A method for identifying performance of a turboshaft engine component with adaptive performance characteristics

Abstract

The application discloses a kind of self-adapting performance characteristic turboshaft engine component performance identification method, comprising: step one: preparing engine performance baseline model and engine test data;Step two: the relationship between characteristic coupling factor correction coefficient and conversion speed is established;Step three: the turboshaft engine component performance identification model of multiple state points is established;Step four: self-adaptingly determine optimization variable range;Step five: output component performance identification result.The application makes full use of the complementary constraint ability of multiple state point data, while realizing the self-adapting determination of identification boundary, effectively avoids the convergence difficulty or solution set divergence problem caused by artificial trial and error setting range, significantly improves the efficiency and precision of turboshaft engine component performance identification.

Description

Technical Field

[0001] This invention relates to the field of aero-engine performance analysis and component identification methods, specifically to a method for identifying the performance of turboshaft engine components with adaptive performance characteristics. Background Technology

[0002] Overall performance simulation of aero-engines is widely used in fields such as model development, performance evaluation, and health management. Its core lies in establishing characteristic models of components such as the compressor, combustion chamber, and turbine, and obtaining overall engine performance parameters through iteratively solving common working equations. Since accurate component test characteristics are difficult to obtain, component characteristics are usually input into the simulation program in the form of general characteristic diagrams. However, due to factors such as machining tolerances, assembly deviations, and wear and tear, different engines of the same model, under the same conditions, exhibit significant deviations between their actual component characteristics and the general curves. This results in a non-negligible error between the calculation results of the simulation model based on general characteristics and the actual test data from bench tests.

[0003] To improve simulation accuracy, engineering typically employs parameter identification strategies. This involves using overall engine performance parameters (such as exhaust temperature, fuel flow, thrust / power, etc.) measured during actual testing to inversely adjust component characteristics, making the simulation output approximate the actual measured values. Currently, a common practice is to select a single stable operating state as a constraint for component performance identification. However, a single state point can only provide a limited number of independent measurement parameters, while the number of component parameters to be identified is usually large, resulting in insufficient constraints from the measurement parameters on the identification parameters and a high degree of freedom in the identification results. Actual engine testing procedures usually include multiple steady-state operating points, such as idle, intermediate state, and maximum state. Fully utilizing multi-state data can provide complementary constraint information under different operating conditions, theoretically effectively alleviating the problems of single-point identification. To unify the data from various state points, a common technique is to establish a function model of component performance correction coefficients varying with the converted engine speed, thereby transforming the independent correction parameters of each state point into a small number of global coefficients in the function. This significantly reduces the number of unknowns to be identified.

[0004] Nevertheless, the reasonable range of values ​​for these global coefficients (i.e., the search boundaries) varies significantly with the individual differences of each engine. Using a fixed empirical range might be too wide for some engines, leading to low search efficiency or even getting trapped in local maxima; while for others it might be too narrow, causing the true solution to fall outside the boundary and resulting in convergence failure. Currently, setting these search boundaries still heavily relies on technicians to trial and error on a per-engine basis, which is both labor-intensive and cannot guarantee consistency and repeatability.

[0005] Therefore, there is an urgent need to propose a method for identifying the performance of engine components that can effectively integrate test measurement information from multiple state points, and at the same time have the ability to automatically and reasonably determine the search range of optimization variables based on the data characteristics of the engine to be identified, thereby eliminating manual intervention and achieving stable, efficient and high-precision identification processing. Summary of the Invention

[0006] This invention aims to provide an adaptive performance characteristic turboshaft engine component performance identification method. Based on the turboshaft engine performance baseline model, it considers the functional relationship between the component characteristic coupling factor correction coefficient and the converted speed. Through unified identification of component performance at multiple state points and adaptive selection of the range of variable variation, it achieves accurate identification of engine component performance.

[0007] This invention is achieved through the following steps: Step 1: Prepare the engine performance baseline model and engine test data.

[0008] A baseline performance model for a turboshaft engine is established using the component method to calculate the overall engine performance under given operating conditions and component performance conditions.

