High-precision shock-resistant pressure sensing signal processing method suitable for aviation scenarios

Abstract

The application discloses a high-precision anti-impact pressure sensing signal processing method suitable for an aviation scene and belongs to the technical field of aviation pressure sensing signal processing. The method deploys a distributed pressure sensing array on the surface of an aircraft, and collects original pressure signals in real time. A transient impact feature separation operation is performed on the original pressure signals of each pressure sensor, and the original pressure signals are decomposed into high-frequency impact components and low-frequency aerodynamic components. According to flight parameters, the low-frequency aerodynamic components at different positions are subjected to spatial gradient reconstruction, and a dynamic pressure distribution atlas of the surface of the aircraft is generated. The dynamic pressure distribution atlas is used to perform energy normalization mapping on the high-frequency impact components, and impact energy coefficients are obtained. Based on the impact energy coefficients and the dynamic pressure distribution atlas, the effective pressure values in the original pressure signals are subjected to nonlinear weighted correction, and the corrected anti-impact pressure signals are output.

Description

Technical Field

[0001] This invention relates to the field of aviation pressure sensing signal processing technology, specifically a high-precision shock-resistant pressure sensing signal processing method suitable for aviation scenarios. Background Technology

[0002] During flight, aircraft surface pressure sensors are subjected to a combination of aerodynamic loads and transient mechanical shocks. Transient shocks originate from structural flutter, foreign object impacts, or localized pressure pulsations caused by flow separation, and their frequency components significantly overlap with the effective aerodynamic pressure frequency band. Existing pressure signal processing methods generally employ low-pass filtering or wavelet thresholding denoising techniques, treating high-frequency components as noise and eliminating or attenuating them. These methods ignore the spatial distribution characteristics of impact energy and the correlation between impact response and real-time flight status, resulting in amplitude distortion and phase lag in the filtered signal during the impact period, failing to reflect true aerodynamic pressure changes. Especially when sensors are deployed in a distributed array, the intensity and time-frequency characteristics of the impacts experienced at each measuring point differ significantly. Single time-frequency domain processing cannot utilize spatial redundancy information for coordinated correction, leading to inconsistent pressure signal correction accuracy at different locations, thus affecting the accuracy of the flight control system's perception of the surface pressure situation.

[0003] Under high-speed flight conditions, impact disturbances in key areas such as the wing leading edge and inlet lip are closely coupled with the local aerodynamic pressure gradient. The impact energy exhibits an unsteady distribution due to the shift of airflow stagnation points and the movement of the separation zone. Existing methods struggle to adjust the spatial reconstruction model of the pressure field in real time based on flight parameters, and cannot characterize the mapping relationship between impact energy and pressure gradient under dynamic aerodynamic environments. This results in a lack of physical field constraints in the impact mitigation process, leading to deviations between the corrected signal and the actual flow physics. Therefore, how to quantitatively describe and implement targeted corrections of impact disturbances based on global information from flight parameters and the spatial pressure field while preserving high-frequency aerodynamic characteristics is a problem that needs to be solved to improve the accuracy of airborne pressure sensing signals. Summary of the Invention

[0004] The technical problem to be solved by the present invention is to provide a high-precision shock-resistant pressure sensing signal processing method suitable for aviation scenarios. Under complex flight conditions, it can comprehensively utilize the spatial layout information of distributed pressure sensors and real-time flight parameters to realize spatial constraint quantification of impact energy and adaptive nonlinear correction of pressure signals, thereby improving the accuracy of pressure signals after shock-resistant processing.

[0005] To achieve the above objectives, the present invention provides the following technical solution: The present invention provides a high-precision shock-resistant pressure sensing signal processing method suitable for aviation scenarios. This method, through the coordinated operation of a distributed pressure sensing array and a multi-level signal processing algorithm, solves the problems of pressure sensing signal distortion and pressure distribution misjudgment caused by transient impact events such as bird strikes, hail, and shock wave oscillations during high-speed aircraft flight.

[0006] In one technical solution of this invention, multiple pressure sensors are deployed on the surface of an aircraft to form a distributed pressure sensing array, which collects raw pressure signals of the aircraft in different flight attitudes in real time. Preferably, based on the predicted aerodynamic load distribution map of the aircraft surface, sensor deployment points are set in areas with concentrated aerodynamic loads and high impact risk, such as the wing leading edge, fuselage nose, and air intake lip. Each deployment point is equipped with a piezoresistive pressure sensor. All piezoresistive pressure sensors are connected to a data preprocessing unit through a synchronous acquisition circuit, which allocates the same sampling clock signal to each pressure sensor. The data preprocessing unit triggers all pressure sensors to simultaneously collect pressure data according to the rising edge of the sampling clock signal. All pressure data collected in each sampling period constitutes a raw pressure signal frame, which contains the raw pressure signal sample value of each pressure sensor at that moment. Through the synchronous acquisition mechanism, the instantaneous consistency of the spatial pressure distribution is ensured, providing an accurate time reference for subsequent spatial gradient reconstruction.

[0007] For the raw pressure signals acquired by each pressure sensor, a transient impact feature separation operation is performed to decompose the raw pressure signals into high-frequency impact components and low-frequency aerodynamic components. The raw pressure signal sequence of each pressure sensor within a continuous time window is acquired, and local extremum point detection is performed on this sequence to identify signal abrupt changes where the amplitude change rate exceeds a preset threshold. Using each signal abrupt change point as a dividing boundary, the raw pressure signal sequence is divided into multiple impact event intervals and non-impact event intervals. Empirical mode decomposition is performed on the signal data within each impact event interval to extract oscillation modes with frequencies higher than the aircraft structure's natural frequencies as high-frequency impact components. The signal data within the non-impact event intervals is then concatenated with the residual signal within the impact event intervals after removing the high-frequency impact components to form the low-frequency aerodynamic components. This separation method can accurately identify and separate local high-frequency oscillations caused by transient impacts, while completely preserving the effective low-frequency information reflecting aerodynamic load changes, effectively avoiding the contamination of the aerodynamic component's authenticity by impact interference.

[0008] As a preferred approach, during the empirical mode decomposition process, the signal data within each impact event interval is first subjected to mirror continuation processing to suppress endpoint effects, and then the intrinsic mode function is extracted as the high-frequency impact component. This avoids the boundary oscillation distortion commonly seen in finite-duration signal decomposition and improves the extraction accuracy of the high-frequency impact component.

[0009] Based on the aircraft's current flight parameters, spatial gradient reconstruction is performed on the low-frequency aerodynamic components corresponding to pressure sensors at different locations to generate a dynamic pressure distribution map of the aircraft surface. The aircraft's current flight parameters, including Mach number, angle of attack, and sideslip angle, are read, and the three-dimensional spatial coordinates of each pressure sensor in the distributed pressure sensor array on the aircraft surface are obtained. Using the three-dimensional spatial coordinates of each pressure sensor as interpolation nodes and the corresponding low-frequency aerodynamic components as node values, a radial basis function interpolation method is used to construct the initial pressure field of the aircraft surface. The locations of airflow stagnation points and separation zone boundaries on the aircraft surface are determined based on the flight Mach number, angle of attack, and sideslip angle. Anisotropic constraints are applied to the initial pressure field using these locations, and the weighting coefficients of the interpolation nodes are adjusted to generate the dynamic pressure distribution map. This spatial reconstruction method, which integrates flight parameters, ensures that the pressure distribution map accurately reflects the aerodynamic effects under the aircraft's current flight attitude, maintains high pressure gradient resolution near stagnation points, and preserves clear transition characteristics at the separation zone boundaries.

[0010] Preferably, an adaptive shape parameter is introduced into the radial basis function interpolation method. This adaptive shape parameter is dynamically adjusted according to the spatial distance between adjacent pressure sensors, making the support radius of the interpolation basis function inversely proportional to the local density of the sensors. This enhances the interpolation locality in dense sensor regions, avoiding long-distance smoothing effects; and expands the interpolation coverage in sparse sensor regions, preventing interpolation gaps.

