Low-altitude unmanned aerial vehicle brake multi-parameter performance dynamic test method

Abstract

The application discloses a low-altitude unmanned aerial vehicle brake multi-parameter performance dynamic testing method and belongs to the field of low-altitude unmanned aerial vehicle brake testing technology. The method takes the time when a brake control instruction arrives as a time reference, synchronously collects time series data of a brake piston displacement, brake hydraulic pressure, friction interface temperature, shell vibration acceleration and motor driving current, divides a brake process into a pre-pressing stage, a pressure building stage, a stable friction stage and a releasing stage according to the rising edge stationary point and the falling edge inflection point of the piston displacement time series, extracts a phase difference sequence between the brake hydraulic pressure and the piston displacement in the pressure building stage, extracts a local correlation coefficient sequence between the shell vibration acceleration envelope and the friction interface temperature slope in the stable friction stage, and generates a brake dynamic performance testing result according to the maximum phase difference in the phase difference sequence and the average correlation coefficient in the local correlation coefficient sequence. The application realizes quantitative evaluation of the dynamic performance of the unmanned aerial vehicle brake multi-parameters.

Description

Technical Field

[0001] This invention relates to the field of low-altitude unmanned aerial vehicle (UAV) brake testing technology, specifically a dynamic testing method for multi-parameter performance of low-altitude UAV brakes. Background Technology

[0002] The braking process of low-altitude UAV brakes involves rapid changes in multiple physical quantities, including hydraulic pressure, mechanical motion, frictional heat, structural vibration, and motor drive. Its performance depends not only on the amplitude or steady-state value of a single parameter but also on the temporal relationships and dynamic coupling between different parameters. Due to the short duration and drastic parameter changes during braking, the time synchronization and accuracy of the testing system's stage division are crucial. Existing performance tests for low-altitude UAV brakes often employ single-parameter measurements or simple multi-channel recording methods. The trigger times for each channel are inconsistent, lacking a unified time reference, making it difficult to compare changes in braking hydraulic pressure, piston displacement, temperature, vibration, and current on the same time axis. Some testing methods rely on fixed time thresholds or manual observation for stage division of the braking process, which cannot adapt to changes in piston displacement waveforms under different braking conditions, leading to inaccurate stage boundaries.

[0003] In terms of performance evaluation, existing methods typically analyze pressure response time, displacement change, peak temperature, or root mean square vibration value separately. They fail to extract the phase difference between braking hydraulic pressure and piston displacement during the pressure build-up phase, nor do they analyze the local correlation between the change in shell vibration energy and the rate of change in friction interface temperature during the stable friction phase. Due to the lack of these parameter interaction characteristics, the test results struggle to reflect the degree of hysteresis between hydraulic pressure transmission and piston movement during rapid pressure build-up, and also fail to reflect the coupling changes between thermal effects and vibration during the stable friction phase. For low-altitude UAV brakes, this hysteresis and thermal-vibration coupling directly affect the consistency of braking response and the stability of the friction process; existing testing methods are insufficient to fully characterize their dynamic performance. Summary of the Invention

[0004] To address the problems of inconsistent time bases for various parameters, reliance on fixed thresholds for braking process stage division, and lack of parameter interaction characteristics in existing testing methods for low-altitude unmanned aerial vehicle (UAV) brakes, this invention proposes a multi-parameter dynamic performance testing method for low-altitude UAV brakes. This method synchronously acquires piston displacement, braking hydraulic pressure, friction interface temperature, housing vibration acceleration, and motor drive current using the arrival time of the braking control command as a unified time base. Based on the rising edge stagnation point and falling edge inflection point of the piston displacement time sequence, it automatically divides the process into pre-compression, pressure-building, stable friction, and release stages. It extracts the phase difference sequence between the braking hydraulic pressure and piston displacement during the pressure-building stage, and the local correlation coefficient sequence between the housing vibration acceleration envelope and the friction interface temperature slope during the stable friction stage. The maximum phase difference and average correlation coefficient are used to generate the brake's dynamic performance test results, characterizing braking response delay and braking friction stability.