[0009] Prepare engine test data, including test performance under multiple operating conditions. For each engine, perform preliminary identification of component performance at single-state points under each operating condition to obtain the relative changes in component characteristic coupling factors. The single-state point identification results are used to determine the baseline values ​​and adaptive search range of the unknown coefficients of subsequent functional relationships, and are not used as the final multi-state point identification results.

[0010] Step 2: Establish the relationship between the characteristic coupling factor correction coefficient and the converted speed.

[0011] Considering that the characteristic coupling factor correction coefficients of the flow rate and efficiency of the compressor, gas turbine, and free turbine satisfy a linear or quadratic function relationship with the corresponding component conversion speeds, the following models are established respectively:

[0012] or:

[0013] In the formula, Δ i For the first i Correction coefficients for the coupling factor of individual component characteristics δ i For the first i The relative change of the coupling factor of the characteristics of each component; The slope is to be determined; For the first i Calculation of rotational speed for each component; The intercept is to be determined; The coefficient of the quadratic term; The coefficient of the linear term; For constant terms; i =1,2,…,6 represent compressor flow rate, compressor efficiency, gas turbine flow rate, gas turbine efficiency, free turbine flow rate, and free turbine efficiency, respectively.

[0014] In practical applications, a suitable model should be selected based on the functional relationship between the characteristic coupling factor correction coefficient of the flow rate and efficiency of the compressor, gas turbine, and free turbine and the corresponding component conversion speed.

[0015] Step 3: Establish a performance identification model for turboshaft engine components at multiple state points.

[0016] For linear and quadratic function relationships, the following identification models are established respectively:

[0017] or:

[0018] In the formula, M This represents the number of operating state points of the engine. N This refers to the number of overall engine performance parameters. The first prediction for engine performance model j The first state k One performance parameter; The first measurement for engine test run j The first state k One performance parameter; For the first j In the state of the th , the th k Weighting coefficients for each performance parameter; j =1,2,…, M Indicates the working status sequence number; k =1,2,…, N Indicates the sequence number of the engine's overall performance parameters; i =1,2,…,6 represents the component performance parameter serial number. and Slope k i The lower and upper bounds of the range of variation; and Intercepts The lower and upper bounds of the range of variation; and The coefficients of the quadratic term are respectively The lower and upper bounds of the range of variation; and The coefficients of the linear term are respectively The lower and upper bounds of the range of variation; and These are constant terms. The lower and upper bounds of the range of variation. The same symbol has a consistent meaning throughout this specification.

[0019] The objective functions of equations (3) and (4) are both M Working conditions N A linear combination of the absolute values ​​of the relative deviations between the model-predicted values ​​and the test-measured values ​​of the overall machine performance parameters. This is achieved by adjusting... This can improve the constraints on specific working states or performance parameters during the identification process. A genetic algorithm is used to perform a global search within the upper and lower bounds of the unknown coefficients to obtain the combination of unknown coefficients that minimizes the objective function. Figure 1 The process for identifying engine component performance based on multi-state point performance data is shown.

[0020] Step 4: Adaptively determine the range of optimization variables.

[0021] The optimization variables in equations (3) and (4) are coefficients in the functional relationship model, and it is usually difficult to give a reasonable range of values ​​directly. This invention proposes to use the preliminary identification results of the engine single-state point component performance to fit the benchmark values ​​of the unknown coefficients, and then adjust the acceptable range of the coefficients according to the given characteristic coupling factor to adaptively determine the lower and upper bounds of each unknown coefficient. The following explanation uses a quadratic function model.

[0022] Assume it exists m The characteristic coupling factor correction coefficients of the performance parameters of each component are given, and each correction coefficient satisfies the quadratic function relationship shown in equation (2) with respect to the converted speed. Then, each component performance parameter corresponds to 3 undetermined coefficients, totaling 3. m There are 10 undetermined coefficients. The baseline value of the undetermined coefficients is denoted as:

[0023] In the formula, For the first i The baseline values ​​for the quadratic coefficient, linear coefficient, and constant term corresponding to the performance parameters of each component. i =1,2,…, m With undetermined coefficients For example, its lower and upper bounds are obtained by the following formula:

[0024] In the formula, for The lower bound of the range of variation, for The upper bound of the range of variation, It is the boundary adjustment factor, and satisfies ;a and b The lower and upper bounds are determined according to the same rules.

