Flight parameter preprocessing method and system based on QAR data

Abstract

The invention relates to a flight parameter preprocessing method and system based on QAR data, and aims to improve the accuracy of flight performance evaluation. According to the method, the problem that the flight parameters in the QAR data cannot be accurately extracted is solved. The method comprises the following steps: firstly, carrying out same-frequency processing on data to ensure the consistency of the data; then, key parameters in the flight process are extracted through parameter identification; thirdly, performing singular value detection and correction on the data by applying residual analysis, filling missing data by adopting an interpolation method, and removing noise in combination with a filtering technology; afterwards, based on the flight phase recognition technology, key data in the landing stage of the aircraft are extracted and intercepted. And finally, archiving and integrating the processed data to generate a high-quality preprocessed data set. According to the method, the utilization rate of the QAR data is effectively improved, the method is particularly suitable for flight data analysis in an aircraft landing stage, and reliable technical support is provided for flight safety evaluation and performance optimization.

Description

Background Art

[0002] In recent years, the aviation transportation industry has developed rapidly. Many data acquisition devices are installed on aircraft to record various parameters, such as altitude, angle of attack, airspeed and other indicators. Among them, the quick access recorder (QAR) is an important flight parameter recording device on the aircraft. This device can collect hundreds or even thousands of data at the same time, covering most of the parameters in aircraft operation. The analysis of QAR data has been applied to the fields of flight technical inspection, flight quality analysis, safety accident investigation, etc. The angle of attack is one of the parameters it records, which refers to the angle between the forward direction of the aircraft body and the lateral plane of the wing. It is an important parameter in aircraft flight and has an impact on the aircraft's lift, drag, stability and maneuverability.

[0003] QAR data records various flight data measured by sensors during the flight of the aircraft. It cannot be used directly in calculation and analysis, and needs to be converted into a coordinate system based on the fuselage before analysis. There are many correction methods involved, such as combining flight parameters and attitude information, using flight dynamics models, etc. Summary of the invention

[0004] In order to solve the above-mentioned problem in the prior art, that is, due to errors caused by sensor recording or measurement, the various collected data deviate from the actual data or even are wrong, and cannot be directly applied to the aircraft aerodynamic calculation research.

[0005] In a first aspect, the present invention provides a flight parameter preprocessing method based on QAR data, the method comprising:

[0006] Step S1, extracting data from a quick access recorder;

[0007] Step S2, performing frequency synchronization processing on the extracted data to obtain frequency synchronized data;

[0008] Step S3, performing parameter identification on the frequency-synchronized data to obtain required data;

[0009] Step S4, performing singular value detection and correction, data filling and data filtering on the data after parameter identification to obtain cleaned and repaired data;

[0010] Step 5, extracting and intercepting the cleaned and repaired data using the aircraft phase identification and data extraction method;

[0011] Step S6, extracting the landing phase data, archiving the intercepted data, and integrating them into output pre-processed data.

[0012] Furthermore, the data frequency synchronization processing includes:

[0013] Based on time series, with 1HZ as the standard;

[0014] The high-frequency data is downconverted to 1HZ frequency, and the excess frequencies are saved by majority renaming to preserve the authenticity of the original data.

[0015] The low-frequency data is supplemented and expanded according to the 1HZ data, and a position for data filling is reserved so that data can be filled in later through data filling and filtering methods;

[0016] Furthermore, the data cleaning and repair process includes:

[0017] Data filling: fill in missing data through Lagrange interpolation method;

[0018] Singular value detection and correction, using residual analysis based on Kalman filtering to test and eliminate singular values;

[0019] Data filtering, filtering out noise in QAR data through Kalman filtering method and Savitzky-Golay (SG) filtering method;

[0020] Furthermore, the data filtering method includes:

[0021] For data with high-frequency oscillation problems, Kalman filtering is used for denoising and smoothing;

[0022] For data that changes steadily and needs filtering, SG filtering is used for denoising and smoothing;

[0023] Furthermore, the Kalman filter and its residual analysis formula are:

[0024] Establish the state equation and observation equation:

[0025]

