A method for identifying aerodynamic parameters of an airplane in level flight and acceleration and deceleration and performance evaluation thereof

Abstract

The application discloses a kind of aircraft level flight acceleration and deceleration aerodynamic parameter identification and its performance evaluation method, comprising: S1, based on the mass m of aircraft, thrust T and aerodynamic resistance D establishes the dynamics model of aircraft level flight acceleration and deceleration, determines the flight parameter required for aerodynamic parameter identification;S2, pre-processing is carried out to flight parameter;S3, construct aerodynamic derivative influence matrix and resistance coefficient vector;S4, based on aerodynamic derivative influence matrix and resistance coefficient vector, combined with bayesian identification module, obtains aerodynamic resistance coefficient;S5, based on the aerodynamic resistance coefficient, combined with dynamics model, evaluates flight performance.The application carries out sampling rate correction, outlier rejection and filtering processing to the flight parameter required for aerodynamic parameter identification, improves the accuracy of flight parameter, and then uses bayesian parameter identification method to obtain aerodynamic resistance, calculates the flight performance of aircraft level flight acceleration and deceleration, and improves the accuracy of flight performance calculation.

Description

Technical Field

[0001] This invention relates to the field of flight performance evaluation technology, and in particular to a method for identifying aerodynamic parameters of aircraft acceleration and deceleration during level flight and evaluating their performance. Background Technology

[0002] Parameter identification methods are broadly classified into time-domain and frequency-domain methods. In the time-domain approach, maximum likelihood estimation is a classic technique. Iliff and Maine achieved high-precision parameter estimation by combining maximum likelihood estimation with Newton-Raphson optimization. Kalman filtering techniques have gained attention due to their applicability to nonlinear systems. Garcia-Velo used extended Kalman filtering to estimate parameters for F / A-1 high angle-of-attack flight data. Yougsu... Han et al. compared two methods for online aircraft parameter identification: the frequency domain equation error method and the time domain Bayesian method. They compared the methods in terms of convergence and accuracy. Simulation results showed that the Bayesian method had better convergence, and its standard error was smaller than that of the method error method. The Bayesian estimation method could quickly converge to the true value, demonstrating that it could achieve accurate and fast online parameter estimation. In the frequency domain, the recurrent Fourier transform technique developed by Klein and Morelli achieved real-time identification with low computational cost through frequency domain least squares regression. Neural networks were also introduced into parameter identification. Linse used a feedforward neural network to estimate aerodynamic coefficients, avoiding the complexity of traditional modeling, while Raisinghani further verified the effectiveness of neural networks in torque derivative identification.

[0003] In domestic research, Lu Xingju et al. achieved online identification of the state-space model of aircraft using the recursive least squares method, providing a solution for real-time estimation of maneuvering flight parameters. Wei Aiyu et al. proposed a recursive least squares method based on a forgetting factor to address the aerodynamic coupling problem in helicopter maneuvering flight, effectively solving the parameter mutation problem under strong coupling conditions. Regarding parameter identification methods, Wu Zehua et al., addressing the high uncertainty of aerodynamic characteristics during aircraft reentry, proposed an offline aerodynamic parameter identification method based on a Bayesian framework using non-parametric modeling. They also used offline flight data to obtain the relationship between aerodynamic coefficients and aircraft state variables. Furthermore, to address the inability of offline methods to reflect aerodynamic changes during reentry in a timely manner, they optimized the offline identification method and proposed an online aerodynamic parameter identification method based on a Bayesian framework, obtaining relatively accurate identification results. Cui Naigang et al. applied Kalman filtering to the aerodynamic parameter identification of reentry vehicles, solving the noise interference problem in high-dynamic environments. In recent years, Zhang Jiaming et al. explored machine learning-driven intelligent correction methods for aerodynamic parameters, providing new ideas for identification under complex operating conditions.

