Helicopter flight test-oriented frequency sweep excitation signal quality evaluation method

Abstract

The invention discloses a helicopter flight test-oriented frequency sweep excitation signal quality evaluation method, and the method comprises the steps: obtaining the excitation characteristics of a frequency sweep excitation signal to a manipulation response through the decoupling partial coherence analysis of manipulation and input, so as to directly judge the effectiveness of a manipulation excitation signal in different frequency bands; and thus, evaluation of the control excitation signal quality is realized. In addition, on the basis of partial coherence analysis, an improved scheme for operating the excitation signal can be obtained by combining time-frequency coupling analysis, and optimization of the sweep frequency excitation signal is completed. According to the method, the problems of low evaluation efficiency and high degree of dependence on personnel experience caused by the fact that excitation signal evaluation needs to be carried out by completing parameter model identification and combining engineering experience in an existing method are solved.

Description

Technical Field

[0001] This invention relates to the field of helicopter flight mechanics and flight test technology, and in particular to a method for evaluating the quality of swept frequency excitation signals for helicopter flight tests. Background Technology

[0002] Helicopters, with their hovering, vertical takeoff and landing, and low-speed maneuverability, have become indispensable aircraft. However, the inherent strong coupling, instability, and aerodynamic interference characteristics of helicopters make helicopter flight dynamics modeling difficult and hinder the accurate capture of their flight dynamic characteristics. Flight test-based system identification methods are one of the important means to improve the accuracy of helicopter flight dynamics modeling, and the primary condition for system identification-based modeling is to obtain high-quality flight test data that can reflect the inherent dynamic characteristics of helicopters.

[0003] Compared to fixed-wing aircraft, helicopter flight testing is more challenging, making it difficult to obtain high-quality helicopter flight test data. This is mainly due to the wide frequency distribution of helicopter motion modes, complex modal characteristics, and significant control and motion coupling. Consequently, there is a strong nonlinear mapping between control inputs and state responses, making it difficult to capture the functional relationship between a specific control and state response. However, in the system identification of helicopter flight dynamics models, it is essential to identify stability derivatives and control derivatives, and to obtain data on the control input signals' effects on principal axis and off-axis responses. Therefore, flight test data needs to reflect the influence characteristics of each control on different state responses. Through long-term theoretical analysis and engineering practice, domestic and international scholars and engineers have established relatively complete categories of helicopter flight test control excitation signals. The most commonly used are multi-stage square waves and frequency sweep excitation signals. Among them, frequency sweep excitation signals are particularly suitable for flight tests requiring longer excitation times and are most frequently used in flight tests oriented towards system identification. Frequency sweep excitation signals require different excitation parameters, including signal amplitude, frequency range, and frequency change rate, depending on the helicopter model and flight test conditions. However, the relationship between the frequency sweep excitation signal and the quality of flight test data is an extremely complex nonlinear mapping relationship. In particular, there are various state disturbances and pilot control behaviors during flight tests, making it difficult to directly design the parameters of the excitation signal through data models.

[0004] Currently, in frequency-sweep excitation flight tests for helicopter flight dynamics system identification, the quality analysis of the excitation signal mostly requires a preliminary model identification round. Analysis and improvement are then based on the identification results. As can be seen from the latest monograph, *Aircraft and Rotorcraft System Identification: Engineering Methods with Flight Test Examples 2nd Edition*, published by AIAA Press, by Mark B. Tischler, a leading expert in aircraft model identification and director of the Flight Mechanics and Flight Control Team at NASA Ames Research Center, internationally, the implementation and evaluation of frequency-sweep flight tests for helicopters still largely rely on the experience of pilots and flight test engineers. This approach is not only inefficient in terms of evaluation and improvement but also heavily dependent on the experience of flight test personnel. Therefore, to improve the efficiency of frequency-sweep excitation signal analysis and improvement, and thus shorten the helicopter flight test cycle, it is necessary to further explore excitation signal quality assessment methods capable of parametric model identification, while simultaneously reducing the reliance on the experience of analysts by adopting automated quantitative assessment methods as much as possible. Summary of the Invention

[0005] The technical problem to be solved by the present invention is to address the deficiencies mentioned in the background art by providing a method for evaluating the quality of swept frequency excitation signals for helicopter flight tests.