[0026] Among them, x k is the system state, x k-1 is the state of the system at the previous moment, A is the state transfer matrix, and B is the control input matrix u k-1 is the control input, w k-1 is the process noise, z k is the estimated value, H is the observation matrix, v k is the measurement noise;

[0027] To make a status prediction:

[0028] X k =AX k-1 +Bu k-1 ;

[0029] Among them, X k is the state prediction at time k, X k-1 is the state estimate at the previous moment;

[0030] Forecast error covariance:

[0031] P k =AP k-1 A T +Q;

[0032] Among them, P k is the prediction error covariance, P k-1 is the covariance of the previous moment, Q is the covariance of the process noise;

[0033] Kalman gain calculation:

[0034] K k =P k H T (HP k H T +R) -1 ;

[0035] Among them, K k is the Kalman gain matrix, R is the covariance of the measurement noise;

[0036] Status Update:

[0037] X k+1 =X k +K k (z k -HX k );

[0038] Error covariance update:

[0039] P k+1 =(IK k H)P k ;

[0040] Among them, x k+1 is the updated system status, P k+1 is the updated system status;

[0041] Through the above recursive prediction and state updating process, the optimal estimated value of the state variable based on Kalman filtering can be obtained.

[0042] Compute the residual:

[0043] Y k =z k -HX k ;

[0044] Among them, Y kis the residual;

[0045] Calculate the innovation covariance:

[0046] S k =HP k H T +R;

[0047] Among them, S k is the innovation covariance;

[0048] Standardized residual calculation:

[0049]

[0050] Among them, v k is the standardized residual;

[0051] Outlier Detection:

[0052] Determine z k is a singular value;

[0053] in, is the residual threshold; through the above residual calculation and judgment process, it can be determined whether there are singular values ​​in the data and the location of the singular values ​​can be determined.

[0054] Furthermore, the SG filtering formula is:

[0055]

[0056] Among them, y′ i is the smoothed data point, y i is the original data point, c j is the Savitzky-Golay filter coefficient, m is half the width of the window, and the window length is 2m+1;

[0057] Among them, c j The calculation involves constructing a Vandermonde matrix and solving its pseudo-inverse, which can be done in linear algebra:

[0058]

[0059] Where n is the order of the fitting polynomial;

[0060] Assume that the pseudo-inverse of matrix A is A + , then the filter coefficient c can be expressed as:

[0061] c=A + e0;

[0062] Among them, e0 is a unit vector, which represents the smoothing weight of the central point of the window;

[0063] Furthermore, in the data filling part, the Lagrange interpolation method is used to fill and complete the missing data. The formula of the Lagrange interpolation method is:

[0064]

[0065] Among them, P n (x) is the constructed interpolation polynomial, y i is the function value corresponding to a given data point, l j (x) is the basis function, defined as:

[0066]

[0067] Among them, x0, x j-1 、x j 、x j+1 、x n are the horizontal coordinates of the interpolation points;

[0068] Furthermore, for data containing missing values ​​that need to be filtered, the above-mentioned Lagrange interpolation method is preferentially used to fill the missing values ​​before performing the SG filtering operation; the selection of this processing order has significant technical advantages; first, the SG filtering algorithm relies on the continuity of the data and performs calculations based on points within a local window, so missing values ​​will cause the filter to be unable to obtain accurate smoothing results; by pre-filling the missing values, it can be ensured that the filtering operation is performed on the complete data set, thereby avoiding inaccurate or failed calculations; secondly, filling the missing values ​​can avoid the boundary effect of the filter near the missing points, so that the filtered data maintains consistency and smoothness; in addition, filling the missing values ​​first can reduce the errors or noise introduced by interpolation in subsequent filtering steps, and enhance the smoothness and accuracy of the overall data; through this process optimization, the integrity of data processing and the filtering effect are effectively guaranteed, which is suitable for data preprocessing in a variety of signal processing scenarios.