[0005] To achieve the above objectives, the present invention provides the following technical solution: The present invention provides a multi-parameter dynamic testing method for the brake of a low-altitude unmanned aerial vehicle (UAV), including multi-parameter synchronous acquisition, dynamic segmentation of the braking process, parameter interaction feature extraction, and dynamic performance evaluation. The multi-parameter synchronous acquisition step uses the arrival time of the braking control command of the low-altitude UAV brake as the time reference, and synchronously acquires time-series data of the brake piston displacement, braking hydraulic pressure, friction interface temperature, shell vibration acceleration, and motor drive current. The dynamic segmentation step divides the braking process into a pre-pressure stage, a pressure-building stage, a stable friction stage, and a release stage based on the rising edge stagnation point and falling edge inflection point of the piston displacement time sequence. The parameter interaction feature extraction step extracts the phase difference sequence between the braking hydraulic pressure and piston displacement during the pressure-building stage, and the local correlation coefficient sequence between the shell vibration acceleration envelope and the slope of the friction interface temperature during the stable friction stage. The dynamic performance evaluation step generates the brake dynamic performance test results based on the maximum phase difference in the phase difference sequence and the average correlation coefficient in the local correlation coefficient sequence. This method incorporates the dynamic changes of multiple parameters of the brake into a unified time reference, which can simultaneously reflect the braking response delay and friction stability, and is suitable for the characteristics of short braking time and rapid parameter changes of low-altitude UAV brakes.

[0006] As a technical solution of the present invention, the multi-parameter synchronous acquisition step includes: installing a pressure sensor on the brake cylinder, a displacement sensor on the piston rod, embedding a thermocouple on the back of the friction pad, installing an acceleration sensor on the brake housing, and installing a current clamp on the drive motor cable; when the field-programmable logic device receives the braking control command, it outputs a sampling pulse synchronized with the rising edge of the command, driving the multi-channel synchronous acquisition card to simultaneously acquire five types of signals output by the pressure sensor, displacement sensor, thermocouple, acceleration sensor, and current clamp; clock correction and resampling are performed on the five sets of time series corresponding to the five types of signals to make the five sets of time series have the same time interval. Preferably, the clock correction and resampling includes: comparing the sampling timestamp of each channel with the reference time count output by the field-programmable logic device to calculate the clock offset of each channel; for channels whose clock offset exceeds a preset offset threshold, linear interpolation is used to correct the sampled values ​​to a unified time grid; when the correction is completed, each channel sequence is resampled to the same sampling frequency, and the first sampling point of each channel sequence is aligned with the arrival time of the braking control command. This process reduces the time inconsistency of multi-channel sampling and improves the accuracy of subsequent phase difference and correlation coefficient calculations.

[0007] In the dynamic segmentation steps of the braking process, the piston displacement time series is low-pass filtered to obtain a smooth displacement series; the first derivative of the smooth displacement series is calculated to obtain the piston motion velocity series; the moment when the velocity in the piston motion velocity series changes from zero to a positive value is taken as the starting moment of the braking process; the moment when the velocity first reaches its peak and then begins to decrease is taken as the ending moment of the pre-compression stage; the moment when the velocity decreases from a positive value to zero and the displacement remains stable is taken as the starting point of the stable friction stage; and the moment when the displacement begins to decrease and the velocity changes from zero to a negative value is taken as the starting point of the release stage. Preferably, the determination of the ending moment of the pre-compression stage includes: finding the maximum velocity value in the piston motion velocity series, and taking a local minimum point before and after the maximum value; taking the moment corresponding to the maximum velocity value as the peak moment, and taking the moments corresponding to the next local minimum points on both sides of the peak moment as the front boundary and the rear boundary, respectively; finding the time point with the largest velocity decrease slope between the peak moment and the front boundary, and taking this time point as the ending moment of the pre-compression stage. Identifying the stage boundaries based on the actual piston motion velocity change can improve the segmentation accuracy and robustness.

