A method and system for diagnosing and correcting abnormal data in an aircraft atmospheric data system

Abstract

This invention discloses a method and system for diagnosing and correcting abnormal data in an aircraft atmospheric data system. The method includes: Step 1) collecting pressure data measured at each pressure gauge port of the aircraft; Step 2) obtaining the pressure deviation value of the corresponding pressure gauge port based on the known true value of each pressure gauge port, and determining whether the pressure gauge port is working normally by comparing it with a pre-set threshold value; Step 3) fitting the corresponding pressure data obtained in Step 1) to obtain a polynomial curve for all pressure gauge ports determined to be normal, and judging whether there is abnormal data based on the monotonicity; Step 4) removing abnormal data and reconstructing the data using the weighted least squares method. This invention has higher purposefulness and reliability in correction; the method is simple; it does not add extra hardware and software facilities, and does not occupy valuable aircraft payload and internal space; detection and diagnosis begin with measurement data, avoiding abnormal data from directly participating in system calculation, and proactively avoiding and mitigating risks.

Description

Technical Field

[0001] This invention belongs to the field of atmospheric data system fault detection technology, and particularly relates to a method and system for diagnosing and correcting abnormal data in an aircraft atmospheric data system. Background Technology

[0002] An atmospheric data system is an essential measurement device used by aircraft in real-time to sense environmental conditions and measure flight speed, attitude, and altitude. It is crucial and indispensable for ensuring flight safety and improving flight quality. Pitot tubes, wind vanes, and the later-developed FADS system are all specific manifestations of the sensing equipment in an atmospheric data system. Among these, pressure measurement at characteristic areas / locations on the aircraft surface is one of the most fundamental and critical technical aspects of atmospheric data system design and use. Only by accurately measuring the pressure data at these characteristic areas / locations can the aircraft's true flight status (speed, attitude angles, altitude, etc.) be retrieved through subsequent calculations.

[0003] During actual flight, aircraft inevitably encounter abnormal flight conditions such as sudden gusts, surface icing and water accumulation, and severe convective weather. These conditions can range from minor turbulence causing injury to serious crashes and loss of life. Furthermore, in near-space (20-100 km above the Earth's surface) and other high-altitude flight conditions, the surface pressure of aircraft flying in a thin gas environment is difficult to measure accurately (ambient pressure is approximately tens of Pa), which can also lead to malfunctions in the atmospheric data system. These problems are unavoidable practical issues during aircraft flight; therefore, it is necessary to conduct abnormal data diagnosis, isolation, and fault-tolerant design based on basic aerodynamic calculation algorithms to improve system reliability.

[0004] Based on problems discovered during practical use, researchers have developed various methods and technologies, including the chi-square test, multi-sensor data fusion, and neural networks. Among these, the chi-square test-based method requires statistical analysis of a large amount of atmospheric measurement data to derive the measurement residual patterns. It may require several iterations to detect fault points and achieve final convergence. Its algorithm and mechanism are complex, significantly impacting real-time performance, and its implementation in airborne systems is challenging. Multi-sensor data fusion methods, based on non-atmospheric measurement principles of GPS / INS, suffer from limitations due to system complexity. Neural network methods offer unique advantages for solving coupled systems with multiple parameter inputs and outputs, but the unpredictability of network node parameters limits their application, generally limiting them to post-event evaluation and simulation.

[0005] The general approach to designing an aircraft atmospheric data system is as follows: based on the flow characteristics of the aircraft surface, in areas where the flow on the aircraft surface is relatively stable, the sensitivity of atmospheric parameters to pressure data is studied, a small number of locations are selected to form an embedded atmospheric data sensing system, and an aerodynamic mathematical model between the pressure at the measuring point and the incoming atmospheric parameters is constructed.

[0006] P i =F i (Ma,P ∞ ,α,β)

[0007] Currently, when designing an atmospheric data system for an aircraft, CFD and wind tunnel testing techniques are mainly used to obtain pressure distribution data on the surface of a specific area of ​​the aircraft. Then, the placement of the atmospheric data system's sensors is selected, and an atmospheric data parameter calculation database is constructed. By solving a specific set of equations and combining them with the calculation database, the atmospheric data parameters during flight are finally obtained.

[0008] Let X be an estimated atmospheric parameter, and W be noise.