[0011] The high-frequency impact components are mapped using a dynamic pressure distribution map to obtain the impact energy coefficient for each pressure sensor. The pressure gradient vector at each sensor location is extracted from the dynamic pressure distribution map; its magnitude indicates the intensity of pressure change at that location, and its direction indicates the direction of the fastest pressure increase. The instantaneous energy envelope of the high-frequency impact component for each pressure sensor is calculated in the time domain, and the ratio of this instantaneous energy envelope to the pressure gradient vector is calculated to obtain the initial energy ratio. The initial energy ratios of all pressure sensors in the distributed pressure sensor array are obtained, and the maximum energy ratio is identified. The initial energy ratio of each pressure sensor is divided by this maximum energy ratio to obtain the impact energy coefficient for each pressure sensor. The impact energy coefficient quantitatively characterizes the intensity level of the impact component at each location relative to the local aerodynamic gradient, providing a normalization basis for subsequent weighted corrections and ensuring the comparability and uniformity of impact effects at different locations and times.

[0012] Based on the impact energy coefficient and dynamic pressure distribution map, the effective pressure value in the original pressure signal is nonlinearly weighted and corrected to output a corrected impact-resistant pressure signal. For each pressure sensor, the exponential parameter of the weighted correction function is determined according to its corresponding impact energy coefficient, which is positively correlated with the impact energy coefficient. The average pressure of the neighborhood of the pressure sensor is read from the dynamic pressure distribution map and used as a reference value. The high-frequency impact component is subtracted from the value of each sampling point in the original pressure signal to obtain a preliminary purified signal. Then, with the reference value as the target value, a nonlinear pull is applied to the preliminary purified signal using an exponentially decaying weighted correction function, causing the corrected sampling point values ​​to converge towards the reference value. The convergence speed is controlled by the exponential parameter. When the impact energy coefficient is small, the exponential parameter is small, and the corrected signal is closer to the preliminary purified signal, preserving the true details of aerodynamic changes. When the impact energy coefficient is large, the exponential parameter increases, and the corrected signal converges to the reference value faster, effectively suppressing the influence of residual impact. The signals of all pressure sensors after nonlinear weighting correction are reassembled in time order to output the corrected impact-resistant pressure signal. This correction method not only eliminates the direct interference of the impact component, but also compensates for the loss of local pressure information caused by the impact by introducing a spatial neighborhood reference value, thus ensuring the continuity and accuracy of the final pressure signal.

[0013] As a technical solution of the present invention, when reading the average neighborhood pressure of the pressure sensor location from the dynamic pressure distribution map as a reference value, a spatial neighborhood window is defined in the dynamic pressure distribution map with the three-dimensional spatial coordinates of the pressure sensor as the center. The size of the spatial neighborhood window is determined by the local radius of curvature of the aircraft surface; the smaller the local radius of curvature, the smaller the size of the spatial neighborhood window. The pressure values ​​of all interpolation points within the spatial neighborhood window are statistically analyzed. After excluding pressure values ​​above the upper quartile and below the lower quartile, the arithmetic mean of the remaining pressure values ​​is calculated, and this arithmetic mean is used as the reference value of the pressure sensor at the current moment. Preferably, the geometry of the spatial neighborhood window is an asymmetric ellipse, with the major axis of the ellipse aligned with the main airflow direction at that location on the aircraft surface, and the minor axis perpendicular to the main airflow direction. This window design, adapted to airflow characteristics, considers the attenuation difference of aerodynamic forces along the streamline direction and spanwise direction, making the reference value more consistent with aerodynamic laws and avoiding unreasonable averaging effects introduced by isotropic windows in high gradient regions.

[0014] As a preferred embodiment of the present invention, before performing the transient impact feature separation operation, the data preprocessing unit receives the original pressure signal frame in each sampling period, performs channel consistency verification on the sampled values ​​of each pressure sensor in the original pressure signal frame, and determines whether there are abnormal channels whose sampled values ​​exceed the range of the pressure sensor. If an abnormal channel exists, spatial interpolation is performed using the sampled values ​​of pressure sensors at adjacent positions of the abnormal channel in the same sampling period to generate a replacement value to replace the sampled value of the abnormal channel. All sampled values ​​that have undergone channel consistency verification and abnormal replacement are reassembled into a preprocessed original pressure signal frame, and then the preprocessed original pressure signal frame is sent to the transient impact feature separation operation in chronological order. This preprocessing step completes the identification and repair of abnormal channels before signal separation, preventing sensor saturation or failure caused by transient impact from having a chain reaction on the subsequent decomposition and reconstruction process, and improving the robustness of the entire signal processing chain.

[0015] In a preferred embodiment of the present invention, after outputting the corrected shock-resistant pressure signal, the corrected shock-resistant pressure signals output by all pressure sensors are arranged in chronological order to form a multi-channel pressure signal matrix. Principal component analysis is performed on this multi-channel pressure signal matrix to extract the principal component with the highest contribution rate as the dominant pressure mode of the aircraft's current flight state. The dominant pressure mode is compared with the expected pressure mode in the aircraft's flight control system. If the difference between the two exceeds a preset flight safety threshold, a pressure sensor array recalibration command is generated. By using principal component analysis to monitor the corrected pressure signal globally, system-level measurement deviations caused by long-term sensor drift or drastic changes in the aerodynamic environment can be detected in a timely manner, providing a trigger basis for recalibration.

[0016] In a preferred embodiment of the present invention, after generating a recalibration command for the pressure sensor array, the command is sent to the built-in calibration circuit of each piezoresistive pressure sensor. The built-in calibration circuit disconnects the pressure sensing diaphragm from the external pressure medium and connects instead to an internal reference voltage source. Under the control of the built-in calibration circuit, each pressure sensor sequentially measures according to multiple different reference voltage levels, obtaining the calibration output value corresponding to each reference voltage level. Based on the calibration output values ​​reported by all pressure sensors, the mean zero-point offset and mean sensitivity drift of the entire distributed pressure sensor array are calculated. These mean zero-point offset and mean sensitivity drift are used to batch correct the subsequently acquired raw pressure signals until the next recalibration command is received. This scheme automatically completes array-level parameter correction during flight intervals or when recalibration conditions are triggered, ensuring the consistency and traceability of pressure measurement benchmarks during long-term service.

[0017] The technical effects and advantages provided by the present invention in the above technical solution are as follows: By reconstructing the spatial gradients of low-frequency aerodynamic components at different locations based on flight parameters to generate a dynamic pressure distribution map, and incorporating this map into the energy mapping process of the impact component, this approach departs from the conventional method of extracting impact features solely through time-frequency domain analysis. Driven by parameters such as flight Mach number, angle of attack, and sideslip angle, and combined with sensor spatial coordinates and radial basis function interpolation to apply anisotropic constraints, the dynamic pressure distribution map reflects real-time changes in airflow stagnation points and separation zone boundaries. The pressure gradient vector at each sensor location is calculated on this physically constrained pressure map. The magnitude of this pressure gradient vector is then compared to the instantaneous energy envelope of the high-frequency impact component. After normalization using the maximum energy ratio across the entire array, the impact energy coefficient is obtained. This coefficient reflects the normalization level of the impact energy relative to the drastic changes in local aerodynamic pressure, extending impact intensity assessment from isolated sensor signal characteristics to the overall framework of the spatial pressure field. This avoids the problem of incorrect impact energy estimation due to flight attitude changes, improving the accuracy of impact energy quantification and the comparability across different flight states.

[0018] When performing nonlinear weighted correction on the original pressure signal based on the impact energy coefficient and dynamic pressure distribution map, the neighborhood average pressure provided by the dynamic pressure distribution map serves as the reference value. The correction function adopts an exponential decay form positively correlated with the impact energy coefficient. The initial purified signal is obtained by subtracting the high-frequency impact component from the original sampled value. During the correction process, this signal is pulled towards the reference value by the exponential weighting function, and the convergence speed is controlled by the exponential parameter. The calculation of the neighborhood average pressure excludes abnormal pressure values, and the neighborhood window adopts an asymmetric ellipse, with its major axis aligned with the main airflow direction and its minor axis perpendicular to the main airflow direction. The window size is adjusted according to the local radius of curvature. This structure allows the correction reference to reflect both the reasonable expectation of the surrounding flow field pressure and fully consider the influence of local geometry and airflow direction on the pressure distribution. In areas with strong impact interference, the impact energy coefficient is higher, and the correction strength is enhanced, causing the signal to converge quickly to the average pressure level of the spatial neighborhood; in areas with weak impact, the correction strength automatically weakens, preserving the true pulsation details in the original signal. This achieves a nonlinear match between the correction intensity and the impact energy distribution and spatial pressure field characteristics. The corrected signal maintains the spatial continuity and dynamic response characteristics of the aerodynamic pressure component while eliminating impact noise. Attached Figure Description

[0019] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this invention. For those skilled in the art, other drawings can be obtained based on these drawings.