[0004] Based on the current research status and development trends at home and abroad, aerodynamic parameter identification methods using the Bayesian framework have many advantages. This application selects Bayesian theory to conduct aerodynamic parameter identification research, aiming to solve the problems of low iteration efficiency, easy divergence of results, and insufficient accuracy of traditional methods. By combining the Bayesian method with flight test data, it explores a way to accurately and quickly extract real aerodynamic characteristics. The research results can not only provide more reliable theoretical support for aircraft dynamics modeling, but also provide technical reference for shortening the flight test cycle and reducing R&D costs, which has important engineering application value. Summary of the Invention

[0005] The purpose of this invention is to overcome the shortcomings of the prior art and provide a method for identifying aerodynamic parameters and evaluating the performance of aircraft during level flight acceleration and deceleration. By simulating engine thrust, flight drag, and aircraft mass, the flight parameters required for identifying aerodynamic parameters during level flight acceleration and deceleration are determined. The flight parameters are then subjected to sampling rate correction, outlier removal, and filtering to improve their accuracy. Finally, the aerodynamic drag is obtained using the Bayesian parameter identification method, and the flight performance of the aircraft during level flight acceleration and deceleration is calculated, thereby improving the accuracy of flight performance calculation.

[0006] To achieve the above objectives, the technical solution adopted by this invention is: a method for identifying aerodynamic parameters and evaluating the performance of an aircraft during level flight acceleration and deceleration, comprising the following steps: Step S1, establishing a dynamic model of the aircraft's level flight acceleration and deceleration based on the aircraft's mass m, thrust T, and aerodynamic drag D, and determining the flight parameters required for aerodynamic parameter identification; Step S2, preprocessing the flight parameters described in Step S1; Step S3, constructing an aerodynamic derivative influence matrix based on the preprocessed flight parameters. and drag coefficient vector Step S4: Based on the aerodynamic derivative influence matrix described in step S3 and drag coefficient vector Step S5: Based on the aerodynamic drag coefficient described in step S4 and the dynamic model described in step S1, evaluate the flight performance of the aircraft during level flight acceleration and deceleration.

[0007] Preferably, the mass m of the aircraft in step S1 is expressed by the following formula:

[0008]

[0009] In the formula: Empty weight of the aircraft; For weapon weight; This refers to the weight of the fuel. Other weight on the aircraft;

[0010] The thrust It can be expressed as follows:

[0011]

[0012] In the formula: The total thrust of the two engines; and These are the maximum thrust of the two engines, respectively. and This is the throttle position coefficient; and This is the rotor speed coefficient; and To correct the parameters;

[0013] The aerodynamic drag D is expressed by the following formula:

[0014]

[0015] In the formula: The drag coefficient of the entire aircraft without rudder deflection. , , These correspond to the drag increment coefficients caused by a unit angle of deflection of the horizontal stabilizer, aileron, and rudder, respectively. , , These correspond to the deflection angles of the horizontal stabilizer, aileron, and rudder, respectively. air density, Let be the airspeed at time i. , , , These represent the areas of the wings, horizontal stabilizer, ailerons, and rudder, respectively.

[0016] Preferably, the dynamic model of the aircraft's acceleration and deceleration is expressed by the following equation:

[0017] In the formula: V pred For predicting speed;

[0018] Based on the dynamic model, the influencing factors of each parameter in the dynamic model are analyzed. The flight parameters required for aerodynamic parameter identification include air pressure, temperature, airspeed, and mass.

[0019] Preferably, the preprocessing of flight parameters required for aerodynamic parameter identification in step S2 includes sampling rate correction, outlier removal, and smoothing filtering. The sampling rate correction converts flight parameters with different sampling frequencies and periods into data with the same sampling period. The outlier removal removes abnormal data by setting an effective range for the flight parameters. The smoothing filtering is a 15-point moving average filter.

[0020] Preferably, the aerodynamic derivative influence matrix The expression is:

[0021]

[0022] In the formula: ;

[0023] The drag coefficient vector The expression is:

[0024] .

[0025] Preferably, the Bayesian identification module includes prior distribution modeling, likelihood function construction, and posterior distribution calculation; the prior distribution modeling adopts a multivariate Gaussian distribution, expressed by the following formula:

[0026]

[0027] In the formula: It is the drag coefficient vector to be identified. ; It is a diagonal covariance matrix. .

[0028] Preferably, the likelihood function is expressed by the following formula:

[0029]

[0030] In the formula: It is the i-th row of the design matrix. These are the heteroscedasticity weights related to dynamic pressure;

[0031] The posterior distribution is expressed by the following formula:

[0032]

[0033] In the formula: ; .

[0034] Preferably, the evaluation of the aircraft's level flight acceleration and deceleration performance requires correction of the aircraft's flight speed, which is expressed by the following formula:

[0035]

[0036] In the formula: , , for acceleration at any moment For time difference.