[0009] X = [Ma, P] ∞ ,α,β] T P = F(X) + W

[0010] Linearization of F(X):

[0011] P = F(X0) + HΔX + W

[0012] Where X0 is the prior value, ΔX = X - X0, and H is the Jacobi matrix.

[0013] Then the optimal linear minimum variance unbiased estimate of ΔX is:

[0014] ΔX=(H T S -1 H) -1 H T S -1 (PF(X0))

[0015] After obtaining ΔX, the prior value is redefined: X0 = X0 + ΔX.

[0016] By combining the previously constructed solution database, atmospheric parameter values ​​can be obtained through iterative calculations.

[0017] The data sources in this computational database are highly "ideal." This "ideal" means that the data are obtained from CFD and wind tunnel tests—conditions that are precisely controlled and free from unexpected or disruptive factors. Based on this database, the computational accuracy can be guaranteed under stable flight conditions with minimal external environmental fluctuations. However, under the aforementioned abnormal conditions, the obtained atmospheric data parameters will be significantly distorted, leading to flight anomalies.

[0018] Inaccurate measurements inevitably lead to inaccurate calculations. If clearly abnormal data is used in a "normal" calculation database, the result will inevitably be inaccurate and unrealistic data. Summary of the Invention

[0019] The purpose of this invention is to overcome the shortcomings of the prior art and to propose a method and system for diagnosing and correcting abnormal data in an aircraft atmospheric data system.

[0020] To achieve the above objectives, this invention proposes a method for diagnosing and correcting abnormal data in an aircraft atmospheric data system, the method comprising:

[0021] Step 1) Collect pressure data measured at each pressure port of the aircraft;

[0022] Step 2) Based on the known true value of each pressure test hole, obtain the pressure deviation value of the corresponding pressure test hole, and determine whether the pressure test hole is working properly by comparing it with the preset threshold value.

[0023] Step 3) For all pressure test holes that are determined to be normal, fit the corresponding pressure data obtained in Step 1) to obtain a polynomial curve, and determine whether there is abnormal data based on the monotonicity.

[0024] Step 4) Remove outlier data and reconstruct the data using weighted least squares.

[0025] Preferably, the known true value of each pressure measuring hole is obtained in advance through CFD and wind tunnel tests.

[0026] Preferably, the pressure deviation value for the corresponding pressure measuring hole is the difference between the pressure data measured at each pressure measuring hole and the known true value of that pressure measuring hole, and the absolute value is taken.

[0027] Preferably, in step 2), the operation of the pressure testing hole is determined by comparing it with a preset threshold value; specifically, this includes:

[0028] The pressure deviation value of the pressure measuring hole is compared with the preset threshold value. If the comparison result is greater than the threshold value, the pressure measuring hole is determined to be malfunctioning; otherwise, it is malfunctioning.

[0029] Preferably, step 3) involves determining whether there is abnormal data based on monotonicity; specifically, this includes:

[0030] For a polynomial curve, the derivative is calculated point by point. If there is a sudden change in the sign of the derivative before and after a certain point, the data at that point is determined to be abnormal data.

[0031] Preferably, step 4) specifically includes:

[0032] The pressure values ​​P of n pressure gauges at a certain moment in flight state are determined by the following formula. n :

[0033]

[0034] In the formula, B(x,y) n×m Let A be the polynomial of each degree in a system of bivariate polynomial equations, m denotes the coefficient dimension, x is the coefficient of the bivariate polynomial representing the numerical value of the surface pressure in the x-direction, y is the coefficient of the bivariate polynomial representing the numerical value of the pressure in the y-direction, j and k are the highest powers of the bivariate polynomials, both taking the value of 3, and A is the highest power of the polynomial. m These are the coefficients of the system of two-variable polynomial equations to be determined.

[0035] According to the least squares criterion for filtering, the following equation is satisfied to obtain A. m The optimal estimate in the least squares sense

[0036] (B(x,y) n×m A m -P n ) T F n×n (B(x,y) n×m A m -P n ) = Min

[0037] Among them, F n×n Let be the weight matrix, defined as 1 for non-fault points and 0 for fault points; T represents the transpose, and Min represents the minimum sum of squares of the differences between the measured values ​​and the optimal estimate at each point.

[0038] After removing outliers, the least squares parabolic fit result is as follows:

[0039]

[0040] Obtain data reconstruction values

[0041]

[0042] On the other hand, this invention proposes an abnormal data diagnosis and correction system for an aircraft atmospheric data system, implemented based on the above method, the system comprising:

[0043] The data acquisition module is used to collect pressure data measured at each pressure port of the aircraft.