[0020] Figure 1This is a flowchart of a high-precision shock-resistant pressure sensing signal processing method applicable to aviation scenarios; Figure 2 This is a flowchart of distributed pressure sensing signal preprocessing and transient impact feature separation; Figure 3 This is a flowchart of dynamic pressure field reconstruction and impact energy coefficient calculation based on anisotropic constraints; Figure 4 This is a flowchart of the pressure sensor array recalibration process based on dominant pressure mode comparison. Detailed Implementation

[0021] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0022] See Figure 1 This invention provides a high-precision shock-resistant pressure sensing signal processing method suitable for aviation scenarios, comprising: deploying multiple pressure sensors on the surface of an aircraft to form a distributed pressure sensing array, and acquiring raw pressure signals of the aircraft in different flight attitudes in real time; performing transient impact feature separation operation on the raw pressure signal acquired by each pressure sensor, decomposing the raw pressure signal into high-frequency impact components and low-frequency aerodynamic components; performing spatial gradient reconstruction on the low-frequency aerodynamic components corresponding to pressure sensors at different locations according to the current flight parameters of the aircraft, generating a dynamic pressure distribution map of the aircraft surface; using the dynamic pressure distribution map to perform energy normalization mapping on the high-frequency impact components to obtain the impact energy coefficient corresponding to each pressure sensor; and performing nonlinear weighted correction on the effective pressure value in the raw pressure signal based on the impact energy coefficient and the dynamic pressure distribution map, outputting the corrected shock-resistant pressure signal.

[0023] Example 1: In practice, a predicted aerodynamic load distribution map of the aircraft surface is obtained. This map is generated from computational fluid dynamics simulations or wind tunnel test data, using a three-dimensional surface mesh of the aircraft as its geometric basis. Each mesh node stores aerodynamic load parameters such as pressure coefficient and shear stress. Based on the predicted aerodynamic load distribution map, high-load areas where the pressure gradient amplitude on the aircraft surface exceeds a preset gradient threshold are identified. These high-load areas are then overlaid with structurally sensitive parts of the aircraft for analysis. Sensor deployment points are set at the wing leading edge, fuselage nose, and air intake lip. At the wing leading edge, sensor deployment points are evenly distributed along the spanwise direction at percentage intervals of the chord length. At the fuselage nose, sensor deployment points are arranged in a circular array around the nose cone stagnation point. At the air intake lip, sensor deployment points are arranged along the circumference of the lip at equal arc length intervals.

[0024] Each sensor deployment point is equipped with a piezoresistive pressure sensor. The piezoresistive pressure sensor comprises a silicon-based piezoresistive sensing element, temperature drift compensation circuitry, and built-in calibration circuitry. The natural frequency of the sensing element in the piezoresistive pressure sensor is an order of magnitude higher than the natural frequency of the aircraft structure. The range of the piezoresistive pressure sensor is determined by multiplying the maximum pressure value predicted at the sensor deployment point in the aerodynamic load distribution prediction map by a safety factor, with the safety factor ranging from 1.5 to 2.0.

[0025] All piezoresistive pressure sensors are connected to the data preprocessing unit via a synchronous acquisition circuit. The synchronous acquisition circuit contains multiple input channels, each corresponding to one piezoresistive pressure sensor. An integrated phase-locked loop (PLL) generates a unified sampling clock signal, which is distributed to the sample-and-hold module of each input channel via equal-length clock traces. The synchronous acquisition circuit assigns the same sampling clock signal to each piezoresistive pressure sensor, ensuring complete timing alignment of the sampling clock signals for all piezoresistive pressure sensors, with clock offset controlled at the nanosecond level.

[0026] The data preprocessing unit receives the sampling clock signal and sensor data bus signal output from the synchronous acquisition circuit. The data preprocessing unit triggers all piezoresistive pressure sensors to simultaneously acquire pressure data according to the rising edge of the sampling clock signal. At each rising edge of the sampling clock signal, all input channels of the synchronous acquisition circuit synchronously enter the sample-and-hold state, temporarily storing the analog voltage signals output by each piezoresistive pressure sensor in the sample-and-hold capacitor, which is then sequentially converted into digital values ​​by the analog-to-digital converter. The pressure data acquired from all piezoresistive pressure sensors in each sampling period constitute a raw pressure signal frame. The data structure of the raw pressure signal frame includes a frame header identifier, sampling period number, timestamp, and several data words. Each data word corresponds to the raw pressure signal sample value of a piezoresistive pressure sensor in that sampling period. The order of the data words in the raw pressure signal frame is consistent with the physical location encoding order of the piezoresistive pressure sensors in the distributed pressure sensor array, so that the sample values ​​of the corresponding piezoresistive pressure sensors can be indexed by spatial location later.

[0027] Example 2: In specific implementation, please refer to Figure 2 The data preprocessing unit receives the raw pressure signal frames for each sampling period. Internally, the data preprocessing unit maintains a channel status table, which records the channel number, three-dimensional coordinates of the installation location, upper and lower range values, and normal operation flag for each piezoresistive pressure sensor in the distributed pressure sensor array. The data preprocessing unit performs frame header verification and timestamp continuity checks on the raw pressure signal frames. Frame header verification is completed by identifying the synchronization word in the frame header identifier. Timestamp continuity checks are performed by comparing the difference between the timestamp of the current raw pressure signal frame and the timestamp of the previous raw pressure signal frame to see if it equals the completion of the sampling period.

[0028] After the frame header check and timestamp continuity check pass, the data preprocessing unit performs channel consistency verification on the sampled values ​​of each piezoresistive pressure sensor in the original pressure signal frame. The specific implementation of the channel consistency verification is as follows: read the sampled value corresponding to the m-th data word in the original pressure signal frame, denoted as... ,in This refers to the channel number of the piezoresistive pressure sensor in the distributed pressure sensing array. The value ranges from 1 to M, where M is the total number of piezoresistive pressure sensors in the distributed pressure sensor array; the upper limit of the range of the m-th piezoresistive pressure sensor is indexed from the channel status table. and lower limit of range , This represents the maximum pressure input value that the m-th piezoresistive pressure sensor can respond normally, as determined during factory calibration. This represents the minimum pressure input value that the m-th piezoresistive pressure sensor can respond normally, determined during factory calibration; it also determines the sampled value. Does it meet the requirements? or If the condition is met, the channel corresponding to the m-th piezoresistive pressure sensor is determined to be an abnormal channel. All M data words in the original pressure signal frame are traversed to complete the consistency check of all piezoresistive pressure sensor channels.

[0029] If an abnormal channel exists, the data preprocessing unit performs spatial interpolation using the sampled values ​​from pressure sensors adjacent to the abnormal channel within the same sampling period. The spatial interpolation is implemented as follows: for the m-th piezoresistive pressure sensor identified as an abnormal channel, the three-dimensional coordinates of its installation position are obtained from the channel status table, denoted as... , This represents the coordinate value of the m-th piezoresistive pressure sensor along the heading axis in the aircraft coordinate system. This represents the coordinate value of the m-th piezoresistive pressure sensor along the spanwise axis in the aircraft coordinate system. This represents the coordinate value of the m-th piezoresistive pressure sensor along the normal axis in the aircraft coordinate system; Set the spatial search radius around the center. spatial search radius The value is 1.5 times the average distance between adjacent piezoresistive pressure sensors in the distributed pressure sensor array; in the spatial search radius Search within the range for all adjacent piezoresistive pressure sensors that are in normal working condition. Suppose K adjacent piezoresistive pressure sensors are found, record the three-dimensional coordinates of the installation positions of each of the K adjacent piezoresistive pressure sensors. Where k is the index number of adjacent piezoresistive pressure sensors, and the value of k ranges from 1 to K, and represents the sampled values ​​of K adjacent piezoresistive pressure sensors in the same sampling period. The substitute value is calculated using inverse distance weighted interpolation. substitution value The calculation formula is: In the formula, This represents the replacement value for the m-th abnormal channel. This represents the sampled value of the k-th adjacent piezoresistive pressure sensor in the same sampling period. This represents the Euclidean spatial distance between the piezoresistive pressure sensor corresponding to the m-th abnormal channel and the k-th adjacent piezoresistive pressure sensor. The calculation formula is K represents the spatial search radius centered on the piezoresistive pressure sensor corresponding to the m-th abnormal channel. The total number of adjacent piezoresistive pressure sensors found within the range. The data preprocessing unit uses the calculated surrogate values. Replace the sampled value of the m-th abnormal channel.