[0037] Compared with the prior art, the present invention has the following advantages:

[0038] 1. This invention determines the flight parameters required for identifying the aerodynamic parameters of an aircraft during level flight acceleration and deceleration by simulating engine thrust, flight drag, and aircraft mass. The flight parameters are then subjected to sampling rate correction, outlier removal, and filtering to improve their accuracy. Finally, the aerodynamic drag is obtained using the Bayesian parameter identification method to calculate the aircraft's level flight acceleration and deceleration performance, thereby improving the accuracy of flight performance calculation.

[0039] 2. This invention treats the parameters to be identified as random variables that follow a specific probability distribution. Based on Bayes' theorem, it constructs a complete probability model of prior distribution, likelihood function, and posterior distribution, outputs the optimal point estimate of the parameters, and fully quantifies the uncertainty of the parameters and the correlation between the parameters through the posterior covariance matrix, calculates the confidence interval of the parameters, and realizes full-dimensional statistics of aerodynamic parameters.

[0040] 3. This invention filters invalid and abnormal data by removing outliers, which reduces the interference of invalid data on subsequent identification algorithms, avoids iterative oscillations and invalid calculations caused by outliers, and improves processing efficiency. At the same time, outlier removal ensures the physical rationality of the input data, eliminates the identification bias caused by outliers, and improves the accuracy of data and identification results.

[0041] 4. This invention achieves rapid batch processing of data through sliding filtering with a fixed window length, while effectively suppressing high-frequency noise from the sensor. This makes the changing trends of fuel mass and speed data conform to actual flight patterns, eliminates calculation errors caused by noise fluctuations, and provides high-precision input data for subsequent aerodynamic parameter identification.

[0042] The present invention will now be described in further detail with reference to the accompanying drawings and embodiments. Attached Figure Description

[0043] Figure 1 This is a comparison chart of the original data and the smoothed and filtered data of the fuel quality change over time according to the present invention.

[0044] Figure 2 This is a diagram showing the correction and processing of the aircraft's level flight plus motion results according to the present invention;

[0045] Figure 3 This is a speed prediction diagram under different aerodynamic drag conditions according to the present invention. Detailed Implementation

[0046] This invention discloses a method for identifying aerodynamic parameters of an aircraft during level flight acceleration and deceleration and for evaluating its performance, comprising the following steps:

[0047] Step S1: Based on the aircraft's mass m, thrust T, and aerodynamic drag D, establish a dynamic model of the aircraft's acceleration and deceleration during level flight, and determine the flight parameters required for aerodynamic parameter identification.

[0048] Based on the characteristics of aircraft acceleration and deceleration during level flight and Newton's second law, treating the aircraft as a point mass, we obtain the dynamic equation of the aircraft's center of mass in the trajectory coordinate system, i.e., the dynamic model of acceleration and deceleration during level flight is as follows:

[0049] (1)

[0050] In the formula: T is the thrust; D is the drag; m is the mass; V pred t represents the predicted velocity; t represents time.

[0051] During flight, an aircraft's mass constantly changes due to factors such as fuel consumption and weapon deployment. Therefore, to obtain accurate data on these mass changes, multiple factors affecting the aircraft's mass must be considered. The total mass of an aircraft consists of four parts, including its empty weight. Weapon weight fuel weight Other weights For a given aircraft type, the empty weight of the aircraft is... The weight of the weapons remains constant throughout the flight and has no effect on acceleration or deceleration during level flight; the weapons carried by the aircraft vary depending on the mission, therefore the weapon weight... The difference is that aircraft may drop weapons during air combat, increasing the weight of the weapons. If the weight of fuel is reduced, the total mass of the aircraft will decrease, and according to formula (1), the acceleration of the aircraft will increase; fuel weight The weight of fuel varies depending on the aircraft's flight mission, and the weight of fuel increases as the aircraft consumes fuel during flight. Consequently, the fuel weight decreases. The reduction of the weight of the aircraft reduces its total mass, thus increasing its acceleration; other weights... Since the weight of the pilots and other factors have a relatively small impact on the total mass of the aircraft, the mass m of the aircraft can be expressed by the following formula:

[0052] (2)

[0053] In the formula: Empty weight of the aircraft; For weapon weight; This refers to the weight of the fuel. Other weight on the aircraft;

[0054] The thrust It can be expressed as follows:

[0055] (3)