[0025] The bisection method is used to solve the problem so that the maximum change in the correction coefficient does not exceed the acceptable range. P i The boundary adjustment factors specifically include: 1) For the first i Each characteristic coupling factor correction coefficient, given its acceptable range of variation. ; Set the maximum value of the boundary adjustment factor =1, minimum value =0, initialize the absolute value of the maximum relative error. and initialize the loop step size. .

[0026] 2) Calculation Substitute it into equation (6) to calculate the lower and upper bounds of the undetermined coefficients corresponding to the performance parameters of the i-th component.

[0027] 3) Randomly select M groups of samples within the range of the upper and lower bounds of the undetermined coefficients, and use equation (2) to calculate the correction coefficients for N different conversion speeds to obtain the correction coefficient sample matrix. .

[0028] 4) Calculate the correction coefficient reference value Δ0 for N conversion speeds based on the undetermined coefficient reference value. (i) And calculate the maximum deviation:

[0029] In the formula, for M 1D unit vector For absolute value calculation functions, This is a function for calculating the maximum value.

[0030] 5) Judgment and The size, and updated according to the following formula and :

[0031] 6) Calculation and execute .

[0032] 7) When and Repeat step 2 if the condition is met; otherwise, output the output. .

[0033] 8) Calculate the optimal lower and upper bounds of each unknown coefficient based on the boundary adjustment factor of the output, thereby determining the search range of the genetic algorithm.

[0034] Step 5: Output component performance identification results.

[0035] Based on the determined range of values ​​for the optimization variables, a genetic algorithm is used to perform a global search on the unknown coefficients of the models in equations (3) and (4) to obtain the combination of unknown coefficients that minimizes the objective function. On this basis, according to the relationship model between the characteristic coupling factor correction coefficient of the flow rate and efficiency of the compressor, gas turbine and free turbine and the converted speed, the characteristic coupling factor correction coefficient at a given converted speed is calculated. Further, it is substituted into the performance baseline model of the turboshaft engine to calculate the corresponding component performance and complete the identification of component performance.

[0036] The advantages and beneficial effects of this invention are as follows: 1. This invention constrains component performance parameters with multi-state point whole machine performance measurement data, and reduces the number of parameters to be identified by the functional relationship between the characteristic coupling factor correction coefficient and the converted speed, thereby improving the solvability of the identification problem.

[0037] 2. Based on the overall performance characteristics of individual engines, this invention adaptively determines the search range of the optimization variables of the identification model, which can greatly reduce the manual analysis, trial and error and adjustment work. Attached Figure Description

[0038] Figure 1 This is a flowchart of the engine performance model correction process based on multi-state point performance data. Detailed Implementation

[0039] The technical solution of the present invention will be described below with reference to the accompanying drawings and embodiments. The embodiments are for illustrative purposes only and are not intended to limit the scope of protection of the present invention; any equivalent substitutions or modifications made by those skilled in the art without departing from the concept of the present invention should fall within the scope of protection of the present invention.

[0040] This embodiment uses test data from 110 sets of a certain type of turboshaft engine as the object. Each engine includes test data under five operating conditions: Ngcr=90.0%, 93.5%, 95.5%, 96.5%, and 97.0%. Each condition includes four overall engine performance parameters: P3 / P0, T45kc, Pdnkc, and SFCkc.

[0041] Implementation Step 1: First, establish a performance baseline model for the turboshaft engine and use the average performance values ​​of 100 engines at multiple state points to complete the performance baseline correction.

[0042] Test data were collected for each engine under five operating conditions: Ngcr = 90.0%, 93.5%, 95.5%, 96.5%, and 97.0%. For each operating condition, the relative changes in the characteristic coupling factors of the compressor, gas turbine, and free turbine for each engine under the five conditions were initially obtained through single-condition model identification.

[0043] Step 2: Based on the single state point, the preliminary identification results are fitted using the linear function relationship shown in Equation (1) and the quadratic function relationship shown in Equation (2) to obtain the baseline values ​​of each unknown coefficient. For the linear function relationship, each engine contains 12 unknown coefficients; for the quadratic function relationship, each engine contains 18 unknown coefficients.