[0030] In some embodiments, if the spatial search radius If no adjacent piezoresistive pressure sensors in normal working condition are found within the search range, then the spatial search radius is increased. The value is increased by 0.5 times each time until at least one adjacent piezoresistive pressure sensor is found, and then inverse distance weighted interpolation calculation is performed.

[0031] The data preprocessing unit reassembles all sampled values ​​that have undergone channel consistency verification and anomaly replacement into a preprocessed original pressure signal frame. The data structure of the preprocessed original pressure signal frame is the same as that of the original pressure signal frame. The preprocessed original pressure signal frame contains M data words, each corresponding to a sampled value from a piezoresistive pressure sensor after channel consistency verification and anomaly replacement processing.

[0032] The data preprocessing unit sends the preprocessed raw pressure signal frames to the transient impact feature separation operation in chronological order. The transient impact feature separation operation is performed independently for each piezoresistive pressure sensor in the distributed pressure sensor array. Taking the m-th piezoresistive pressure sensor as an example, the specific implementation of the transient impact feature separation operation is as follows: The raw pressure signal sequence of the m-th piezoresistive pressure sensor within a continuous time window is acquired. The length of the continuous time window is N sampling periods, where N is determined based on the sampling frequency of the piezoresistive pressure sensor and the reciprocal of the aircraft structure's natural frequency. The value of N ensures that the duration of the continuous time window covers at least one complete aerodynamic pressure change cycle. From the preprocessed raw pressure signal frame sequence output by the data preprocessing unit, N frames are extracted in chronological order. The sampled value of the m-th data word is extracted from each preprocessed raw pressure signal frame to form the raw pressure signal sequence of the m-th piezoresistive pressure sensor, denoted as . Where n is the sampling point number, and the value of n ranges from 1 to N. This represents the pressure sample value of the m-th piezoresistive pressure sensor at the n-th sampling point.

[0033] For the original pressure signal sequence Local extremum detection is performed. The implementation method for local extremum detection is as follows: calculate the original pressure signal sequence. First-order forward difference sequence The value of n ranges from 1 to N-1; calculate the first-order forward difference sequence. The absolute value of the amplitude change rate is used to obtain the amplitude change rate sequence; the amplitude change rate of each sampling point in the amplitude change rate sequence is compared with a preset change threshold. Comparison, preset change threshold The value is 5% of the range of the piezoresistive pressure sensor, and the range is... ; The rate of change of amplitude exceeds the preset change threshold The sampling points are marked as candidate mutation points; non-maximum suppression is applied to the candidate mutation points. If multiple candidate mutation points appear within a range of three consecutive sampling points, only the candidate mutation point with the largest amplitude change rate is retained as the signal mutation point. The signal mutation point is the point whose amplitude change rate exceeds a preset change threshold.

[0034] Using each signal abrupt change point as a dividing boundary, the original pressure signal sequence is divided. The signal is divided into multiple impact event intervals and non-impact event intervals. The segmentation method is as follows: the signal segment between two adjacent signal abrupt change points is divided into impact event intervals, with the starting point of the impact event interval being the previous signal abrupt change point and the ending point being the next signal abrupt change point; the original pressure signal sequence is then... The signal segment from the initial sampling point to the first signal abrupt change point is divided into non-impact event intervals; the segment from the last signal abrupt change point to the original pressure signal sequence is divided into non-impact event intervals. The signal segment between the termination sampling points is divided into non-impact event intervals; the signal segment between two non-adjacent signal abrupt change points without signal abrupt change points is also divided into non-impact event intervals.

[0035] Empirical Mode Decomposition (EMD) is performed on the signal data within each impact event interval. During EMD, mirror extension processing is first applied to the signal data within the impact event interval to suppress endpoint effects. The mirror extension processing is implemented as follows: using the signal data within the impact event interval as the original data segment, several sampling points are extended outwards from both ends of the original data segment, with the number of extended sampling points being one-tenth of the original data segment length; outside the starting point of the original data segment, the signal data inside the original data segment is mirrored with the starting point as the center of symmetry, generating starting-side extended data; outside the ending point of the original data segment, the signal data inside the original data segment is mirrored with the ending point as the center of symmetry, generating ending-side extended data; the starting-side extended data, the original data segment, and the ending-side extended data are then concatenated to form the extended data segment. Empirical Mode Decomposition (EMD) is performed on the extended data segment. EMD extracts intrinsic mode functions (EMFs) through a screening process. This screening process includes: identifying all maxima and minima of the extended data segment; fitting cubic spline curves to the maxima to obtain the upper envelope; fitting cubic spline curves to the minima to obtain the lower envelope; calculating the mean curves of the upper and lower envelopes; subtracting the mean curves from the extended data segment to obtain candidate EMFs; and repeating the screening process until the candidate EMFs satisfy the EMF conditions. These conditions include a zero-mean condition and that the number of extrema and the number of zero-crossing points differ by no more than one. After EMD, the portion of the extended data segment corresponding to the original data segment is selected as the valid EMFs.

[0036] Oscillation modes with frequencies higher than the natural frequencies of the aircraft structure are extracted from the effective eigenmode functions as high-frequency impact components. The extraction method is as follows: a Hilbert transform is performed on each effective eigenmode function to calculate the instantaneous frequency; effective eigenmode functions with a mean instantaneous frequency higher than the natural frequencies of the aircraft structure are identified as high-frequency impact components. The natural frequencies of the aircraft structure are obtained through modal analysis of the aircraft structure and are the lowest values ​​among the first-order bending mode frequencies of the wing and the first-order torsional mode frequencies of the fuselage.

[0037] The signal data within the non-impact event interval is spliced ​​with the residual signal within the impact event interval after removing the high-frequency impact component, forming the low-frequency aerodynamic component. The splicing method is as follows: the original pressure signal sequence is... The signal is decomposed into multiple signal segments in chronological order. For each non-impact event interval, the signal data within that interval is directly extracted as a component of the low-frequency aerodynamic component. For each impact event interval, the residual signal after empirical mode decomposition and removal of high-frequency impact components is extracted. The residual signal is the original signal data within that impact event interval minus all high-frequency impact components. The signal segments are then analyzed according to their position within the original pressure signal sequence. The original time sequence is used to connect the signal data in the non-impact event interval and the residual signals corresponding to each impact event interval in sequence to form a complete low-frequency aerodynamic component sequence.

[0038] Example 3: In specific implementation, please refer to Figure 3 The system reads the aircraft's current flight parameters. This reading is achieved through the aircraft's avionics bus interface, which interacts with the air data computer and the inertial navigation system. The air data computer outputs the flight Mach number, angle of attack, and sideslip angle, while the inertial navigation system outputs the aircraft's three-axis attitude angles and three-axis angular rates. The air data computer's update frequency is synchronized with the sampling frequency of the data preprocessing unit. The flight Mach number is denoted as... , representing the ratio of an aircraft's speed to the speed of sound at its altitude, denoted by . The angle between the aircraft's longitudinal axis and the direction of the incoming flow is denoted as the sideslip angle. , which represents the angle between the longitudinal axis of the aircraft and the projection of the incoming flow direction onto the horizontal plane.