[0056] In the formula: The total thrust of the two engines; and These are the maximum thrust of the two engines, respectively. and This is the throttle position coefficient; and This is the rotor speed coefficient; and To correct the parameters; all of the above parameters can be obtained from the aircraft's design specifications;

[0057] The aerodynamic drag D is expressed by the following formula:

[0058] (4)

[0059] In the formula: The drag coefficient of the entire aircraft without rudder deflection. , , These correspond to the drag increment coefficients caused by a unit angle of deflection of the horizontal stabilizer, aileron, and rudder, respectively. , , These correspond to the deflection angles of the horizontal stabilizer, aileron, and rudder, respectively. air density, Let be the airspeed at time i. , , , These represent the areas of the wings, horizontal stabilizer, ailerons, and rudder, respectively.

[0060] By combining the aircraft's level flight acceleration and deceleration dynamics model formula (1) with the aircraft's mass formula (2), thrust formula (3), and aerodynamic drag formula (4), a level flight acceleration and deceleration model based on flight parameter data is obtained, which is expressed by the following formula:

[0061] (5)

[0062] In the formula: V pred For predicting speed;

[0063] Analyze the intrinsic relationship between parameters and flight parameters in the level flight acceleration / deceleration model. and Related to the throttle lever angle, and Related to engine rotor speed, exhaust temperature, etc. Related to flight altitude, , , It refers to the change in the deflection angle of the control surfaces during aircraft flight. and These parameters are reflected in the flight parameters. Therefore, by extracting these parameters from the flight parameter data, the characteristics of flight speed variation can be solved, and the flight performance of the aircraft can be predicted.

[0064] Based on the dynamic model, the influencing factors of each parameter in the dynamic model are analyzed. The flight parameters required for aerodynamic parameter identification include air pressure, temperature, airspeed, and mass.

[0065] Step S2: Preprocess the flight parameters described in step S1;

[0066] The preprocessing of flight parameters required for aerodynamic parameter identification includes sampling rate correction, outlier removal, and smoothing filtering.

[0067] The sampling rate correction is to convert flight parameters with different sampling frequencies and periods into data with the same sampling period.

[0068] Because different sensors on the aircraft have different sampling frequencies and periods, while parameter identification requires the same sampling period, sampling rate correction is necessary. This means unifying flight parameters with different sampling frequencies and periods to a specific frequency and period. To effectively solve this problem and achieve data unification, the flight test data obtained from different sensors needs to be converted into a data format with the same sampling period. Generally, a shorter period is selected as the standard. To ensure that the integrity and authenticity of the processed signals are not compromised, interpolation is required for flight test data with relatively long sampling periods. Through a reasonable interpolation algorithm, the long-period data is refined within the required time period, so that the sampling period of these data is consistent with the data with the shortest sampling period, thus preparing for subsequent accurate analysis and system identification.

[0069] The outlier removal is achieved by setting a valid range for flight parameters and removing abnormal data.

[0070] During the recording of flight parameter data, some abnormal data may exist in the flight parameters due to sensor noise, random pulses, etc. These abnormal data may cause the identification parameters to deviate from the real physical characteristics, or even cause the identification algorithm to diverge. Therefore, it is necessary to remove outliers from the flight parameters required for aerodynamic parameter identification, effectively reduce the impact of abnormal data on modeling accuracy, improve data quality, and ensure the accuracy of subsequent aerodynamic parameter identification. According to formula (5), the level flight acceleration and deceleration dynamic model based on flight parameter data usually needs to remove outliers from the measured flight parameters such as air pressure, temperature, speed, and acceleration. Among them, air pressure and temperature are used to determine atmospheric environment parameters, while speed and acceleration directly reflect the motion state of the aircraft. These parameters are the basic input quantities for calculating flight altitude, Mach number, atmospheric density, aerodynamic force, and torque. Their data quality directly determines the reliability of the identification results. Therefore, it is necessary to remove outliers from them. The key to outlier removal is to set a reasonable range for each parameter, effectively remove abnormal data, and establish a constraint set B based on prior knowledge.

[0071] (6)

[0072] In the formula For static pressure, For static temperature, Airspeed, It is acceleration.

[0073] In actual flight data acquisition, there may be a situation where the amount of effective data is insufficient. When the number of effective samples n < 10, the identification algorithm may fail due to insufficient data. At this time, the fault tolerance mechanism is activated. First, the above constraints are automatically removed, and the first 500 sampling points of the original data are retained, which ensures both data quality and data integrity.