[0006] To solve the above-mentioned technical problems, the present invention adopts the following technical solution:

[0007] A method for evaluating the quality of swept-frequency excitation signals for helicopter flight testing includes the following steps:

[0008] Step 1) Based on the collected helicopter frequency sweep excitation test flight data, obtain the stable trim values ​​of the control input and the stable trim values ​​of the state response under stable conditions; and perform low-pass filtering on all control input data and state response data.

[0009] Step 2) Perform mean-removal processing on all low-pass filtered manipulation input data and state response data according to the following formula:

[0010] ;

[0011] in, , These are the i-th manipulation input data and the j-th state response data after low-pass filtering, respectively. , Let be the stable balance value of the i-th manipulation input and the stable balance value of the j-th state response, respectively, in a steady state. , These are the i-th manipulation input data and the j-th state response data after removing the mean;

[0012] Step 3) Determine the range of modal frequencies that the flight test data should reflect based on the mission objectives of the frequency sweep flight test, and then determine the window function width T of the flight test data. w Then, windowing is applied to all the mean-reduced manipulation input data and state response data.

[0013] Step 4) Perform Fourier transform on all windowed manipulation input data and state response data according to the following formula to obtain the frequency domain data of all manipulation inputs and the frequency domain data of the state response:

[0014] ;

[0015] in, , These are the frequency domain data vectors of the manipulation input and the state response, respectively. FFT(·) is the Fast Fourier Transform. For windowed manipulation input vectors, The data vector of the state response after windowing processing;

[0016] Step 5), for each set of frequency domain data of the manipulation input. Frequency domain data of state response Calculate the partial coherence function value of the manipulation input to the state response after removing coupling effects. f is , The frequency;

[0017] Step 5.1), calculate the autospectrum of the manipulated input. Self-spectrum of state response Cross spectrum of manipulator input to state response In the formula, T is the time of one frequency sweep excitation flight test;

[0018] Step 5.2), according to the following formula, respectively , , Average the results based on the time window:

[0019] ;

[0020] In the formula, n w The number of time windows used for a set of flight test data; Let f represent the autospectral density of the m-th and n-th manipulation inputs at frequency f within the i-th time window. The autospectral density of the m-th and n-th control inputs at frequency f after averaging; This represents the autospectral density of the state response at frequency f within the i-th time window. This is the autospectral of the state response at frequency f after taking the mean; This represents the cross spectrum of the state response of the m-th manipulation input pair at frequency f within the i-th time window. This is the cross spectrum of the state response of the m-th manipulation input pair after averaging at frequency f;

[0021] Step 5.3), calculate the residual autospectrum of the manipulated input data according to the following formula. Residual autospectrum of state response The residual cross spectrum of the manipulator input to the state response :

[0022] ;

[0023] ;

[0024] Step 5.4): Calculate the partial coherence function value of the manipulation input to the state response after removing coupling effects using the following formula. :

[0025] ;

[0026] Step 6): Determine the frequency range of the data according to the needs of the frequency sweep test flight mission, and set the partial coherence function threshold for the current test flight data analysis based on the characteristics of the identification model; compare the partial coherence function value corresponding to the frequency domain data of each set of control inputs and the frequency domain data of the state response with the partial coherence function threshold. If the partial coherence function value corresponding to the frequency domain data of each set of control inputs and the frequency domain data of the state response is greater than or equal to the partial coherence function threshold, the frequency sweep excitation signal quality is deemed qualified; otherwise, it is deemed unqualified.

[0027] As a further optimization of the sweep frequency excitation signal quality assessment method for helicopter flight tests according to the present invention, low-pass filtering is performed in step 1) according to the following formula:

[0028] ;

[0029] In the formula, f(·) represents the original manipulation input data and state response data before filtering. ω represents the filtered manipulation input data and state response data. c This represents the filter cutoff frequency, where t is the time variable, and sinc(·) is the filter function. ; This is the time variable for integration.

[0030] As a further optimization of the sweep frequency excitation signal quality evaluation method for helicopter flight tests according to the present invention, step 3) uses the Hanning window as shown in the following formula for windowing processing:

[0031] ;

[0032] in, , These are the manipulation input vector at the nth sampling time point after mean removal and the state response data vector at the nth sampling time point, respectively. , These are the manipulation input vector at the nth sampling time point after windowing and the state response data vector at the nth sampling time point, respectively. The number of data points within a window. The sampling rate for the experimental data.

[0033] Compared with the prior art, the present invention, employing the above technical solution, has the following technical effects:

[0034] 1) By using decoupled frequency domain control response partial coherence analysis, the time-domain excitation characteristics of the control excitation signal to the helicopter state response are obtained, and the quality assessment results of the swept frequency excitation signal are directly obtained. This eliminates the need for model parameter identification in existing assessment methods and improves the efficiency of control excitation signal quality assessment.