[0044] The pressure test hole determination module is used to obtain the pressure deviation value of the corresponding pressure test hole based on the known true value of each pressure test hole, and to determine whether the pressure test hole is working properly by comparing it with a preset threshold value.

[0045] The abnormal data detection module is used to fit the corresponding pressure data obtained by the acquisition module to all pressure measurement holes determined to be normal, obtain a polynomial curve, and determine whether there is abnormal data based on the monotonicity; and

[0046] The reconstruction module is used to remove abnormal data and reconstruct the data using weighted least squares.

[0047] Compared with the prior art, the advantages of the present invention are:

[0048] 1. The method of the present invention is based on a known (trustworthy) database to correct or even reconstruct abnormal data, and the correction is more targeted and reliable.

[0049] 2. The method of the present invention is simple;

[0050] 3. The method of this invention does not require additional hardware or software facilities, nor does it occupy valuable aircraft payload and internal space;

[0051] 4. The method of the present invention starts with measurement data for detection and diagnosis, avoiding abnormal data from directly participating in system calculation, thus avoiding and mitigating risks in advance. Attached Figure Description

[0052] Figure 1 This is a schematic diagram of the pressure distribution of an aircraft under normal conditions;

[0053] Figure 2 This is a diagram illustrating abnormal data;

[0054] Figure 3(a) shows the data under normal conditions, and Figure 3(b) shows the data under abnormal conditions;

[0055] Figure 4 This is a schematic diagram of the reconstructed data;

[0056] Figure 5 This is a flowchart of the method in Embodiment 1 of the present invention. Detailed Implementation

[0057] The method of the present invention identifies the rationality and reliability of the measured data, and then eliminates the obviously abnormal data so that it is not allowed to participate in the parameter calculation; or, the measured abnormal data is corrected as necessary to return it to the normal value, and then the atmospheric data parameters are calculated to obtain reasonable atmospheric data parameters.

[0058] Conducting CFD and wind tunnel tests on a specific aircraft is an essential step in designing an atmospheric data system. These techniques allow us not only to acquire pressure data for specific regions of the aircraft, enabling the construction of a computational database, but also to comprehensively understand the pressure distribution characteristics, patterns, and even specific numerical values ​​in characteristic areas of the aircraft. These characteristics, patterns, and numerical values ​​can be considered "true values," forming a known database that serves as the basis for subsequent identification, isolation, and removal of outlier data.

[0059] This section uses the pressure distribution at the nose of a typical blunt-nosed aircraft and the design and correction of its atmospheric data system as an example. Through prior CFD simulation analysis and wind tunnel experiments, the pressure distribution characteristics and specific data of the aircraft's characteristic region (generally the nose region) can be obtained. Generally, this pressure distribution (displayed as a pressure coefficient Cp) will exhibit certain patterns, such as... Figure 1 As shown, this pattern and specific data are the "true values" and basis for judging whether the actual measurement values ​​during flight are "out of line".

[0060] When the data measured by the sensing equipment of the in-flight atmospheric data system deviates from this "true value" distribution pattern, such as... Figure 2 As shown, it should be processed through software and hardware techniques such as identification, elimination, correction, and reconstruction.

[0061] 1. Fault point detection and diagnosis

[0062] 1.1 Fault point detection / identification—threshold and threshold detection

[0063] like Figure 2 As shown, when the measured data deviates significantly from the "true value," the deviation is compared with a pre-set "threshold value." If this deviation exceeds the threshold value, the pressure measurement port is considered malfunctioning, and the obtained data is unreliable. This "threshold value" can be determined through actual flight testing, acquiring a large amount of data. Based on practical experience, this threshold value should be set moderately leniently, not too strictly, while ensuring that the data does not deviate significantly and flight safety is maintained; otherwise, the flight control system will frequently report errors, which is detrimental to actual use.

[0064] 1.2 Fault Point Detection / Discrimination—Monotonicity / Derivative Detection

[0065] As can be seen from the pressure distribution characteristics of the surface, the pressure distribution in a certain direction can be fitted as a polynomial curve. According to CFD simulation calculations and wind tunnel test results, fitting it as a cubic polynomial parabola can generally ensure its accuracy.