[0074] Taking flight altitude parameters as an example, normal outliers are eliminated using a defined range judgment method, removing values ​​that exceed the preset legal range [H]. min H max The sampling points are identified as outliers and removed. When sensor failure or strong disturbance causes a large number of altitude data to exceed the limit, and the effective data volume n < 10 after removal, the system automatically starts the fault tolerance mechanism: removes the altitude determination range constraint, stops outlier removal, directly retains the original altitude sampling data, and extracts the first 500 consecutive points as identification input. This ensures data integrity while taking into account data quality, and meets the minimum sample size requirement for flight parameter identification.

[0075] Because flight parameters inevitably contain errors such as sensor noise, random interference, and high-frequency jitter, the flight parameter data curve fluctuates violently and has many local abrupt changes, making it impossible to accurately reflect the true change pattern of the parameters. Therefore, a smoothing filter is applied to the flight parameter data. The smoothing filter is a 15-point moving average filter. The 15-point moving average filter uses a "window" of length 15 on the time series, which slides from left to right. Each time, the arithmetic mean of the 15 data points in the window is taken, and this average value is used as the filtered output for the current point.

[0076] like Figure 1 As shown, the blue line represents the original change in the total mass of the aircraft (due to changes in fuel mass over time). The curve is stepped. To improve the quality of flight parameter data, the original measurement signal is smoothed and filtered. Smoothing filtering can effectively suppress high-frequency noise and random disturbances, and retain the true trend of parameter changes. The total mass of the aircraft is processed using a 15-point moving average filtering method. The specific calculation formula is as follows:

[0077] (7)

[0078] After smoothing and filtering, the data curves are smooth and continuous, noise and interference are significantly suppressed, and the true changing trends of flight status and aerodynamic parameters are more clearly reflected. Figure 1 The red curve in the middle represents the change curve after smoothing filtering. Compared with the original data, the aircraft mass after smoothing filtering is approximately a monotonically decreasing curve, which is consistent with the actual flight situation of the aircraft. It has a higher signal-to-noise ratio and stability, effectively avoiding problems such as distortion of identification results and difficulty in algorithm convergence caused by excessive noise, and improving the accuracy of aerodynamic parameter identification.

[0079] Step S3: Based on the preprocessed flight parameters, construct the aerodynamic derivative influence matrix. and drag coefficient vector ;

[0080] The aerodynamic drag coefficients in the aerodynamic drag formula are separated from dynamic pressure, area, and control surface deflection angle, forming a combination of four parameters. ,in The above four parameters are combined into an aerodynamic derivative influence matrix. The expression is:

[0081] (8)

[0082] In the formula: ; ; ; , , All are left-side rudder deflection angles; , , All are right-side rudder deflection angles.

[0083] Will , , , The four drag coefficients, written as a vector, are represented by the drag coefficient vector, as shown by the following formula:

[0084] (9)

[0085] Based on equations (8) and (9), the aerodynamic drag is converted into a linear form, which is expressed by the following equation:

[0086] (10)

[0087] In the formula: Y is the resistance value D, Indicates random error. It follows a pattern with a mean of 0 and a covariance of . The weight matrix W follows a normal distribution and is a diagonal matrix with diagonal elements of 1. By weighting the differences in measurement noise under different flight conditions, heteroscedastic noise can be effectively suppressed to suppress the interference of heteroscedastic noise on the accuracy of parameter identification.

[0088] For a certain period of level flight acceleration and deceleration, due to the continuous changes in dynamic pressure and control surface deflection, N sets of flight parameters are selected during the motion, and the influence matrix of the aerodynamic derivatives is calculated based on the above. ,matrix This is transformed into an N*4 matrix, specifically represented as follows:

[0089] (11)

[0090] Based on this, Bayesian parameter identification is performed.

[0091] Step S4: Based on the aerodynamic derivative influence matrix described in step S3 and drag coefficient vector Combined with the Bayesian identification module, the aerodynamic drag coefficient is obtained;

[0092] The Bayesian identification module includes prior distribution modeling, likelihood function construction, and posterior distribution calculation.