[0035] 2) Using the partial coherent calculation results of the helicopter state response to the frequency sweep excitation signal as the basis for excitation signal quality assessment greatly reduces the reliance on the experience of assessment technicians in the assessment process.

[0036] 3) Based on the partial coherent calculation results of the helicopter state response to frequency sweep excitation signals of different frequency bands, and combined with the time domain characteristics of the control excitation signal, the improvement scheme of the frequency sweep excitation signal can be directly analyzed to improve the efficiency of frequency sweep excitation flight test. Attached Figure Description

[0037] Figure 1 This is the implementation process for evaluating the quality of the sweep frequency excitation signal of the present invention;

[0038] Figure 2 It is a partial coherence analysis process for helicopter control response numbers;

[0039] Figure 3 This is a schematic diagram comparing the results of low-pass filtering of flight test data by the method of the present invention before and after the low-pass filtering in this embodiment.

[0040] Figure 4 This is a comparative schematic diagram showing the amplitude, phase, and off-coherence analysis results of a certain set of swept frequency excitation data using the method of the present invention in this embodiment;

[0041] Figure 5 This is a schematic diagram comparing the amplitude, phase, and off-coherence analysis results of the improved sweep frequency excitation data using the method of this aspect in the embodiments. Detailed Implementation

[0042] The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings:

[0043] This invention can be implemented in many different forms and should not be considered limited to the embodiments described herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully express the scope of the invention to those skilled in the art. In the drawings, components are enlarged for clarity.

[0044] This invention discloses a method for evaluating the quality of swept-frequency excitation signals for helicopter flight tests, the process of which is as follows: Figure 1 As shown, it includes the following steps:

[0045] Step 1) Based on the collected helicopter frequency sweep excitation test flight data, obtain the stable trim values ​​of the control input and the stable trim values ​​of the state response under stable conditions; and perform low-pass filtering on all control input data and state response data.

[0046] Perform low-pass filtering according to the following formula:

[0047] ;

[0048] In the formula, f(·) represents the original manipulation input data and state response data before filtering. ω represents the filtered manipulation input data and state response data. c This represents the filter cutoff frequency, where t is the time variable, and sinc(·) is the filter function. ; For integration time;

[0049] Step 2) Perform mean-removal processing on all low-pass filtered manipulation input data and state response data according to the following formula:

[0050] ;

[0051] in, , These are the i-th manipulation input data and the j-th state response data after low-pass filtering, respectively. , Let be the stable balance value of the i-th manipulation input and the stable balance value of the j-th state response, respectively, in a steady state. , These are the i-th manipulation input data and the j-th state response data after removing the mean, respectively.

[0052] Step 3) Determine the range of modal frequencies that the flight test data should reflect based on the mission objectives of the frequency sweep flight test, and then determine the window function width T of the flight test data. w Then, windowing is applied to all the mean-reduced manipulation input data and status response data.

[0053] This invention uses the Hanning window shown in the following formula for windowing:

[0054] ;

[0055] in, , These are the manipulation input vector at the nth sampling time point after mean removal and the state response data vector at the nth sampling time point, respectively. , These are the manipulation input vector at the nth sampling time point after windowing and the state response data vector at the nth sampling time point, respectively. The number of data points within a window. The sampling rate for the experimental data.

[0056] Step 4) Perform Fourier transform on all windowed manipulation input data and state response data according to the following formula to obtain the frequency domain data of all manipulation inputs and the frequency domain data of the state response:

[0057] ;

[0058] in, , These are the frequency domain data vectors of the manipulation input and the state response, respectively. FFT(·) is the Fast Fourier Transform. For windowed manipulation input vectors, The data vector of the state response after windowing processing;

[0059] Step 5), for each set of frequency domain data of the manipulation input. Frequency domain data of state response Calculate the partial coherence function value of the manipulation input to the state response after removing coupling effects. f is , The frequency;

[0060] Step 5.1), calculate the autospectrum of the manipulated input. Self-spectrum of state response Cross spectrum of manipulator input to state response In the formula, T is the time of one frequency sweep excitation flight test;

[0061] Step 5.2), according to the following formula, respectively , , Average the results based on the time window:

[0062] ;

[0063] In the formula, n w The number of time windows used for a set of flight test data; Let f represent the autospectral density of the m-th and n-th manipulation inputs at frequency f within the i-th time window. This is the autospectral density of the m-th and n-th manipulation inputs at frequency f after averaging. This represents the autospectral density of the state response at frequency f within the i-th time window. This is the autospectral of the state response at frequency f after taking the mean. This represents the cross spectrum of the state response of the m-th manipulation input pair at frequency f within the i-th time window. This is the cross spectrum of the state response of the m-th manipulation input pair after averaging at frequency f.