[0069] Furthermore, the specific method of the aircraft phase identification and data extraction is:

[0070] By analyzing the key flight parameters in the QAR data, including radio altitude, speed, pitch angle, heading angle, reverse thrust device status, etc., combined with the preset flight phase standards, the various phases of the flight landing process are identified; the phase identification process divides the data of the flight landing process into four main phases: approach phase, air phase, transition phase and deceleration phase. The specific characteristics of each phase division include:

[0071] Approach phase: The criterion for identifying the approach phase is that the radio altitude during the landing phase is between 1500 feet and 50 feet;

[0072] Air segment: The criterion for identifying the air segment is that the radio altitude during the landing phase is between 50 feet and 0 feet;

[0073] Transition phase: The standard for identifying the transition phase is that the radio altitude during the landing phase is 0 feet or negative, which means that the aircraft has touched down, until the main wheels and nose wheels of the aircraft are completely on the ground and the engine conversion ratio (EPR) and exhaust gas temperature (EGT) data change, and the reverse thrust device and spoilers begin to operate;

[0074] Deceleration segment: The standard for identifying the deceleration segment is that the aircraft is completely touched down and the reverse thrust device and spoilers begin to operate until the aircraft heading angle deviates from the runway reference azimuth by 90 degrees.

[0075] The runway reference azimuth in the deceleration section is an average value of the heading angle data of the aircraft in the transition section.

[0076] In a second aspect, the present invention provides a flight parameter preprocessing system based on QAR data, the system comprising:

[0077] a data extraction module configured to extract data from the QAR;

[0078] A data frequency synchronization module is configured to perform frequency synchronization on the extracted data to obtain frequency synchronized data so that the data has the same time series;

[0079] A parameter identification module is configured to perform parameter identification on the frequency-synchronized data to obtain required data;

[0080] Based on the data after parameter identification, singular value detection and correction, data filtering, and data filling are performed to obtain cleaned and repaired data;

[0081] A data extraction and interception module is configured to extract and intercept data based on the cleaned and repaired data using an aircraft phase identification and data extraction method;

[0082] The data output module is configured to extract the landing phase data, archive the intercepted data, and integrate them into output pre-processed data.

[0083] Beneficial effects of the present invention:

[0084] (1) The present invention designs a unique flight parameter preprocessing method based on QAR data, which can effectively make up for the shortcomings of existing QAR data processing methods, that is, they cannot be directly used and have problems such as singular values, noise, and data missing. The QAR data processed by this method can be directly used to analyze and calculate the aerodynamic characteristics of the aircraft.

[0085] (2) The present invention can not only detect singular values ​​in flight data in real time through the residual analysis method based on Kalman filtering, but also automatically correct them, thereby reducing the impact of data errors on subsequent analysis and ensuring accurate monitoring of flight status and reliability of data analysis. BRIEF DESCRIPTION OF THE DRAWINGS

[0086] Other features, objects and advantages of the present application will become more apparent by reading the detailed description of non-limiting embodiments made with reference to the following drawings:

[0087] Figure 1 is a flow chart of a flight parameter preprocessing method based on QAR data in an embodiment of the present invention;

[0088] Figure 2 is a flow chart of the calculation process of Kalman filtering in an embodiment of the present invention;

[0089] Figure 3 is a comparison diagram of the result of Kalman filtering and the original data in an embodiment of the present invention;

[0090] Figure 4 is a flow chart of the calculation process of Savitzky-Golay filtering in an embodiment of the present invention;

[0091] Figure 5 is a comparison diagram of the result of Savitzky-Golay filtering and the original data in an embodiment of the present invention;

[0092] Figure 6 is a flowchart of searching for radio altitude singular values ​​in an embodiment of the present invention; DETAILED DESCRIPTION

[0093] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. The embodiments described herein are only used to explain the relevant inventions, but not to limit the invention. It should also be noted that the drawings only show some embodiments related to the invention, not all embodiments.

[0094] In order to more clearly explain the flight parameter preprocessing method based on QAR data of the present invention, the following is combined with Figure 1 Each step in the embodiment of the present invention is described in detail.