[0066] Except for singular points (or points where the derivative is zero), the polynomial curve of pressure distribution in a certain direction will not show abrupt changes in trend—on both sides of a data point, the data exhibits a definite monotonically increasing or decreasing characteristic, as shown in Figure 3(a). Based on this characteristic, the validity of the data can be determined by taking the derivative before and after a certain data point and observing the change in its sign; furthermore, the normality or abnormality of the fault point can be determined. Figure 3(b) shows the data under abnormal conditions.

[0067] 2. Fault data correction / reconstruction

[0068] After using the above methods to detect and identify fault points, any possible fault points and their data should be removed; and weighted least squares method should be used for stress reconstruction.

[0069] Let's take a bivariate polynomial of multiple degree as an example. Its expression is:

[0070]

[0071] In the formula, P n Let B(x,y) represent the pressure values ​​at each pressure measurement point under different flight conditions. n×m Let be the polynomials of a system of bivariate polynomial equations. j and k represent the highest power of the bivariate polynomials; generally, a power of 3 is sufficient to meet the accuracy requirements. A m These are the coefficients of the system of two polynomial equations that we need to find.

[0072] According to the least squares criterion of filtering, when If:

[0073] (B(x,y) n×m A m -P n ) T F n×n (B(x,y) n×m A m -P n ) = Min

[0074] Then it can be called It is A m The optimal estimate of in the least squares sense.

[0075] Among them, F n×nThe weight matrix assigns weights to the fault locations detected by pressure measurement faults in Section 1.2, defining "1" as usable and "0" as usable. After removing "bad points," the least-squares parabolic fitting result is:

[0076]

[0077] This completes the reconstruction of the removed "bad pixels":

[0078]

[0079] Completed reconstructed data, such as Figure 4 As shown, the data has now been corrected to a reasonable range. The corrected data can then be used to calculate atmospheric parameters, providing reliable input parameters for the flight control system.

[0080] The technical solution of the present invention will be described in detail below with reference to the accompanying drawings and embodiments.

[0081] Example 1

[0082] like Figure 5 As shown, Embodiment 1 of the present invention proposes a method for diagnosing and correcting abnormal data in an aircraft atmospheric data system, the method comprising:

[0083] Step 1) Collect pressure data measured at each pressure port of the aircraft.

[0084] Step 2) Based on the known true value of each pressure test hole, obtain the pressure deviation value of the corresponding pressure test hole. Compare it with the preset threshold value. If the comparison result is greater than the threshold value, the pressure test hole is determined to be malfunctioning. Otherwise, it is malfunctioning.

[0085] The known true values ​​for each pressure gauge hole are obtained in advance through CFD and wind tunnel tests.

[0086] The pressure deviation value for each pressure test hole is the difference between the pressure data measured at each pressure test hole and the known true value of that pressure test hole, and the absolute value is taken.

[0087] Step 3) For all pressure testing holes determined to be normal, fit the corresponding pressure data obtained in Step 1) to obtain a polynomial curve, and determine whether there is abnormal data based on the monotonicity. Specifically:

[0088] For a polynomial curve, the derivative is calculated point by point. If there is a sudden change in the sign of the derivative before and after a certain point, the data at that point is determined to be abnormal data.

[0089] Step 4) Remove outlier data and reconstruct the data using weighted least squares. Specifically:

[0090] The pressure values ​​P of n pressure gauges at a certain moment in flight state are determined by the following formula. n :

[0091]

[0092] In the formula, B(x,y) n×m Let A be the polynomial of each degree in a system of bivariate polynomial equations, m denotes the coefficient dimension, x is the coefficient of the bivariate polynomial representing the numerical value of the surface pressure in the x-direction, y is the coefficient of the bivariate polynomial representing the numerical value of the pressure in the y-direction, j and k are the highest powers of the bivariate polynomials, both taking the value of 3, and A is the highest power of the polynomial. m These are the coefficients of the system of two-variable polynomial equations to be determined.

[0093] According to the least squares criterion for filtering, the following equation is satisfied to obtain A. m The optimal estimate in the least squares sense

[0094] (B(x,y) n×m A m -P n ) T F n×n (B(x,y) n×m A m -P n ) = Min

[0095] Among them, F n×n Let be the weight matrix, defined as 1 for non-fault points and 0 for fault points; T represents the transpose, and Min represents the minimum sum of squares of the differences between the measured values ​​and the optimal estimate at each point.