[0093] Prior distribution modeling is a probabilistic distribution modeling of the aerodynamic parameters to be identified. In the identification of aerodynamic parameters during level flight acceleration and deceleration, the approximate range of variation of parameters such as drag coefficient is determined based on aerodynamic theory. A reasonable prior distribution is then constructed based on this, providing initial parameter probability information for Bayesian identification, thus improving the stability, convergence, and reliability of parameter identification. Especially when measurement data is limited or noise is high, it can effectively constrain parameters within a reasonable physical range, improving the accuracy of the identification results. The prior distribution modeling uses a multivariate Gaussian distribution, expressed by the following formula:

[0094] (12)

[0095] In the formula: It is the drag coefficient vector to be identified. ; It is a diagonal covariance matrix. .

[0096] The drag coefficient vector is obtained from the prior distribution. The approximate range of values ​​and probability distribution of .

[0097] The likelihood function quantitatively measures whether the measured flight data matches the aerodynamic parameter identification results under the prior distribution, reflecting the degree to which the observed data supports the parameter values. Within the Bayesian identification framework, the likelihood function describes the probability of the measured data occurring under given parameter conditions, providing a data-driven correction basis for parameter identification. The smaller the probability, the higher the drag coefficient to be identified. Compared with measured flight data The less it conforms, the more it needs to be corrected; the higher the probability, the greater the drag coefficient to be identified. Compared with measured flight data The more consistent the model, the more likely the weighted Gaussian function is selected for parameter identification of the level flight acceleration / deceleration model.

[0098] (13)

[0099] In the formula: It is the aerodynamic derivative influence matrix, which is the collection of measured flight data; It is a whole The i-th independent data point, , It is the dynamic pressure at time i; , Let be the airspeed at time i.

[0100] noise variance With the square of dynamic pressure Inversely proportional ( This indicates that measurement accuracy improves at high speeds during aircraft flight. When the aircraft flies at high speeds, the dynamic pressure is greater, resulting in relatively lower noise and higher measurement accuracy. Conversely, when the aircraft flies at low speeds, the dynamic pressure is smaller, resulting in relatively higher noise and lower measurement accuracy.

[0101] By combining prior knowledge and measured flight data, the probability distribution of the aerodynamic parameters to be identified is calculated through posterior distribution. This not only yields the optimal estimates of the parameters but also quantitatively describes their uncertainty, dispersion, and confidence intervals, enabling more robust statistical inference of the aerodynamic parameters. For parameter identification in level flight acceleration / deceleration models, a Gaussian posterior distribution is selected, expressed by the following formula:

[0102] (14)

[0103] In the formula: ; .

[0104] The confidence intervals of parameters calculated by the posterior covariance matrix can statistically verify the significance of the drag effect of each control surface (the influence is significant if the confidence interval does not contain zero), providing statistical support for aircraft aerodynamic characteristics analysis and maintenance support plan formulation.

[0105] By combining the prior distribution and the likelihood function, we obtain the analytical form of the posterior distribution and its mean. It is the optimal estimate of aerodynamic parameters and the final identification result of the drag coefficient under level flight acceleration and deceleration scenarios, covariance matrix. This reflects the uncertainty of the estimate.

[0106] The aforementioned prior distribution describes the prior knowledge and probability distribution of the aerodynamic parameters to be identified before the acquisition of observation data, providing initial constraint information for the parameters; the likelihood function characterizes the degree of matching between the measured flight data and the model parameters, reflecting the corrective effect of the observation data on the parameters; and the posterior distribution is the final result after integrating the prior information and the observation information. Together, these three constitute the Bayesian identification method.

[0107] Step S5: Based on the aerodynamic drag coefficient described in step S4 and combined with the dynamic model described in step S1, evaluate the flight performance of the aircraft during level flight acceleration and deceleration.

[0108] Based on the aircraft's acceleration and deceleration dynamics model during level flight, and given the aircraft's mass and thrust, the drag coefficient obtained from Bayesian parameter identification is imported. , The acceleration of the aircraft is obtained according to formula (5). , To predict the aircraft's speed, the acceleration needs to be integrated to obtain the aircraft's speed, which is expressed by the following formula:

[0109] (15)

[0110] In the formula: for The velocity value at that moment, for Previous moment speed value, for acceleration at any moment For time difference.