[0095] The flight parameter preprocessing method based on QAR data provided by the present invention includes steps S1 to S6, and each step is described in detail as follows:

[0096] Step S1, extracting parameters from a quick access recorder;

[0097] Step S2, the extracted data is processed into the same frequency to obtain the data after the same frequency:

[0098] After obtaining the QAR data, we first need to match it with the data frequency in the QAR data version library to obtain the original frequency of data collection;

[0099] After obtaining the high-frequency data, the data is classified and processed according to the multiples of the data frequency relative to the 1HZ frequency. First, the data is down-converted according to the 1HZ frequency, and then the excess frequencies are saved by renaming them in the majority to preserve the authenticity of the original data;

[0100] After obtaining the low-frequency data, expand it according to the 1HZ data, and reserve the position for data filling so that data can be filled in later through data filling and filtering methods;

[0101] Step S3, performing parameter identification on the frequency-synchronized data to obtain required data;

[0102] The frequency-synchronized data set is screened to select parameters closely related to flight performance, including altitude, speed, pitch angle, roll angle, and vertical speed. In this process, the method will classify the data according to the data type. For example, speed parameters such as airspeed, ground speed, and vertical speed are grouped together to facilitate subsequent parameter query and operation. Step S4, perform singular value detection and correction, data filling, and data filtering on the data after parameter identification to obtain cleaned and repaired data;

[0103] Data filling: data may be empty at some point during the recording process. The data filling method used in this paper is Lagrange interpolation. Lagrange interpolation is a commonly used polynomial interpolation method. Its basic idea is: construct a polynomial function in given discrete data points so that its curve passes through all data points, thereby achieving the purpose of smooth interpolation. The specific method is:

[0104]

[0105] Among them, P n (x) is the constructed interpolation polynomial, y i is the function value corresponding to a given data point, l j (x) is the basis function used to ensure that the interpolation polynomial passes through every known data point and is defined as:

[0106]

[0107] Among them, x0, x j-1 、x j 、x j+1 、x n are the horizontal coordinates of the interpolation points;

[0108] In the data filling process, we first need to determine the set of known data points (x0, y0), (x1, y1), ..., (x n ,y n ), these data points are the adjacent points before and after the missing data. Through the Lagrange interpolation method, a polynomial function is constructed so that the function passes through all known data points. Then, the missing moment data is estimated through the polynomial function to complete the data at that moment. Data filtering, the data collected by the sensor will inevitably have data problems such as high-frequency random noise caused by sensor accuracy, pulse noise caused by instrument failure, external factor noise such as electromagnetic interference, and discrete errors generated during data sampling. For data with high-frequency oscillation problems, the method adopted by the present invention is the Kalman filtering method. Combined with Figure 2 The calculation process flow chart of Kalman filter, the specific method of obtaining it is:

[0109] Establish the state equation and observation equation:

[0110]

[0111] Among them, x k is the system state, x k-1 is the state of the system at the previous moment, A is the state transfer matrix, and B is the control input matrix u k-1 is the control input, w k-1 is the process noise, z k is the estimated value, H is the observation matrix, v k is the measurement noise;

[0112] To make a status prediction:

[0113] X k =AX k-1 +Bu k-1 ;

[0114] Among them, X k is the state prediction at time k, X k-1 is the state estimate at the previous moment;

[0115] Forecast error covariance:

[0116] P k =AP k-1 A T +Q;

[0117] Among them, P k is the prediction error covariance, P k-1 is the covariance of the previous moment, Q is the covariance of the process noise;

[0118] Kalman gain calculation:

[0119] K k =P k H T (HP k H T +R) -1 ;

[0120] Among them, K k is the Kalman gain matrix, R is the covariance of the measurement noise;

[0121] Status Update:

[0122] X k+1 =X k +K k (z k -HX k );

[0123] Error covariance update:

[0124] P k+1 =(IK k H)P k ;

[0125] Among them, x k+1 is the updated system status, P k+1 is the updated system status;

[0126] Through the above recursive prediction and state update process, the optimal estimate of the state variable based on Kalman filtering can be obtained. Figure 3 As shown, the comparison between the RALT_LH data after Kalman filtering and the original RALT_LH data is shown. Figure 3 In the figure, the solid line represents the original RALT_LH data, and the dotted line represents the RALT_LH data after Kalman filtering. After Kalman filtering, the data presents a smoother and more stable characteristic, effectively eliminating noise interference and improving the accuracy and reliability of the data. This processing result is more in line with the actual application needs, providing basic support for subsequent data analysis and utilization, while restoring the authenticity of the data.