[0044] Implementation Step 3: Set the acceptable variation range P of the characteristic coupling factor correction coefficient. i =0.02, meaning the correction coefficient variation range does not exceed ±2%. Following the method in step four, the boundary adjustment factor is calculated for the performance parameters of each engine and each component, and the upper and lower bounds of the unknown coefficients required by the genetic algorithm are formed accordingly.

[0045] Step 4: Solve equations (3) and (4) using a genetic algorithm. The population size is set to 120 for linear functions and 180 for quadratic functions. The maximum number of iterations is set to 25 generations for both.

[0046] Step 5: Compare the identification results obtained from linear and quadratic function relationships. Using the requirement that the relative deviation of each performance parameter under five operating states for each engine does not exceed 1%, 95 engines were determined to use quadratic function relationships, while the remaining 15 engines used linear function relationships. The relative deviation between the predicted multi-state performance values ​​and the actual test performance values ​​for all engines was within ±1%.

[0047] The identification results of 110 engines were statistically analyzed, and the maximum relative deviation and average relative deviation of the four overall engine performance parameters under various operating conditions are shown in Table 1.

[0048] Table 1. Performance Identification and Verification Results of Engine Components

[0049] As can be seen from the table above, the maximum relative deviation of each performance parameter under all working conditions is less than 1%, and the average relative deviation is less than 0.36%, indicating that the present invention can maintain high component performance identification accuracy under large sample conditions.

Claims

1. A method for performance identification of turboshaft engine components based on adaptive performance characteristics, characterized in that, Includes the following steps: Step 1: Prepare the engine performance baseline model and engine test data; Step 2: Establish the relationship between the characteristic coupling factor correction coefficient and the converted speed; Step 3: Establish a performance identification model for turboshaft engine components at multiple state points; Step 4: Adaptively determine the range of optimization variables; Step 5: Output component performance identification results.

2. The method for performance identification of turboshaft engine components based on adaptive performance characteristics according to claim 1, characterized in that: Step one further includes: Step 2.1 Establish a baseline performance model of the turboshaft engine using the component method to calculate the overall engine performance under given operating conditions and component performance conditions; Step 2.2 Prepare engine test data, which should include test performance under multiple operating conditions. Perform preliminary identification of component performance at single-state points for each operating condition of each engine to obtain the relative change of component characteristic coupling factors. The single-state point identification results are used to determine the benchmark value and adaptive search range of the unknown coefficients of the subsequent functional relationship, and are not used as the final multi-state point identification results.

3. The method for performance identification of turboshaft engine components based on adaptive performance characteristics according to claim 1 or 2, characterized in that: The linear function relationship between the characteristic coupling factor correction coefficient and the converted speed is as follows: ； The quadratic function relationship between the characteristic coupling factor correction coefficient and the converted speed is: ； In the formula, Δ i δ is the correction coefficient for the coupling factor of the i-th component characteristic. i Let be the relative change in the coupling factor of the i-th component characteristic. The slope is undetermined. Calculate the rotational speed for the i-th component. The intercept is to be determined. The coefficient of the quadratic term, The coefficient of the linear term, For constant terms, i =1,2,…,6 represent compressor flow rate, compressor efficiency, gas turbine flow rate, gas turbine efficiency, free turbine flow rate, and free turbine efficiency, respectively.

4. The method for performance identification of turboshaft engine components with adaptive performance characteristics according to claim 3, characterized in that: When using a linear function relationship, the component performance identification model is as follows: ； In the formula, M This represents the number of operating state points of the engine. N This refers to the number of overall engine performance parameters. The first prediction for engine performance model j The first state k One performance parameter; The first measurement for engine test run j The first state k One performance parameter; For the first j In the state of the th , the th k Weighting coefficients for each performance parameter; j =1,2,…, M Indicates the working status sequence number; k =1,2,…, N Indicates the sequence number of the engine's overall performance parameters; i =1,2,…,6 represents the component performance parameter serial number; and Slope k i The lower and upper bounds of the range of variation; and Intercepts The lower and upper bounds of the range of variation; and The coefficients of the quadratic term are respectively The lower and upper bounds of the range of variation; and The coefficients of the linear term are respectively The lower and upper bounds of the range of variation; and These are constant terms. The lower and upper bounds of the range of variation.