[0111] Verification has shown that the maximum relative error in speed prediction using this method is only 0.71%, and the performance evaluation results are highly consistent with actual flight conditions. Most existing methods calculate drag based on wind tunnel test data and design manual parameters when the aircraft leaves the factory, which deviates significantly from the actual aerodynamic characteristics caused by skin aging, control surface clearance, and aerodynamic wear after the aircraft enters service. In contrast, the core input drag coefficient of this application comes from Bayesian identification of the aircraft's actual flight parameter data, which perfectly matches the current actual aerodynamic state of the aircraft. Therefore, the performance identification of this application is superior to traditional methods.

[0112] like Figure 2 As shown in the figure, the vertical axis represents the corrected airspeed, and the horizontal axis represents time. The black curve represents the measured airspeed of the aircraft, the green curve represents the predicted airspeed after smoothing correction, and the purple curve represents the predicted airspeed after integration only. It can be seen that the predicted airspeed without smoothing correction has a large error compared to the measured airspeed, and the integration result gradually diverges. In contrast, the airspeed curve predicted by the data preprocessing + Bayesian identification method in this application closely matches the measured airspeed curve. The maximum relative error throughout the entire process under this condition is only 0.54%, and the airspeed prediction deviation is controlled within 3 m / s across the entire acceleration range from Mach 0.58 to 0.84, demonstrating a high goodness of fit. It can accurately reproduce the actual level flight acceleration and deceleration performance of an aircraft; it can be seen that the error between the predicted speed of the aircraft after smoothing correction and the actual speed of the aircraft is small, the fitting effect is good, and the reliability of parameter identification is improved.

[0113] Increased aerodynamic drag on an aircraft prolongs its acceleration and deceleration time during level flight, hindering its ability to quickly pursue and escape targets, thus impacting flight performance and combat effectiveness. As an aircraft's service life increases, the surface roughness of its skin may increase due to various factors (such as airflow erosion and foreign object impacts), leading to a continuous increase in aerodynamic drag and a gradual decline in acceleration and deceleration performance during level flight. Furthermore, the clearances of control surfaces such as ailerons and flaps widen, resulting in increased induced drag and further performance degradation. Therefore, it is necessary to assess flight performance under different drag conditions. Traditional methods, based on static wind tunnel data from the factory, cannot capture time-varying drag changes, thus failing to achieve accurate performance assessment throughout the aircraft's lifecycle. The Bayesian parameter identification method proposed in this application can identify the current aerodynamic drag coefficient in real time based on flight parameter data from different service stages and flight counts, accurately quantifying the performance changes caused by increased drag.

[0114] Based on the calculation methods and steps above, taking a certain type of aircraft as an example, we simulate the speed change over time under conditions of increased drag of 10%, 20%, 30%, and 40%. Figure 3 This graph shows the predicted speeds of a certain type of aircraft during level flight acceleration, with Mach numbers ranging from 0.87 to 0.94 and an altitude of 9250 meters, under different aerodynamic drag conditions. The vertical axis represents the corrected airspeed, and the horizontal axis represents time. The graph contains five curves, from top to bottom, illustrating the predicted speeds under different aerodynamic drag conditions. It can be seen from the graph that, for the same acceleration time, the final speeds differ under different aerodynamic drag conditions; the greater the increase in drag, the lower the final speed. Furthermore, within the same acceleration time, the aircraft cannot reach Mach 0.94 after the drag increases. Additionally, before 30 seconds, the impact of increased drag on speed is minimal. Therefore, increased aerodynamic drag will affect the aircraft's level flight acceleration performance after a certain period; the greater the increase in aerodynamic drag, the worse the acceleration performance, directly impacting air combat effectiveness.

[0115] Table 1 shows the acceleration timetable for reaching 680 m / s under different aerodynamic drag conditions.

[0116] Table 1. Comparison of acceleration times under different aerodynamic drag conditions during level flight acceleration at Mach numbers from 0.87 to 0.94 and an altitude of 9250 meters.

[0117] Aerodynamic drag conditions Acceleration time Reference resistance 255s Aerodynamic drag increases by 10% 260s Aerodynamic drag increases by 20% 264s Aerodynamic drag increases by 30% 268s Aerodynamic drag increased by 40% 273s

[0118] As can be seen from the table, as aerodynamic drag increases, the acceleration time in level flight becomes longer. When aerodynamic drag increases by 40%, the acceleration time increases by 18 seconds, resulting in decreased speed maneuverability and poorer maneuverability.

[0119] The above description is merely a preferred embodiment of the present invention and does not constitute any limitation on the present invention. Any simple modifications, alterations, or equivalent structural transformations made to the above embodiments based on the technical essence of the present invention shall still fall within the protection scope of the present invention.