[0127] For data that changes steadily and needs filtering, SG filtering is used for denoising and smoothing. SG filtering is an effective method for smoothing data. It reduces noise through local polynomial fitting while retaining the main trends and local features of the data. This filtering method is suitable for data smoothing while keeping spikes or other important features from being blurred.

[0128] like Figure 4 The calculation process of Savitzky-Golay filtering is shown, and the SG filtering formula is:

[0129]

[0130] Among them, y′ i is the smoothed data point, y i is the original data point, c j is the Savitzky-Golay filter coefficient, m is half the width of the window, and the window length is 2m+1;

[0131] Among them, c j The calculation involves constructing a Vandermonde matrix and solving its pseudo-inverse, which can be done in linear algebra:

[0132]

[0133] Where n is the order of the fitting polynomial;

[0134] Assume that the pseudo-inverse of matrix A is A + , then the filter coefficient c can be expressed as:

[0135] c=A + e0;

[0136] Among them, e0 is a unit vector, which represents the smoothing weight for the center point of the window. This unit vector ensures that the data point in the center of the window has a higher weight during the smoothing process, so that the fitting result is more biased towards the central data point. In this way, the SG filter can smooth the data within the window while maintaining the main trend of the data near the central data point;

[0137] like Figure 5 As shown, the comparison between the IAS data after Savitzky-Golay filtering and the original IAS data is shown. Figure 5 In the figure, the solid line represents the IAS data after Savitzky-Golay filtering, and the dotted line represents the original IAS data. By constructing the Vandermonde matrix and solving the pseudo-inverse, the Savitzky-Golay filter can perform local polynomial fitting on the data to achieve a smoothing effect. The calculation process and application scenarios of the filter are shown in detail above. It has a good application effect in actual data processing, especially for QAR data with noise and time series.

[0138] For data that contains missing values ​​and needs to be filtered, the above-mentioned Lagrange interpolation method is preferably used to fill the missing values ​​before performing the SG filtering operation; the choice of this processing order has significant technical advantages; first, the SG filtering algorithm relies on the continuity of the data and is calculated based on points within a local window, so missing values ​​will cause the filter to be unable to obtain accurate smoothing results; by pre-filling the missing values, it can be ensured that the filtering operation is performed on the complete data set, thereby avoiding inaccurate or failed calculations; secondly, filling the missing values ​​can avoid the boundary effect of the filter near the missing points, so that the filtered data maintains consistency and smoothness; in addition, filling the missing values ​​first can reduce the errors or noise introduced by interpolation in subsequent filtering steps, and enhance the smoothness and accuracy of the overall data; through this process optimization, the integrity of data processing and the filtering effect are effectively guaranteed, which is suitable for data preprocessing in a variety of signal processing scenarios.

[0139] Singular value detection and correction, through the residual analysis method based on Kalman filter, singular value detection and correction are performed on QAR data, and the position of singular values ​​is marked. The specific method is as follows:

[0140] Compute the residual:

[0141] Y k =z k -HX k ;

[0142] Among them, Y k is the residual;

[0143] Calculate the innovation covariance:

[0144] S k =HP k H T +R;

[0145] Among them, S k is the innovation covariance;

[0146] Standardized residual calculation:

[0147]

[0148] Among them, v k is the standardized residual;

[0149] Outlier Detection:

[0150] Determine z k is a singular value;

[0151] in, is the residual threshold. If the residual threshold is too small, the Kalman filter may frequently mistake normal noise fluctuations for abnormalities. If the residual threshold is too large, the Kalman filter may not be able to detect the real abnormal conditions or failures of the system in time.

[0152] like Figure 6 As shown, for the radio altitude singular value retrieval process, the radio altitude data predicted by Kalman filtering and the observed value are subjected to residual analysis, and the singular values ​​are corrected and marked. The present invention is adjusted according to the noise level, system performance requirements and application scenarios, and 2-3 times the standard deviation is selected to determine the residual threshold through experimental tuning;

[0153] The flight data at each moment is predicted, and the predicted value is compared with the actual observed value to calculate the residual. If the residual exceeds the preset threshold, the data point is judged to be a singular value. The detected singular value will be automatically marked and corrected accordingly according to the system model to ensure the accuracy of subsequent data analysis. The correction process is based on the predicted value of the Kalman filter, and uses its estimated result of the normal data state to replace the abnormal point with the most likely reasonable value.