5. The method for performance identification of turboshaft engine components with adaptive performance characteristics according to claim 3, characterized in that: When using a quadratic function relationship, the component performance identification model is as follows: ； In the formula, M This represents the number of operating state points of the engine. N This refers to the number of overall engine performance parameters. The first prediction for engine performance model j The first state k One performance parameter; The first measurement for engine test run j The first state k One performance parameter; For the first j In the state of the first k Weighting coefficients for each performance parameter; j =1,2,…, M Indicates the working status sequence number; k =1,2,…, N Indicates the sequence number of the engine's overall performance parameters; i =1,2,…,6 represents the component performance parameter serial number; and Slope k i The lower and upper bounds of the range of variation; and Intercepts The lower and upper bounds of the range of variation; and The coefficients of the quadratic term are respectively The lower and upper bounds of the range of variation; and The coefficients of the linear term are respectively The lower and upper bounds of the range of variation; and These are constant terms. The lower and upper bounds of the range of variation.

6. The method for performance identification of turboshaft engine components with adaptive performance characteristics according to claim 3, characterized in that: In step four, the process for adaptively determining the range of optimization variables is as follows: Suppose there exists m The characteristic coupling factor correction coefficients of the performance parameters of each component are given, and each correction coefficient satisfies the quadratic function relationship shown in equation (2) with respect to the converted speed. Then, each component performance parameter corresponds to 3 undetermined coefficients, totaling 3. m One undetermined coefficient; The baseline value of the undetermined coefficients is denoted as: ； In the formula, For the first i The baseline values ​​for the quadratic coefficient, linear coefficient, and constant term corresponding to the performance parameters of each component. i =1,2,…, m ; Undetermined coefficients The lower and upper bounds are obtained by the following formula: ； In the formula, for The lower bound of the range of variation for The upper bound of the range of variation, It is the boundary adjustment factor, and satisfies ; a and b The lower and upper bounds are determined according to the same rules.

7. The method for performance identification of turboshaft engine components with adaptive performance characteristics according to claim 6, characterized in that: The bisection method is used to solve the problem so that the maximum change in the correction coefficient does not exceed the acceptable range. P i The boundary adjustment factors specifically include: Step 1) For the first i Each characteristic coupling factor correction coefficient, given an acceptable range of variation. ; set the maximum value of the boundary adjustment factor =1, minimum value =0, initialize the absolute value of the maximum relative error. and initialize the loop step size. ; Step 2) Calculation Substitute into equation (6) to calculate the lower and upper bounds of the undetermined coefficients corresponding to the performance parameters of the i-th component; Step 3) Randomly select M groups of samples within the range determined by the upper and lower bounds of the undetermined coefficients, and use Equation (2) to calculate the correction coefficients for N different conversion speeds, thus obtaining the correction coefficient sample matrix. ; Step 4) Calculate the correction coefficient reference value Δ0 for N conversion speeds based on the undetermined coefficient reference value. (i) And calculate the maximum deviation: ； In the formula, for M 1D unit vector For absolute value calculation functions, This is a function for calculating the maximum value. Step 5) Judgment and The size, and updated according to the following formula and : ； Step 6) Calculation and execute ; Step 7) When and Repeat step 2 if necessary; otherwise, output... ; Step 8) Calculate the optimal lower and upper bounds of each unknown coefficient based on the boundary adjustment factor of the output, thereby determining the search range of the genetic algorithm.

8. The method for performance identification of turboshaft engine components with adaptive performance characteristics according to claim 1, characterized in that: In step five, based on the determined range of values ​​for the optimization variables, a genetic algorithm is used for a global search to obtain the combination of unknown coefficients that minimizes the objective function. On this basis, according to the relationship model between the characteristic coupling factor correction coefficient of the flow rate and efficiency of the compressor, gas turbine, and free turbine and the converted speed, the characteristic coupling factor correction coefficient at a given converted speed is calculated. This coefficient is then substituted into the turboshaft engine performance baseline model to calculate the corresponding component performance, thus completing the identification of component performance.

Images