[0039] Obtain the three-dimensional spatial coordinates of each piezoresistive pressure sensor in the distributed pressure sensor array on the aircraft surface. The method for obtaining the three-dimensional spatial coordinates is as follows: After the distributed pressure sensor array is installed, a laser tracker is used to calibrate the spatial position of the geometric center of the pressure-sensing diaphragm of each piezoresistive pressure sensor. A transformation relationship is established between the aircraft body coordinate system and the sensor installation coordinate system. The spatial coordinates of the geometric center of the pressure-sensing diaphragm of each piezoresistive pressure sensor measured by the laser tracker are transformed into the aircraft body coordinate system to obtain the three-dimensional spatial coordinates of each piezoresistive pressure sensor. The three-dimensional spatial coordinates of the m-th piezoresistive pressure sensor are denoted as follows: ,in This represents the coordinate value of the m-th piezoresistive pressure sensor along the heading axis in the aircraft body coordinate system. This represents the coordinate value of the m-th piezoresistive pressure sensor along the spanwise axis in the aircraft body coordinate system. This represents the coordinate value of the m-th piezoresistive pressure sensor along the normal axis in the aircraft body coordinate system. The value of m ranges from 1 to M, where M is the total number of piezoresistive pressure sensors in the distributed pressure sensor array.

[0040] Using the three-dimensional spatial coordinates of each piezoresistive pressure sensor as interpolation nodes and the corresponding low-frequency aerodynamic components as node values, the initial pressure field on the aircraft surface is constructed using the radial basis function interpolation method. The implementation of the radial basis function interpolation method is as follows: a three-dimensional curved surface mesh is defined for the aircraft surface. This mesh is obtained by discretizing the geometric model of the aircraft surface and contains Q mesh nodes. The three-dimensional spatial coordinates of the q-th mesh node are denoted as follows: The value of q ranges from 1 to Q; for any mesh node on the three-dimensional curved surface mesh... Calculate the distance from the grid node to each piezoresistive pressure sensor mounting location. spatial distance The interpolated pressure value of the initial pressure field at the q-th grid node. It is obtained by weighted summation of radial basis functions, using multiple quadratic functions, and its expression is: In the formula, This represents the interpolated pressure value of the initial pressure field at the q-th grid node on the three-dimensional curved surface mesh of the aircraft, in Pascals. This represents the total number of piezoresistive pressure sensors in a distributed pressure sensing array. This represents the radial basis function weighting coefficients corresponding to the m-th piezoresistive pressure sensor. The linear equations formed by solving a system of linear equations consisting of the node values ​​and spatial coordinates of M piezoresistive pressure sensors are obtained. The linear equations are as follows: ,in This represents the low-frequency aerodynamic component value corresponding to the j-th piezoresistive pressure sensor. This represents the spatial distance between the j-th piezoresistive pressure sensor and the m-th piezoresistive pressure sensor, where the value of j ranges from 1 to M. This represents the spatial distance from the q-th grid node on the three-dimensional curved surface mesh of the aircraft to the installation location of the m-th piezoresistive pressure sensor, in meters. This represents the adaptive shape parameter corresponding to the m-th piezoresistive pressure sensor, in meters.

[0041] Adaptive shape parameters introduced in radial basis function interpolation methods Dynamically adjusted based on the spatial distance between adjacent piezoresistive pressure sensors. Adaptive shape parameters. The acquisition method is as follows: For the m-th piezoresistive pressure sensor, search in the distributed pressure sensor array for the one with the closest spatial distance to the m-th piezoresistive pressure sensor. Two adjacent piezoresistive pressure sensors. The value of is set to 4; calculate the m-th piezoresistive pressure sensor and Average spatial distance between adjacent piezoresistive pressure sensors Adaptive shape parameters The value is set to ,in This is the shape parameter scaling factor. The value is set to 0.8. The basis for setting it to 0.8 is that, under this value, the support radius of the quadratic radial basis function is inversely proportional to the local density of the sensor. In dense sensor regions, the support radius of the interpolation basis function shrinks to preserve local pressure details, while in sparse sensor regions, the support radius of the interpolation basis function expands to ensure the smoothness of the interpolation surface.

[0042] According to the flight Mach number Angle of attack and sideslip angle Determine the location of airflow stagnation points and the boundary of the separation zone on the aircraft surface. The location of airflow stagnation points is determined based on the angle of attack. and sideslip angle In the aircraft body coordinate system, the incoming flow direction vector is synthesized, and the intersection of the incoming flow direction vector and the three-dimensional curved surface mesh of the aircraft surface is the location of the airflow stagnation point; the boundary of the separation zone is determined by: based on the flight Mach number. and angle of attack Query the pre-stored separation zone boundary database, which is generated through computational fluid dynamics simulations at different flight Mach numbers. and angle of attack The results are obtained through combined calculations. The separation zone boundary database stores a list of separation zone boundary grid node indices corresponding to each combination of flight parameters.

[0043] Anisotropic constraints are applied to the initial pressure field using the location of the airflow stagnation point and the boundary of the separation zone, and the weight coefficients of the interpolation nodes are adjusted. The anisotropic constraints are applied as follows: at grid nodes within the boundary of the separation zone, the weights of piezoresistive pressure sensors that are far from the airflow stagnation point during radial basis function interpolation are reduced. This reduction is achieved by adjusting the radial basis function weight coefficients corresponding to the m-th piezoresistive pressure sensor. Multiply by a decay factor Attenuation factor The calculation formula is ,in This represents the shortest spatial distance from the installation location of the m-th piezoresistive pressure sensor to the boundary of the separation zone. This represents the attenuation scale parameter. The value of is set to one-fifth of the characteristic dimension of the region enclosed by the separation zone boundary, where the characteristic dimension is the semi-major axis length of the circumscribed ellipsoid of the separation zone boundary; at the airflow stagnation point, the interpolated pressure value of the initial pressure field at that grid node is forced to be equal to the total pressure value, which is determined by the flight Mach number. It is obtained by calculation using the isentropic relationship.

[0044] After adjusting for anisotropic constraints, the weighting coefficients of the radial basis function interpolation are resolved to obtain the adjusted weighting coefficients. Using these adjusted weighting coefficients, the interpolated pressure values ​​are recalculated for all mesh nodes of the 3D surface mesh of the aircraft, generating a dynamic pressure distribution map. The dynamic pressure distribution map uses the 3D surface mesh of the aircraft as its carrier, with each mesh node storing the pressure value after anisotropic constraint adjustment, denoted as... .

[0045] The pressure gradient vector at the location of each piezoresistive pressure sensor is extracted from the dynamic pressure distribution map. The extraction method is as follows: for the m-th piezoresistive pressure sensor, the installation location of the m-th piezoresistive pressure sensor is located in the dynamic pressure distribution map. The spatial gradient of the dynamic pressure distribution pattern is calculated at the grid node where the node is located. The spatial gradient is calculated using a central difference scheme, and the pressure gradient component along the heading axis is... The pressure gradient component along the spanwise axis is The pressure gradient component along the normal axis is The pressure gradient vector at the location of the m-th piezoresistive pressure sensor is denoted as... Pressure gradient vector Size The pressure gradient vector represents the degree of pressure change at the location of the m-th piezoresistive pressure sensor. The direction indicates the direction of the fastest pressure increase at the location of the m-th piezoresistive pressure sensor. The calculation formula is .

[0046] Calculate the instantaneous energy envelope in the time domain for the high-frequency impact component corresponding to each piezoresistive pressure sensor. The instantaneous energy envelope is calculated as follows: Perform a Hilbert transform on the high-frequency impact component corresponding to the m-th piezoresistive pressure sensor to obtain the analytic signal of the high-frequency impact component. Take the modulus of the analytic signal to obtain the instantaneous amplitude envelope. Then, square the instantaneous amplitude envelope to obtain the instantaneous energy envelope. The instantaneous energy envelope corresponding to the m-th piezoresistive pressure sensor is denoted as […]. t represents time.

[0047] The instantaneous energy envelope corresponding to the m-th piezoresistive pressure sensor With respect to the magnitude of the pressure gradient vector Perform a ratio calculation to obtain the initial energy ratio corresponding to the m-th piezoresistive pressure sensor. The initial energy ratio is denoted as... , The calculation formula is .when When it approaches zero, Set directly to .