Claims

1. A method for identifying aerodynamic parameters of an aircraft during level flight acceleration and deceleration and for evaluating its performance, characterized in that, Includes the following steps: Step S1: Based on the aircraft's mass m, thrust T, and aerodynamic drag D, establish a dynamic model of the aircraft's acceleration and deceleration during level flight, and determine the flight parameters required for aerodynamic parameter identification. Step S2: Preprocess the flight parameters described in step S1; Step S3: Based on the preprocessed flight parameters, construct the aerodynamic derivative influence matrix. and drag coefficient vector ; Step S4: Based on the aerodynamic derivative influence matrix described in step S3 and drag coefficient vector Combined with the Bayesian identification module, the aerodynamic drag coefficient is obtained; Step S5: Based on the aerodynamic drag coefficient described in step S4 and combined with the dynamic model described in step S1, evaluate the flight performance of the aircraft during level flight acceleration and deceleration.

2. The method for identifying aerodynamic parameters of aircraft acceleration and deceleration during level flight and evaluating their performance according to claim 1, characterized in that, The mass m of the aircraft mentioned in step S1 is expressed by the following formula: In the formula: Empty weight of the aircraft; For weapon weight; This refers to the weight of the fuel. Other weight on the aircraft; The thrust It can be expressed as follows: In the formula: The total thrust of the two engines; and These are the maximum thrust of the two engines, respectively. and This is the throttle position coefficient; and This is the rotor speed coefficient; and To correct the parameters; The aerodynamic drag D is expressed by the following formula: In the formula: The drag coefficient of the entire aircraft without rudder deflection. , , These correspond to the drag increment coefficients caused by a unit angle of deflection of the horizontal stabilizer, aileron, and rudder, respectively. , , These correspond to the deflection angles of the horizontal stabilizer, aileron, and rudder, respectively. air density, Let be the airspeed at time i. , , , These represent the areas of the wings, horizontal stabilizer, ailerons, and rudder, respectively.

3. The method for identifying aerodynamic parameters of aircraft acceleration and deceleration during level flight and evaluating their performance according to claim 2, characterized in that, The dynamic model of the aircraft's acceleration and deceleration is expressed by the following equation: In the formula: V pred For predicting speed; Based on the dynamic model, the influencing factors of each parameter in the dynamic model are analyzed. The flight parameters required for aerodynamic parameter identification include air pressure, temperature, airspeed, and mass.

4. The method for identifying aerodynamic parameters of aircraft acceleration and deceleration during level flight and evaluating their performance according to claim 1, characterized in that, The preprocessing of flight parameters required for aerodynamic parameter identification in step S2 includes sampling rate correction, outlier removal, and smoothing filtering. The sampling rate correction is to convert flight parameters with different sampling frequencies and periods into data with the same sampling period. The outlier removal is achieved by setting a valid range for flight parameters and removing abnormal data. The smoothing filter is a 15-point moving average filter.

5. The method for identifying aerodynamic parameters of aircraft acceleration and deceleration during level flight and evaluating their performance according to claim 1, characterized in that, The aerodynamic derivative influence matrix The expression is: In the formula: ; The drag coefficient vector The expression is: 。 6. The method for identifying aerodynamic parameters of aircraft acceleration and deceleration during level flight and evaluating their performance according to claim 1, characterized in that, The Bayesian identification module includes prior distribution modeling, likelihood function construction, and posterior distribution calculation. The prior distribution modeling adopts a multivariate Gaussian distribution, which is expressed by the following formula: In the formula: It is the drag coefficient vector to be identified. ; It is a diagonal covariance matrix. .

7. The method for identifying aerodynamic parameters of aircraft acceleration and deceleration during level flight and evaluating their performance according to claim 6, characterized in that, The likelihood function is expressed by the following equation: In the formula: It is the i-th row of the design matrix. These are the heteroscedasticity weights related to dynamic pressure; The posterior distribution is expressed by the following formula: In the formula: ; .

8. The method for identifying aerodynamic parameters of aircraft acceleration and deceleration during level flight and evaluating their performance according to claim 6, characterized in that, The evaluation of the aircraft's level flight acceleration and deceleration performance requires a correction to the aircraft's speed, which is expressed by the following formula: In the formula: , , for acceleration at any moment For time difference.

Images