[0096] After removing outliers, the least squares parabolic fit result is as follows:

[0097]

[0098] Obtain data reconstruction values

[0099]

[0100] Example 2

[0101] Embodiment 2 of the present invention proposes a correction system for anomaly data diagnosis and correction method of aircraft atmospheric data system, which is based on the method of Embodiment 1. The system includes:

[0102] The data acquisition module is used to collect pressure data measured at each pressure port of the aircraft.

[0103] The pressure test hole determination module is used to obtain the pressure deviation value of the corresponding pressure test hole based on the known true value of each pressure test hole, and to determine whether the pressure test hole is working properly by comparing it with a preset threshold value.

[0104] The abnormal data judgment module is used to fit the corresponding pressure data obtained by the acquisition module to all pressure measurement holes that are judged to be normal, and obtain a polynomial curve, and judge whether there is abnormal data based on the monotonicity.

[0105] The reconstruction module is used to remove abnormal data and reconstruct the data using weighted least squares.

[0106] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to the embodiments, those skilled in the art should understand that modifications or equivalent substitutions to the technical solutions of the present invention do not depart from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. An aircraft atmospheric data system abnormal data diagnosis correction method, the method comprising: Step 1) collecting pressure data measured by each pressure hole of an aircraft; Step 2) obtaining a pressure deviation value of a corresponding pressure hole according to a known true value of each pressure hole, and determining whether the work of the pressure hole is normal by comparing with a pre-set threshold value; Step 3) for all pressure holes determined to be normal, fitting the corresponding pressure data obtained in Step 1) to obtain a polynomial curve, and determining whether there is abnormal data according to monotonicity; Step 4) removing abnormal data and reconstructing data using a weighted least square method; the Step 4) specifically comprises: The pressure values P of the n pressure holes in the flight state at a certain time are determined according to the following formula n : In the formula, B(x, y) n×m is each polynomial of the binary polynomial equation group, m represents the coefficient dimension, x is the binary polynomial coefficient, represents the value of the surface pressure in the x direction, y is the binary polynomial coefficient, represents the value of the pressure in the y direction, j and k are the highest powers of the binary polynomial, both of which are 3, A m is the coefficient of the binary polynomial equation group to be solved; According to the least square criterion of filtering, A is obtained by satisfying the following equation m Optimal estimate in the least square sense (B(x, y) n×m A m -P n ) T F n×n (B(x, y) n×m A m -P n ) = Min where F n×n is a weight matrix, defined as 1 for non-faulty points and 0 for faulty points; T denotes transposition, and Min denotes the minimum of the sum of squares of the difference between the measured values and the optimal estimate values; After removing the abnormal data, the least square parabolic fitting result solved is: obtaining data reconstruction values 2. The aircraft air data system aberrant data diagnostic correction method of claim 1 wherein, The known true value of each pressure hole is obtained in advance by CFD and wind tunnel test.

3. The aircraft air data system aberrant data diagnostic correction method of claim 2 wherein, The pressure deviation value of the corresponding pressure hole is the difference between the pressure data measured by each pressure hole and the known true value of the pressure hole, and the absolute value is taken.

4. The aircraft atmospheric data system aberrant data diagnostic correction method of claim 3 wherein, In the Step 2), whether the work of the pressure hole is normal is determined by comparing with a pre-set threshold value; specifically comprising: The pressure deviation value of the pressure hole is compared with the pre-set threshold value, and when the comparison result is greater than the threshold value, it is determined that the work of the pressure hole is abnormal, otherwise, it is normal.

5. The aircraft atmospheric data system aberrant data diagnostic correction method of claim 1 wherein, In the Step 3), whether there is abnormal data is determined according to monotonicity; specifically comprising: For the polynomial curve, the derivative is calculated point by point, and when the sign of the derivative on both sides of a point suddenly changes, the data of the point is determined to be abnormal data.

6. A correction system based on the method of diagnosing and correcting abnormal data of an aircraft atmospheric data system according to any one of claims 1 to 5, characterized in that, The system comprises: a collection module for collecting pressure data measured by each pressure hole of an aircraft; a pressure hole determination module for obtaining a pressure deviation value of a corresponding pressure hole according to a known true value of each pressure hole, and determining whether the work of the pressure hole is normal by comparing with a pre-set threshold value; an abnormal data determination module for fitting the corresponding pressure data obtained by the collection module to obtain a polynomial curve for all pressure holes determined to be normal, and determining whether there is abnormal data according to monotonicity; and a reconstruction module for removing abnormal data and reconstructing data using a weighted least square method.

Images