[0154] Step S5, extracting and intercepting the cleaned and repaired data using an aircraft phase identification and data extraction method;

[0155] In this embodiment, the cleaned and repaired data is extracted and intercepted using the aircraft phase identification and data extraction method, and the acquisition method is:

[0156] By analyzing the key flight parameters in the QAR data, including radio altitude, speed, pitch angle, heading angle, reverse thrust device status, etc., combined with the preset flight phase standards, the various phases of the flight landing process are identified; the phase identification process divides the data of the flight landing process into four main phases: approach phase, air phase, transition phase and deceleration phase. By combining the example QAR data, the phases are divided into:

[0157] 1. Approach phase: The radio altitude is between 1495.9ft and 52.9ft, and the time period of the approach phase is 8246 to 8373s (this is the time originally recorded by the QAR data), a total of 127s, and the data is sampled at 1Hz;

[0158] 2. Air segment: The radio altitude is between 48.9ft and 0.4ft, 8373s to 8382s is the time period of the air segment, a total of 9s, and the data is sampled at 1Hz;

[0159] 3. Transition: When the radio altitude is 0ft or a small negative value, the aircraft touches down, and t=8382s is the start of the transition. During 8387s-8392s, the engine conversion ratio EPR and temperature EGT values ​​change, and the reverse thrust device and spoilers start to operate, for a total of 5s. At t=8387s, the spoilers open, and 8382s-8392s is the transition period, for a total of 10s. The data of the transition is sampled at 1Hz.

[0160] Take the aircraft's heading angle data during the transition period and calculate the average value as the runway reference azimuth.

[0161] 4. Deceleration phase: from the end of the transition phase at t = 8392s, when the aircraft touches down completely and the reverse thrust device and spoilers start to operate, until the aircraft heading angle deviates from the runway reference azimuth by more than or equal to 90 degrees, which is the time period of the deceleration phase, a total of 48s;

[0162] In this embodiment, the basis for data interception and extraction is shown in Table 1:

[0163] Table 1 Data extraction basis (approach section + air section + transition section + deceleration section)

[0164]

[0165] Step S6, extracting the landing phase data, archiving the intercepted data, and integrating them into output pre-processed data;

[0166] Although the various steps in the above embodiment are described in the above-mentioned order, those skilled in the art can understand that in order to achieve the effect of this embodiment, different steps do not have to be executed in such an order. They can be executed simultaneously (in parallel) or in a reverse order. These simple changes are within the scope of protection of the present invention.

[0167] The flight parameter preprocessing system based on QAR data provided by the present invention comprises:

[0168] a data extraction module configured to extract data from the QAR;

[0169] A data frequency synchronization module is configured to perform frequency synchronization on the extracted data to obtain frequency synchronized data so that the data has the same time series;

[0170] A parameter identification module is configured to perform parameter identification on the frequency-synchronized data to obtain required data;

[0171] Based on the data after parameter identification, singular value detection and correction, data filtering, and data filling are performed to obtain cleaned and repaired data;

[0172] A data extraction and interception module is configured to extract and intercept data based on the cleaned and repaired data using an aircraft phase identification and data extraction method;

[0173] The data output module is configured to extract the landing phase data, archive the intercepted data, and integrate them into output pre-processed data.

[0174] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working process and related instructions of the system described above can refer to the corresponding process in the aforementioned method embodiment, and will not be repeated here.

[0175] It should be noted that the QAR data preprocessing method system provided in the above embodiment is only illustrated by the division of the above functional modules. In practical applications, the above functions can be assigned to different functional modules as needed, that is, the modules or steps in the embodiments of the present invention can be decomposed or combined. For example, the modules in the above embodiments can be combined into one module, or further divided into multiple sub-modules to complete all or part of the functions described above. The names of the modules and steps involved in the embodiments of the present invention are only for distinguishing the modules or steps, and are not regarded as improper limitations of the present invention.

[0176] The terms "first", "second", etc. are used to distinguish similar objects rather than to describe or indicate a particular order or sequence.