[0064] Step 5.3), calculate the residual autospectrum of the manipulated input data according to the following formula. Residual autospectrum of state response The residual cross spectrum of the manipulator input to the state response :

[0065] ;

[0066] ;

[0067] Step 5.4), as follows Figure 2 As shown, the partial coherence function value of the manipulation input to the state response after removing coupling effects is calculated according to the following formula. :

[0068] ;

[0069] Step 6) Based on the needs of the frequency sweep test flight mission, determine the frequency range of the data, and set the partial coherence function threshold for the current test flight data analysis according to the characteristics of the identification model. Compare the partial coherence function value corresponding to the frequency domain data of each set of control inputs and the frequency domain data of the state response with the partial coherence function threshold. If the partial coherence function value corresponding to the frequency domain data of each set of control inputs and the frequency domain data of the state response is greater than or equal to the partial coherence function threshold, the frequency sweep excitation signal quality is deemed qualified; otherwise, it is deemed unqualified.

[0070] By combining the time-domain curves of the collected control input signal data, we can also analyze the deficiencies of the control input in the corresponding frequency band in terms of amplitude, frequency change rate, excitation time, etc., and propose improvement suggestions for another frequency sweep test flight.

[0071] The following describes the application of the frequency-sweeping excitation signal quality assessment method for helicopter flight tests of this invention to the evaluation of excitation signal quality for a specific helicopter model during frequency-sweeping flight tests. First, the original flight test data is low-pass filtered to obtain smooth time-domain flight test data. Figure 3 The filtering results for two sets of low signal-to-noise ratio data show that the filtering effect is very good. Next, the partial coherence calculation of the control input signal to the spindle response was performed according to the method of the present invention. Based on the need that this set of flight test data is mainly used to identify mid-to-high frequency modes, the data frequency range was set to 1-10 rad / s, and the partial coherence value of the spindle control response in this frequency band was required not to be lower than 0.6. Figure 4 The first set of flight test data yielded the partial coherence calculation results for the spindle control response. The three sub-graphs from top to bottom represent the changes in control response amplitude with frequency, control response phase with frequency, and partial coherence function value with frequency, respectively. It can be seen that the partial coherence function value in the 1-10 rad / s frequency range is mostly below 0.6 and exhibits severe oscillation. Therefore, the evaluation result indicates that this set of sweep excitation signals is unqualified. Combining the time-domain curves, it was found that the pilot control amplitude was too small and the frequency change rate of the sweep was too high. Improvement suggestions were proposed, and a second set of flight test data was obtained through a re-flight. The sweep excitation signal quality analysis was performed again using the method of this invention, and the results are as follows. Figure 5 As shown in the figure, the partial coherence function value of the main axis control response in this set of flight test data remains above 0.6 in the range of 1-10 rad / s, and is almost smooth without oscillation. Therefore, the evaluation result is that the frequency sweep excitation signal of this set is qualified and can meet the requirements for subsequent use.

[0072] This invention addresses the problem of quality assessment of swept-frequency excitation signals in helicopter flight tests. It proposes a method for assessing the quality of swept-frequency excitation signals in helicopter flight tests. By using decoupled partial coherence analysis of the control and input, the excitation characteristics of the swept-frequency excitation signal in response to the control are obtained. This allows for direct determination of the effectiveness of the control excitation signal in different frequency bands, thereby achieving quality assessment. Furthermore, based on partial coherence analysis, time-frequency coupling analysis can be combined to obtain improved control excitation signal schemes, optimizing the swept-frequency excitation signal. This invention avoids obtaining signal quality assessment through model parameter identification, improving analysis efficiency. Moreover, the quality of swept-frequency excitation signals in different frequency bands can be directly obtained from the partial correlation calculation results of the decoupled control response, significantly reducing the reliance on the experience of flight test technicians during signal analysis. This invention solves the problem that existing methods require parameter model identification and engineering experience for excitation signal assessment, leading to low assessment efficiency and high dependence on personnel experience.