[0048] Obtain the initial energy ratio of all piezoresistive pressure sensors in the distributed pressure sensor array, find the maximum energy ratio among all initial energy ratios, and denot the maximum energy ratio as . , The calculation formula is ,in This indicates the operation of taking the maximum value. The initial energy ratio of each piezoresistive pressure sensor is used. Divide by the maximum energy ratio The impact energy coefficient corresponding to each piezoresistive pressure sensor is obtained, and the impact energy coefficient corresponding to the m-th piezoresistive pressure sensor is denoted as . , The calculation formula is Impact energy coefficient The value range is from 0 to 1. The closer the value of is to 1, the more significant the impact on the location of the m-th piezoresistive pressure sensor is in the distributed pressure sensing array.

[0049] Example 4: In practical implementation, for each piezoresistive pressure sensor in the distributed pressure sensing array, the impact energy coefficient corresponding to the m-th piezoresistive pressure sensor is used. Determine the exponent parameter of the weighted correction function. The exponent parameter of the weighted correction function is denoted as... , Impact energy coefficient They exhibit a positive correlation, defined by the exponential parameter mapping function, the expression of which is: ,in This indicates the lower limit of the exponential parameter. The value is set to 1.0. This indicates the upper limit of the exponential parameter. The value is set to 5.0. The value is set to 5.0 because, at this value, the upper limit of the convergence rate of the weighted correction function is sufficient to allow the pressure sensor most severely affected by the shock to bring the initial clean signal to the neighborhood of the reference value within ten sampling periods. This represents the impact energy coefficient corresponding to the m-th piezoresistive pressure sensor. When the impact energy coefficient... As it approaches its maximum value of 1.0, the exponential parameter... Approaching the upper limit of the exponential parameter The correction strength of the weighted correction function reaches its maximum when the impact energy coefficient... When it approaches the minimum value of 0, the exponential parameter Approaching the lower limit of the exponential parameter The correction strength of the weighted correction function is reduced to the minimum.

[0050] The average pressure of the neighborhood of the m-th piezoresistive pressure sensor is read from the dynamic pressure distribution map. The specific implementation of reading the average neighborhood pressure is as follows: using the three-dimensional spatial coordinates of the m-th piezoresistive pressure sensor... Centered on a dynamic pressure distribution map, a spatial neighborhood window is defined. The size of the spatial neighborhood window is determined by the local radius of curvature of the aircraft surface at the installation location of the m-th piezoresistive pressure sensor. The local radius of curvature is obtained as follows: on the three-dimensional curved surface mesh of the aircraft surface, taking the mesh node where the m-th piezoresistive pressure sensor is installed as the reference point, the normal curvature is calculated along the two principal directions of the aircraft surface. The reciprocal of the normal curvature with the larger absolute value is taken as the local radius of curvature at the reference point, denoted as . Feature size of spatial neighborhood window With local radius of curvature The mapping relationship is ,in This represents the upper limit of the feature size of the spatial neighborhood window. The value is three times the average distance between adjacent piezoresistive pressure sensors in the distributed pressure sensor array. This represents the window size scaling factor. The value is set to 0.5. The basis for setting it to 0.5 is that, under this value, the spatial neighborhood window narrows in areas with large curvature to accurately capture local pressure features, and widens in areas with small curvature to smooth random disturbances. This indicates the operation of selecting the smaller of two values. Local radius of curvature. The smaller the feature size of the spatial neighborhood window The smaller.

[0051] The spatial neighborhood window has an asymmetric ellipse geometry. The parameters of the asymmetric ellipse are determined as follows: the main airflow direction at the grid node containing the m-th piezoresistive pressure sensor is extracted from the dynamic pressure distribution map. This main airflow direction is obtained by projecting the gradient direction of the dynamic pressure distribution map onto the tangent plane of the aircraft surface at that grid node. The major axis of the asymmetric ellipse is aligned with the main airflow direction at the location of the m-th piezoresistive pressure sensor, and its length is set as the characteristic size of the spatial neighborhood window. 1.2 times; the minor axis of the asymmetric ellipse is perpendicular to the direction of the main airflow, and the length of the minor axis is set as the characteristic size of the spatial neighborhood window. 0.8 times.

[0052] After defining the spatial neighborhood window, the pressure values ​​of all interpolation points within the spatial neighborhood window are statistically analyzed. The spatial neighborhood window contains a total of The pressure value at the v-th interpolation point is extracted from the dynamic pressure distribution map and denoted as vv. The value of v ranges from 1 to... ;Will The pressure values ​​at each interpolation point are arranged in ascending order, and the upper quartiles are calculated. and lower quartiles Upper quartile For the arranged number Pressure values ​​at each location, lower quartile For the arranged number The pressure value at each location; excluding all interpolation points within the spatial neighborhood window whose pressure values ​​are higher than the upper quartile. The interpolation point and pressure value are below the lower quartile. The number of interpolation points is denoted as , and the remaining interpolation points are retained. ;right The arithmetic mean of the pressure values ​​at the remaining interpolation points is calculated, and the arithmetic mean is denoted as... The calculation formula is ,in This represents the pressure value at the v-th remaining interpolation point after excluding outliers. The calculated arithmetic mean... This serves as the reference value for the m-th piezoresistive pressure sensor at the current moment.

[0053] The preliminary purified signal is obtained by subtracting the high-frequency impact component from the value of each sampling point in the original pressure signal. Taking the nth sampling point of the mth piezoresistive pressure sensor as an example, the original pressure signal sample value of the mth piezoresistive pressure sensor at the nth sampling point is extracted from the preprocessed original pressure signal frame and denoted as . ; Obtain the value of the high-frequency impact component of the m-th piezoresistive pressure sensor at the n-th sampling point, denoted as The initial purification signal is recorded as: The calculation formula is: .

[0054] Based on reference benchmark value For the target value, a weighted correction function in the form of exponential decay is used to pre-purify the signal. Nonlinear traction is applied. The expression for the weighted correction function is: In the formula, This represents the corrected sampling point value of the m-th piezoresistive pressure sensor at the n-th sampling point, in Pascals. This represents the initial purification signal of the m-th piezoresistive pressure sensor at the n-th sampling point, in Pascals. This represents the reference value of the m-th piezoresistive pressure sensor at the current moment, in Pascals. This represents the exponential parameter corresponding to the m-th piezoresistive pressure sensor. The value range is from 1.0 to 5.0. Impact energy coefficient Mapping function through exponential parameters Sure; Indicates the sampling period, in seconds. The value is the reciprocal of the sampling clock signal period in the synchronous acquisition circuit; This represents an exponential function with the natural constant e as its base.

[0055] The weighted correction function makes the sample point values ​​corrected. To reference base value Convergence, the convergence rate is determined by the exponential parameter Control. When the exponential parameter When the value is large, coefficient of the term The value of the sampled point approaches 1 within a sampling period, and the corrected sampled point value is then... Approaching the reference value The correction strength is large; when the exponential parameter When the value is small, coefficient of the term Approaching Corrected sampling point values Towards the reference value in smaller steps Closer proximity results in weaker correction strength. (Exponential parameter) The value of and the impact energy coefficient The positive correlation means that the more significant the impact, the greater the correction intensity, and the faster the initial purification signal is pulled to the reference value.

[0056] Iterate through all N sampling points of the m-th piezoresistive pressure sensor within a continuous time window, and sequentially perform preliminary purification signal calculation and weighted correction function traction for each sampling point to obtain the corrected sampling point value of the m-th piezoresistive pressure sensor at the n-th sampling point. The value of n ranges from 1 to N.

[0057] Iterate through all M piezoresistive pressure sensors in the distributed pressure sensor array, perform nonlinear weighted correction on each piezoresistive pressure sensor, and obtain the corrected sampling point value of each piezoresistive pressure sensor at each sampling point.

[0058] The signals from all piezoresistive pressure sensors, after nonlinear weighting correction, are reassembled in chronological order. The reassembly method is as follows: within each sampling period, the corrected sampling point values ​​of M piezoresistive pressure sensors are assembled into a corrected pressure signal frame according to their channel numbers. The data structure of the corrected pressure signal frame is the same as that of the original pressure signal frame. The corrected pressure signal frames from all sampling periods are then connected in chronological order to output the corrected shock-resistant pressure signal.

