisely along its original trajectory; conversely, it enhances the particle's ability to change its original search trajectory and find new paths. The improved PSO optimization algorithm achieves faster convergence and higher solution efficiency, reducing the time spent repeatedly debugging algorithm parameters. It is more suitable for deployment and application in aero-engine health management systems.

[0030] S204: Update velocity and position, i.e. calculate the inertial weight function value and update the particle's flight velocity and position at the next moment; S205: Determine whether the set algebra n has been reached. If the maximum algebra has been reached, jump to step S202; otherwise, jump back to step S206. S206: Output the best particle, outputting the position and fitness of the historical best particle.

[0031] This application also discloses an aero-engine performance model correction system, characterized in that it includes: A building module configured to build an engine performance model; The correction module is configured to correct the constructed launch performance model based on the PSO optimization algorithm, in which the particle inertia weight is a function of the iteration number, particle fitness value, and population fitness value.

[0032] This application also discloses a method for deploying an engine performance model, referring to... Figure 3 ,include: Construct an engine performance model; The constructed launch performance model is modified based on the PSO optimization algorithm. In the PSO optimization algorithm, the particle inertia weight is a function of the iteration number, particle fitness value, and population fitness value; and The modified engine performance model is cross-compiled to generate a dynamic link library file, and then deployed to the airborne platform.

[0033] Specifically, in the deployment of the engine performance model, firstly, based on the operating system of the target airborne health management and monitoring equipment platform (hereinafter referred to as the airborne platform), a specific cross-compilation tool is used to cross-compile the engine performance model after characteristic modification, generating a dynamic link library file for the target operating system. Then, through the communication interface between the airborne platform and the ground maintenance equipment, the modified engine performance model is deployed to the airborne platform. By recording the function interfaces and dynamic link library file calling patterns of the platform application software, the airborne online operation of the engine model is achieved. Considering the portability and security in the actual application of the aero-engine health management system, modifications and changes at the software source code level are avoided to a minimum unless absolutely necessary. In this application, engine characteristics (compressor component characteristic data, gas turbine component characteristic data, and power turbine component characteristic data) are extracted into specific files and loaded and managed in the form of configuration files. The external files defined in this application are shown in the table below. Simultaneously, optimization process information is recorded to facilitate tracking and analysis by designers.

[0034] Table 1 Description of Externally Called Files

[0035] In some embodiments, the method further includes updating the engine performance model. Specifically, as engine operating time increases, engine performance gradually changes. When the engine has been running for a specified period or when the engine performance has degraded to a certain extent as assessed by the designers (measured by the deviation between the actual measured values ​​of engine characteristic parameters and the output values ​​of the health monitoring performance model), the initially deployed airborne engine model will deviate significantly from the actual engine performance data, requiring adaptive correction. Based on the latest engine operating data, the designers correct the engine characteristics again on the ground according to step S2. The update deployment only requires uploading the corrected engine characteristic data configuration file to the airborne platform, replacing the old characteristic configuration file. No modifications to the program are required, ensuring the safety of the upgrade operation and greatly simplifying the operational complexity for engineers. This is suitable for the effective application of the aero-engine health management system during the use and maintenance phase.

[0036] Although embodiments of this application have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and variations can be made to these embodiments without departing from the principles and spirit of this application, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A method for correcting an aero-engine performance model, characterized in that, include: Construct an engine performance model; The constructed launch performance model is modified based on the PSO optimization algorithm. In the PSO optimization algorithm, the inertia weight of the particle is a function of the number of iterations, the particle fitness value, and the population fitness value.

2. The method for correcting an aero-engine performance model according to claim 1, characterized in that: In the PSO algorithm, the inertia weight of a particle decreases as the number of iterations increases.

3. The method for correcting an aero-engine performance model according to claim 1, characterized in that: The particle inertia weight function in the PSO optimization algorithm is as follows: In the formula, Representing the In the nth iteration The inertial weight value of each particle. and These are the maximum and minimum values ​​of the inertia weight. This is the total number of algorithm iterations set. and Representing the The maximum and minimum values ​​of the fitness of the population particles in each iteration. Representing the In the nth iteration The fitness value of each particle.

4. The method for correcting an aero-engine performance model according to claim 1, characterized in that: The engine performance model is modified based on the PSO optimization algorithm, including: Initialize the initial position and velocity of the particles in the PSO optimization algorithm; Evaluate the particles; The optimization target is determined based on the particle evaluation results. When the optimization target is achieved, the particle position output is optimized, and the engine performance model is corrected based on the optimized particle position.

5. The method for correcting an aero-engine performance model according to claim 4, characterized in that: The evaluation of particles includes: Engine characteristic correction coefficients are obtained based on particle positions, and these coefficients are then substituted into the engine performance model. The deviation between the simulation results of the engine performance model and the actual output of the real engine is calculated, and the particles are evaluated by calculating the fitness function and the penalty function.

6. The method for correcting an aero-engine performance model according to claim 4, characterized in that: The modification of engine performance models based on the PSO optimization algorithm also includes: When the particle evaluation results do not meet the optimization target, the particle position and velocity are updated. Determine if the maximum generation has been reached. If the maximum generation has been reached, output the optimized particle position and correct the engine performance model based on the optimized particle position.

7. A method for deploying an engine performance model, characterized in that, include: Construct an engine performance model; The constructed launch performance model is modified based on the PSO optimization algorithm. In the PSO optimization algorithm, the inertia weight of the particle is a function of the number of iterations, the particle fitness value, and the population fitness value. as well as The modified engine performance model is cross-compiled to generate a dynamic link library file, and then deployed to the airborne platform.

8. The method for deploying an engine performance model according to claim 7, characterized in that: The method also includes extracting engine characteristic data into a characteristic configuration file.

9. The method for deploying an engine performance model according to claim 8, characterized in that: The method also includes further revising the engine performance model based on the PSO optimization algorithm, uploading the revised engine characteristic data to the airborne platform, and replacing the characteristic configuration file.

10. A performance model correction system for an aero-engine, characterized in that, include: A building module configured to build an engine performance model; The correction module is configured to correct the constructed launch performance model based on the PSO optimization algorithm, in which the particle inertia weight is a function of the iteration number, particle fitness value, and population fitness value.

Images