[0073] It will be understood by those skilled in the art that, unless otherwise defined, all terms used herein (including technical and scientific terms) have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains. It should also be understood that terms such as those defined in general dictionaries should be understood to have the same meaning as in the context of the prior art, and should not be interpreted in an idealized or overly formal sense unless defined as herein.

[0074] The specific embodiments described above further illustrate the purpose, technical solution, and beneficial effects of the present invention. It should be understood that the above description is only a specific embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for evaluating the quality of swept-frequency excitation signals for helicopter flight testing, characterized in that, Includes the following steps: Step 1), based on the collected helicopter frequency sweep excitation test flight data, obtain the stable trim value of the control input and the stable trim value of the state response under stable conditions; All manipulation input data and status response data are low-pass filtered. Step 2) Perform mean-removal processing on all low-pass filtered manipulation input data and state response data according to the following formula: ； in, , These are the i-th manipulation input data and the j-th state response data after low-pass filtering, respectively. , Let be the stable balance value of the i-th manipulation input and the stable balance value of the j-th state response, respectively, in a steady state. , These are the i-th manipulation input data and the j-th state response data after removing the mean; Step 3) Determine the range of modal frequencies that the flight test data should reflect based on the mission objectives of the frequency sweep flight test, and then determine the window function width T of the flight test data. w Then, windowing is applied to all the mean-reduced manipulation input data and state response data. Step 4) Perform Fourier transform on all windowed manipulation input data and state response data according to the following formula to obtain the frequency domain data of all manipulation inputs and the frequency domain data of the state response: ； in, , These are the frequency domain data vectors of the manipulation input and the state response, respectively. FFT(·) is the Fast Fourier Transform. For windowed manipulation input vectors, The data vector of the state response after windowing processing; Step 5), for each set of frequency domain data of the manipulation input. Frequency domain data of state response Calculate the partial coherence function value of the manipulation input to the state response after removing coupling effects. f is , The frequency; Step 5.1), calculate the autospectrum of the manipulated input. Self-spectrum of state response Cross spectrum of manipulator input to state response In the formula, T is the time of one frequency sweep excitation flight test; Step 5.2), according to the following formula, respectively , , Average the results based on the time window: ； In the formula, n w The number of time windows used for a set of flight test data; Let f represent the autospectral density of the m-th and n-th manipulation inputs at frequency f within the i-th time window. The autospectral density of the m-th and n-th control inputs at frequency f after averaging; This represents the autospectral density of the state response at frequency f within the i-th time window. This is the autospectral of the state response at frequency f after taking the mean; This represents the cross spectrum of the state response of the m-th manipulation input pair at frequency f within the i-th time window. This is the cross spectrum of the state response of the m-th manipulation input pair after averaging at frequency f; Step 5.3), calculate the residual autospectrum of the manipulated input data according to the following formula. Residual autospectrum of state response The residual cross spectrum of the manipulator input to the state response : ； ； Step 5.4): Calculate the partial coherence function value of the manipulation input to the state response after removing coupling effects using the following formula. : ； Step 6): Determine the frequency range of the data according to the needs of the frequency sweep test flight mission, and set the partial coherence function threshold for the current test flight data analysis based on the characteristics of the identification model; compare the partial coherence function value corresponding to the frequency domain data of each set of control inputs and the frequency domain data of the state response with the partial coherence function threshold. If the partial coherence function value corresponding to the frequency domain data of each set of control inputs and the frequency domain data of the state response is greater than or equal to the partial coherence function threshold, the frequency sweep excitation signal quality is deemed qualified; otherwise, it is deemed unqualified.

2. The method for evaluating the quality of swept-frequency excitation signals for helicopter flight testing according to claim 1, characterized in that, In step 1), low-pass filtering is performed according to the following formula: ； In the formula, f(·) represents the original manipulation input data and state response data before filtering. ω represents the filtered manipulation input data and state response data. c This represents the filter cutoff frequency, where t is the time variable, and sinc(·) is the filter function. ; This is the time variable for integration.

3. The method for evaluating the quality of swept-frequency excitation signals for helicopter flight testing according to claim 1, characterized in that, In step 3), the Hanning window is used for windowing as shown in the following formula: ； in, , These are the manipulation input vector at the nth sampling time point after mean removal and the state response data vector at the nth sampling time point, respectively. , These are the manipulation input vector at the nth sampling time point after windowing and the state response data vector at the nth sampling time point, respectively. The number of data points within a window. The sampling rate for the experimental data.

Images