[0059] Example 5: In specific implementation, please refer to Figure 4 The corrected shock-resistant pressure signals output from all piezoresistive pressure sensors are arranged in chronological order to form a multi-channel pressure signal matrix. The multi-channel pressure signal matrix is ​​constructed by setting an analysis time window, the length of which is... Each sampling period, the length of the analysis time window The value of is determined based on the response time requirements of the aircraft flight control system for pressure field mode identification. The value is set so that the duration of the analysis time window covers one-twentieth of the pressure field variation range within a complete flight maneuver cycle of the aircraft; the current moment is extracted from the corrected pressure signal frame sequence and traced backward. The corrected pressure signal frame for each sampling period, each frame containing the corrected sampling point values ​​of M piezoresistive pressure sensors; The corrected pressure signal frames are arranged in columns to construct a multi-channel pressure signal matrix, denoted as [the matrix is ​​missing in the original text]. Multi-channel pressure signal matrix The dimension is Multi-channel pressure signal matrix The Middle Line 1 Column elements Indicates the first The piezoresistive pressure sensor in the first... The value of the sampled points of the corrected shock resistance pressure signal output in each sampling period. The value range is 1 to , The value range is from 1 to M.

[0060] For multi-channel pressure signal matrix Principal component analysis is performed. The implementation of principal component analysis involves calculating the multi-channel pressure signal matrix. The column mean vector, the first column mean vector Each element is a multi-channel pressure signal matrix. The Middle The mean of all elements in the column; the multi-channel pressure signal matrix. Subtracting the value of the corresponding element of the column mean vector from each column yields the mean-reduced matrix. ; Calculate the mean-reduced matrix The covariance matrix, denoted as covariance matrix The dimension is covariance matrix The Middle Line 1 The elements of the column represent the first... The first piezoresistive pressure sensor and the first The covariance between individual piezoresistive pressure sensors; the covariance matrix Eigenvalue decomposition yields M eigenvalues ​​and M eigenvectors. The M eigenvalues ​​are then arranged in descending order, denoted as . The corresponding M feature vectors are denoted as ; Calculate the contribution rate of each eigenvalue, the th The contribution rate of each eigenvalue ,in The value range is from 1 to M. For summation indexing; extract the eigenvector corresponding to the eigenvalue with the highest contribution rate as the first principal component direction, and then perform mean-reduced matrix calculation. Projecting onto the direction of the first principal component, we obtain the first principal component component sequence, the length of which is... The mean of the first principal component sequence within the analysis time window is taken as the dominant pressure mode of the aircraft's current flight state, denoted as . Dominant pressure mode Let be an M-dimensional vector representing the typical distribution of pressure values ​​from M piezoresistive pressure sensors in a distributed pressure sensing array under the dominant change mode.

[0061] The dominant pressure mode The expected pressure modes are compared with those in the aircraft's flight control system. The expected pressure modes are obtained as follows: The aircraft's flight control system stores an expected pressure mode database. This database is pre-calculated based on combinations of different flight Mach numbers, angles of attack, and sideslip angles within the aircraft's flight envelope. The generation process includes obtaining a baseline value for the aircraft surface pressure distribution through computational fluid dynamics simulation for each set of flight parameters, extracting the first principal component from the baseline value using principal component analysis, and using this first principal component as the expected pressure mode corresponding to that set of flight parameters. The current flight parameters are then read, and the expected pressure mode with the highest matching degree to the current flight parameters is retrieved from the database. This expected pressure mode is denoted as […]. .

[0062] The comparison method is as follows: calculate the dominant pressure mode. With expected stress mode The difference measure is calculated using the complement form of the modal confidence criterion, and the expression is: In the formula, Indicates the dominant pressure mode With expected stress mode The difference between them The value range is from 0 to 1; The dominant pressure mode representing the current flight state of the aircraft. For one 3D column vector, Each element corresponds to a pressure value weight of a piezoresistive pressure sensor in the dominant change mode; This indicates the expected pressure modes retrieved from the expected pressure mode database that match the current flight parameters. For one 3D column vector; superscript This represents the transpose operation of a column vector. This indicates the modulo operation.

[0063] Difference value It is compared with the preset flight safety threshold, which is denoted as: , The value is set to 0.15. The value of 0.15 is based on the difference in range when more than 20% of the piezoresistive pressure sensors in a distributed pressure sensor array exhibit a small range drift. Capable of triggering a recalibration process, within the range of pressure field changes caused by normal aerodynamic disturbances. Keep below the flight safety threshold. If the difference is... Exceeding the preset flight safety threshold If this is the case, a pressure sensor array recalibration command is generated. The pressure sensor array recalibration command is a digital command packet containing a recalibration trigger identifier, a command generation timestamp, and recalibration parameter type fields.

[0064] A pressure sensor array recalibration command is sent to the built-in calibration circuit of each piezoresistive pressure sensor. The transmission of the pressure sensor array recalibration command is achieved through a bidirectional data bus between the data preprocessing unit and each piezoresistive pressure sensor. The data preprocessing unit broadcasts the pressure sensor array recalibration command to all piezoresistive pressure sensors in the distributed pressure sensor array.

[0065] Upon receiving a recalibration command from the pressure sensor array, the built-in calibration circuit of each piezoresistive pressure sensor disconnects the pressure sensing diaphragm from the external pressure medium. This disconnection is achieved as follows: the built-in calibration circuit controls a microelectromechanical system (MEMS) three-way valve. This valve includes a common port, a measurement port, and a calibration port. The common port connects to the back chamber of the pressure sensing diaphragm, the measurement port connects to the external pressure medium's pressure tap, and the calibration port connects to the gas path of the internal reference voltage source. Under normal operating conditions, the MEMS three-way valve connects the common port to the measurement port. Upon receiving the recalibration command, the MEMS three-way valve switches the common port to connect to the calibration port, isolating the pressure sensing diaphragm from the external pressure medium and connecting it to the internal reference voltage source.

[0066] An internal reference voltage source provides a stable reference pressure. The output pressure value of the internal reference voltage source is controlled by the microcontroller inside the piezoresistive pressure sensor via a digital-to-analog converter, enabling it to output multiple different reference pressure levels according to instructions. Each piezoresistive pressure sensor, under the control of its built-in calibration circuit, sequentially measures according to multiple different reference voltage levels. The number of reference voltage levels is set to Q, where Q is 5. These five reference voltage levels represent five pressure values ​​evenly distributed within the measurement range of the piezoresistive pressure sensor, and each pressure value is a lower limit of the measurement range. The 25th percentile, 50th percentile, 75th percentile, and upper limit of the measuring range. In the first At the reference voltage level, the first The calibration output value obtained by measuring the pressure sensor is denoted as . , The value range is from 1 to Q.

[0067] Based on the calibration output values ​​reported by all piezoresistive pressure sensors, the mean zero-point offset and mean sensitivity drift of the entire distributed pressure sensor array are calculated. The mean zero-point offset is calculated as follows: for each piezoresistive pressure sensor, the reference voltage level is extracted as the lower limit of the range. Calibration output value at time ,calculate Compared with the factory-calibrated zero-point output value The difference between them is used as the first Zero offset value of a piezoresistive pressure sensor The arithmetic mean of the zero-point offset values ​​of all M piezoresistive pressure sensors is calculated to obtain the mean zero-point offset of the distributed pressure sensor array. The mean sensitivity drift is calculated as follows: For each piezoresistive pressure sensor, least-squares linear fitting is performed using the calibration output values ​​at Q reference voltage levels. The slope of the fitted line is used as the mean value. Current sensitivity of a piezoresistive pressure sensor , obtain the The factory-calibrated sensitivity of the piezoresistive pressure sensor Calculate the sensitivity drift rate The arithmetic mean of the sensitivity drift rates of all M piezoresistive pressure sensors is used to obtain the mean sensitivity drift of the distributed pressure sensing array. .

[0068] Using the mean of zero offset and sensitivity drift mean The raw pressure signals acquired subsequently are batch corrected. The batch correction method is as follows: for any subsequent sampling period, the first... The raw pressure signal sampled by a piezoresistive pressure sensor The corrected formula is as follows: ,in This indicates the pressure value after batch correction for zero offset and sensitivity drift.