[0177] The term "comprise" or any other similar term is intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus / device that includes a list of elements includes not only those elements but also other elements not expressly listed, or also includes elements inherent to such process, method, article, or apparatus / device.

[0178] So far, the technical solutions of the present invention have been described in conjunction with the preferred embodiments shown in the accompanying drawings. However, it is easy for those skilled in the art to understand that the protection scope of the present invention is obviously not limited to these specific embodiments. Without departing from the principle of the present invention, those skilled in the art can make equivalent changes or substitutions to the relevant technical features, and the technical solutions after these changes or substitutions will fall within the protection scope of the present invention.

Claims

1. A flight parameter preprocessing method based on QAR data, characterized in that: The method comprises: Step S1, extracting data from a quick access recorder (QAR); Step S2, performing frequency synchronization processing on the extracted data to obtain frequency synchronized data; Step S3, performing parameter identification on the frequency-synchronized data to obtain required data; Step S4, performing singular value detection and correction, data filling and data filtering on the data after parameter identification to obtain cleaned and repaired data; Step S5, extracting and intercepting the cleaned and repaired data using an aircraft phase identification and data extraction method; Step S6, extracting the landing phase data, archiving the intercepted data, and integrating them into output pre-processed data.

2. The flight parameter preprocessing method based on QAR data according to claim 1, characterized in that: The frequency synchronization process is highly accurate and flexible, and the specific implementation method is as follows: First, perform time synchronization processing, that is, accurately match the reference frequency recorded in the QAR data version library to determine the original acquisition frequency of each parameter data; on this basis, use 1HZ as the reference frequency limit to clearly divide the data into two categories: high frequency and low frequency; For high-frequency data, a refined frequency reduction processing strategy is implemented to ensure that the data is frequency-reduced according to the 1HZ frequency standard without losing key information; for the redundant frequency information generated during the frequency reduction process, a majority renaming mechanism is used to properly preserve it, thereby ensuring that the integrity and authenticity of the original data are maintained; For low-frequency data, scientific frequency-filling technology is used to appropriately expand the data according to the 1HZ frequency standard; In this process, a blank area for data filling is specially reserved to provide convenient conditions for subsequent data filling and filtering processing, thereby ensuring the continuity and consistency of the data in the time series; Through the above-mentioned frequency-synchronized data processing flow, the present invention realizes the effective integration and unification of data with different frequencies, laying a solid foundation for subsequent data analysis work.

3. The flight parameter preprocessing method based on QAR data according to claim 1, characterized in that: The data cleaning and repairing includes: Singular value detection and correction: Use residual analysis based on Kalman filtering to test and eliminate singular values; Data filling; Data filtering; Among them, in the data filling and data filtering part, for data with high-frequency oscillation problems, the Kalman filter method is used for denoising and smoothing; For data that changes steadily and needs filtering, Savitzky-Golay (SG) filtering is used for denoising and smoothing; For small missing data, Lagrange interpolation method is used to fill in the missing data, and the formula is: Among them, P n (x) is the constructed interpolation polynomial, y i is the function value corresponding to a given data point, l j (x) is the basis function, defined as: Among them, x0, x j-1 、x j 、x j+1 、x n are the horizontal coordinates of the interpolation points; The n-degree polynomial P constructed in this way n x passes through all given data points and has good smooth interpolation characteristics, avoiding the problem of high-order polynomial operation, and the calculation is relatively simple; During the data processing, for data that contains missing values ​​and needs to be filtered, the above-mentioned Lagrange interpolation method is used to fill the missing values ​​first and then perform the SG filtering operation; this processing order ensures the continuous data that the SG filter relies on and avoids inaccurate calculations caused by missing values. By pre-filling the missing values, the filter is prevented from producing boundary effects and the consistency and smoothness of the data are ensured. This process reduces the error or noise introduced by interpolation and improves the overall accuracy of the data and the filtering effect.