[0069] The batch correction operation continues until the aircraft flight control system generates the next pressure sensor array recalibration command. When the next pressure sensor array recalibration command is generated and sent to the distributed pressure sensor array, the complete recalibration process of disconnecting the external pressure medium connection, connecting the internal reference voltage source, measuring multiple reference voltage levels, calculating the mean zero offset and the mean sensitivity drift, and updating the batch correction parameters is repeated.

[0070] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application.

Claims

1. A high-precision shock-resistant pressure sensing signal processing method suitable for aviation scenarios, characterized in that, include: Multiple pressure sensors are deployed on the surface of an aircraft to form a distributed pressure sensing array, which collects the raw pressure signals of the aircraft in different flight attitudes in real time. For the raw pressure signal acquired by each pressure sensor, a transient impact feature separation operation is performed to decompose the raw pressure signal into a high-frequency impact component and a low-frequency aerodynamic component. Based on the aircraft's current flight parameters, spatial gradient reconstruction is performed on the low-frequency aerodynamic components corresponding to pressure sensors at different locations to generate a dynamic pressure distribution map of the aircraft surface. The high-frequency impact component was normalized and mapped using a dynamic pressure distribution spectrum to obtain the impact energy coefficient corresponding to each pressure sensor. Based on the impact energy coefficient and dynamic pressure distribution spectrum, the effective pressure value in the original pressure signal is nonlinearly weighted and corrected to output the corrected impact-resistant pressure signal.

2. The high-precision shock-resistant pressure sensing signal processing method for aviation scenarios according to claim 1, characterized in that, The specific implementation method for deploying multiple pressure sensors on the surface of an aircraft to form a distributed pressure sensing array and collect the raw pressure signals of the aircraft under different flight attitudes in real time is as follows: Based on the predicted aerodynamic load distribution map of the aircraft surface, sensor deployment points are set at the leading edge of the wing, the nose of the fuselage, and the inlet lip of the aircraft, and a piezoresistive pressure sensor is installed at each deployment point. All piezoresistive pressure sensors are connected to the data preprocessing unit via a synchronous acquisition circuit, which assigns the same sampling clock signal to each pressure sensor. The data preprocessing unit triggers all pressure sensors to simultaneously acquire pressure data according to the rising edge of the sampling clock signal. All pressure data acquired in each sampling period constitutes a raw pressure signal frame, which contains the raw pressure signal sample value of each pressure sensor at that moment.

3. The high-precision shock-resistant pressure sensing signal processing method for aviation scenarios according to claim 1, characterized in that, The specific implementation of performing transient impact feature separation operation on the raw pressure signal acquired by each pressure sensor, decomposing the raw pressure signal into high-frequency impact components and low-frequency aerodynamic components, is as follows: The original pressure signal sequence of each pressure sensor within a continuous time window is acquired, and local extreme points are detected in the original pressure signal sequence to identify signal abrupt points where the amplitude change rate exceeds a preset change threshold. Using each signal abrupt change point as a dividing boundary, the original pressure signal sequence is divided into multiple impact event intervals and non-impact event intervals; Empirical mode decomposition is performed on the signal data within each impact event interval to extract oscillation modes with frequencies higher than the natural frequency of the aircraft structure as high-frequency impact components. The signal data within the non-impact event interval is spliced ​​with the residual signal within the impact event interval after removing the high-frequency impact component to form the low-frequency aerodynamic component.

4. The high-precision shock-resistant pressure sensing signal processing method for aviation scenarios according to claim 3, characterized in that, In the empirical mode decomposition process, the signal data within each impact event interval is first subjected to mirror continuation processing to suppress the endpoint effect, and then the intrinsic mode function is extracted as the high-frequency impact component.

5. The high-precision shock-resistant pressure sensing signal processing method for aviation scenarios according to claim 3, characterized in that, The specific implementation method for generating a dynamic pressure distribution map of the aircraft surface by reconstructing the spatial gradient of the low-frequency aerodynamic components corresponding to pressure sensors at different locations based on the aircraft's current flight parameters is as follows: Read the aircraft's current flight parameters, including flight Mach number, angle of attack, and sideslip angle, and obtain the three-dimensional spatial coordinates of each pressure sensor in the distributed pressure sensor array on the aircraft surface; Using the three-dimensional spatial coordinates of each pressure sensor as interpolation nodes and the low-frequency aerodynamic components corresponding to the pressure sensor as node values, the radial basis function interpolation method is used to construct the initial pressure field on the aircraft surface. The location of the airflow stagnation point and the boundary of the separation zone on the aircraft surface are determined based on the flight Mach number, angle of attack, and sideslip angle. Anisotropic constraints are then applied to the initial pressure field using these airflow stagnation point locations and separation zone boundaries. The weight coefficients of the interpolation nodes are adjusted to generate a dynamic pressure distribution map.

6. The high-precision shock-resistant pressure sensing signal processing method for aviation scenarios according to claim 5, characterized in that, The radial basis function interpolation method introduces an adaptive shape parameter, which is dynamically adjusted according to the spatial distance between adjacent pressure sensors, so that the support radius of the interpolation basis function is inversely proportional to the local density of the sensor.

7. The high-precision shock-resistant pressure sensing signal processing method for aviation scenarios according to claim 5, characterized in that, The specific implementation method for obtaining the impact energy coefficient corresponding to each pressure sensor by normalizing the energy mapping of high-frequency impact components using dynamic pressure distribution spectrum is as follows: The pressure gradient vector at the location of each pressure sensor is extracted from the dynamic pressure distribution map. The magnitude of the pressure gradient vector indicates the degree of pressure change at that location, and the direction indicates the direction of the fastest pressure increase. Calculate the instantaneous energy envelope of the high-frequency impact component corresponding to each pressure sensor in the time domain, and calculate the ratio of the instantaneous energy envelope to the pressure gradient vector to obtain the initial energy ratio. Obtain the initial energy ratio of all pressure sensors in the distributed pressure sensor array, find the maximum energy ratio, and divide the initial energy ratio of each pressure sensor by the maximum energy ratio to obtain the impact energy coefficient corresponding to each pressure sensor.

8. The high-precision shock-resistant pressure sensing signal processing method for aviation scenarios according to claim 7, characterized in that, The specific implementation method for nonlinearly weighting and correcting the effective pressure value in the original pressure signal based on the impact energy coefficient and dynamic pressure distribution spectrum, and outputting the corrected impact-resistant pressure signal, is as follows: For each pressure sensor, the exponential parameter of the weighted correction function is determined based on its corresponding impact energy coefficient. This exponential parameter is positively correlated with the impact energy coefficient. Read the average pressure of the neighborhood where the pressure sensor is located from the dynamic pressure distribution map, and use the average pressure of the neighborhood as the reference value; The initial purified signal is obtained by subtracting the high-frequency impact component from the value of each sampling point in the original pressure signal. Then, with the reference value as the target value, a weighted correction function in the form of exponential decay is used to nonlinearly pull the initial purified signal so that the corrected sampling point values ​​converge to the reference value. The convergence speed is controlled by the exponential parameter. The signals from all pressure sensors, after nonlinear weighting correction, are reassembled in chronological order to output the corrected shock-resistant pressure signal.

9. The high-precision shock-resistant pressure sensing signal processing method for aviation scenarios according to claim 8, characterized in that, The specific implementation method for reading the average neighborhood pressure of the pressure sensor location from the dynamic pressure distribution map and using the average neighborhood pressure as a reference value is as follows: Centered on the three-dimensional spatial coordinates of the pressure sensor, a spatial neighborhood window is defined in the dynamic pressure distribution map. The size of the spatial neighborhood window is determined by the local radius of curvature of the aircraft surface. The smaller the local radius of curvature, the smaller the size of the spatial neighborhood window. The pressure values ​​of all interpolation points within the spatial neighborhood window are statistically analyzed. After excluding pressure values ​​that are higher than the upper quartile and lower than the lower quartile, the arithmetic mean of the remaining pressure values ​​is calculated. The arithmetic mean of this value is used as the reference value for the pressure sensor at the current moment.

10. The high-precision shock-resistant pressure sensing signal processing method for aviation scenarios according to claim 9, characterized in that, The geometry of the spatial neighborhood window is an asymmetric ellipse, with the major axis of the ellipse aligned with the main airflow direction at that location on the aircraft surface, and the minor axis perpendicular to the main airflow direction.

Images