4. The flight parameter preprocessing method based on QAR data according to claim 3, characterized in that: The Kalman filtering method includes: Establish the state equation and observation equation: Among them, x k is the system state, x k-1 is the state of the system at the previous moment, A is the state transfer matrix, and B is the control input matrix u k-1 is the control input, w k-1 is the process noise, z k is the estimated value, H is the observation matrix, v k is the measurement noise; To make a status prediction: X k =AX k-1 +Bu k-1 4 Among them, X k is the state prediction at time k, X k-1 is the state estimate at the previous moment; Forecast error covariance: P k =AP k-1 From T +Q; Among them, P k is the prediction error covariance, P k-1 is the covariance of the previous moment, Q is the covariance of the process noise; Kalman gain calculation: K k =P k H T (HP k H T +R) -1 ; Among them, K k is the Kalman gain matrix, R is the covariance of the measurement noise; Status Update: X k+1 =X k +K k (z k -HX k ); Error covariance update: P k+1 =(I-K k H)P k ; Among them, x k+1 is the updated system status, P k+1 is the updated system status; Through the above recursive prediction and state updating process, the optimal estimated value of the state variable based on Kalman filtering can be obtained.

5. The flight parameter preprocessing method based on QAR data according to claim 3, characterized in that: The Kalman filter-based residual analysis, the acquisition method thereof comprises: Compute the residual: Y k =z k -HX k ; Among them, Y k is the residual; Calculate the innovation covariance: S k =HP k H T +R; Among them, S k is the innovation covariance; Standardized residual calculation: Among them, v k is the standardized residual; Outlier Detection: Determine z k is a singular value; in, is the residual threshold; through the above residual calculation and judgment process, it can be determined whether there are singular values ​​in the data and the location of the singular values ​​can be determined.

6. The flight parameter preprocessing method based on QAR data according to claim 3, characterized in that: The Savitzky-Golay (SG) filtering method includes: The specific formula is: Among them, y′ i is the smoothed data point, y i is the original data point, c j is the Savitzky-Golay filter coefficient, m is half the width of the window, and the window length is 2m+1; Among them, c j The calculation involves constructing a Vandermonde matrix and solving its pseudo-inverse, which can be done in linear algebra: Where n is the order of the fitting polynomial; Assume that the pseudo-inverse of matrix A is A + , then the filter coefficient c can be expressed as: c=A + e0; Among them, e0 is a unit vector, which represents the smoothing weight of the central point of the window.

7. The flight parameter preprocessing method based on QAR data according to claim 1, characterized in that: The cleaned and repaired data is extracted and intercepted using the aircraft phase recognition and data extraction method. The specific implementation method is as follows: First, by analyzing the key flight parameters in the QAR data, including radio altitude, speed, pitch angle, heading angle, reverse thrust device status, etc., combined with the preset flight phase standards, the various phases of the flight landing process are identified; the phase identification process divides the data of the flight landing process into four main phases: approach phase, air phase, transition phase and deceleration phase. The specific characteristics of each phase division include: Approach phase: The criterion for identifying the approach phase is that the radio altitude during the landing phase is between 1500 feet and 50 feet; Air segment: The criterion for identifying the air segment is that the radio altitude during the landing phase is between 50 feet and 0 feet; Transition phase: The standard for identifying the transition phase is that the radio altitude during the landing phase is 0 feet or negative, which means that the aircraft has touched down, until the main wheels and nose wheels of the aircraft are completely on the ground and the engine conversion ratio (EPR) and exhaust gas temperature (EGT) data change, and the reverse thrust device and spoilers begin to operate; Deceleration segment: The standard for identifying the deceleration segment is that the aircraft is completely touched down and the reverse thrust device and spoiler start to operate until the aircraft heading angle deviates from the runway reference azimuth by 90 degrees; The runway reference azimuth in the deceleration section is an average value of the heading angle data of the aircraft in the transition section.

8. A flight parameter preprocessing system based on QAR data, characterized in that: The system comprises: a data extraction module configured to extract data from the QAR; A data frequency synchronization module is configured to perform frequency synchronization on the extracted data to obtain frequency synchronized data so that the data has the same time series; A parameter identification module is configured to perform parameter identification on the frequency-synchronized data to obtain required data; Based on the data after parameter identification, singular value detection and correction, data filtering, and data filling are performed to obtain cleaned and repaired data; A data extraction and interception module is configured to extract and intercept data based on the cleaned and repaired data using an aircraft phase identification and data extraction method; The data output module is configured to extract the landing phase data, archive the intercepted data, and integrate them into output pre-processed data.

Images