[0008] In the parameter interaction feature extraction step, during the pressure build-up stage, bandpass filtering is applied to the brake hydraulic time series and piston displacement time series respectively. Using the piston displacement time series as a reference series, the phase difference between the brake hydraulic pressure and piston displacement in each window is obtained through sliding window cross-correlation peak detection, forming a phase difference sequence. During the stable friction stage, envelope detection is performed on the shell vibration acceleration time series to obtain the vibration envelope curve. The vibration envelope curve is divided into multiple integration windows with a fixed window length, and the vibration energy value in each integration window is calculated. At the same time, the slope value of the friction interface temperature time series in each integration window is calculated. The vibration energy value and the temperature slope value are normalized and aligned, and the local correlation coefficient between the two is calculated window by window, forming a local correlation coefficient sequence. Preferably, bandpass filtering of the brake hydraulic time series and piston displacement time series includes: determining the center frequency based on the spectral peak value of the brake hydraulic time series during the pressure build-up phase, and using 0.8 to 1.2 times the center frequency as the passband boundary; filtering the brake hydraulic time series and piston displacement time series using a finite-length unit impulse response filter designed with a Kaiser window; and performing zero-phase round-trip filtering on the two filtered series. This feature extraction method can highlight the dynamic following relationship between hydraulic pressure and displacement during the pressure build-up phase, as well as the local correlation characteristics between vibration and temperature rise during the stable friction phase.

[0009] In the dynamic performance evaluation step, the maximum phase difference and its occurrence time are extracted from the phase difference sequence as braking response delay features; the average correlation coefficient and correlation coefficient fluctuation variance are extracted from the local correlation coefficient sequence as braking friction stability features; the duration of the pre-pressure stage, pressure-building stage, and stable friction stage are calculated, and the three durations are used as stage duration features; the braking response delay features, braking friction stability features, and stage duration features are combined and encoded to obtain the dynamic performance test results of this braking process; the dynamic performance test results of multiple braking processes are connected in chronological order to obtain the multi-parameter performance dynamic test curve of the brake. Preferably, the combined encoding includes: using the braking response delay feature value as the abscissa component and the braking friction stability feature value as the ordinate component to generate dynamic performance feature points in a two-dimensional coordinate system; concatenating the three durations in the stage duration features in sequence into attribute byte segments; and appending the attribute byte segments to the end of the dynamic performance feature points to form the dynamic performance test results of this braking process. When generating multiple test curves, at the end of each braking process, the braking response delay characteristic in the current dynamic performance test result can be used as the ordinate, and the cumulative test time as the abscissa to generate a curve point corresponding to this test. The curve points corresponding to two adjacent braking processes are connected by a straight line, and a moving average filter is applied at the connection point. The complete curve formed by the connection is used as the dynamic test curve of the brake's multi-parameter performance. Through combined coding and curve connection, the test results can reflect both the comprehensive state of multiple parameters in a single braking process and the performance change trend in multiple braking processes.

[0010] The technical effects and advantages provided by the present invention in the above technical solution are as follows: During the pressure build-up phase, bandpass filtering is applied to the brake hydraulic pressure time series and piston displacement time series. Using the piston displacement time series as a reference, the phase difference between the brake hydraulic pressure and piston displacement within each window is obtained through sliding window cross-correlation peak detection, forming a phase difference sequence. The maximum phase difference in the phase difference sequence is extracted as the brake response delay feature. This processing method places the brake hydraulic pressure change and the actual piston displacement response within the same time window for cross-correlation analysis, which can reflect the hysteresis caused by the transmission of hydraulic pressure changes to piston movement during the pressure build-up process and its fluctuation over time. The maximum phase difference corresponds to the most unfavorable response delay state during the pressure build-up phase. Compared to only recording the pressure rise time or displacement arrival time, this phase difference feature directly characterizes the dynamic following relationship between hydraulic pressure input and mechanical displacement output, and can reflect the response consistency of the brake during the rapid pressure build-up phase.

[0011] During the stable friction phase, envelope detection is performed on the shell vibration acceleration time series to obtain the vibration envelope curve. Integration windows are divided according to a fixed window length, and the vibration energy value within each integration window is calculated. Simultaneously, the slope value of the friction interface temperature time series within each integration window is calculated. After amplitude normalization and alignment of the vibration energy value and temperature slope value, local correlation coefficients are calculated window by window to form a local correlation coefficient sequence. The average correlation coefficient and the variance of the correlation coefficient fluctuation are extracted as characteristics of braking friction stability. This processing method correlates the friction interface temperature change rate with the shell vibration energy change through windowing, reflecting the local coupling effect of frictional heat change on the structural vibration level. The average correlation coefficient characterizes the overall strength of thermal-vibration coupling during the stable friction phase, and the variance of the correlation coefficient fluctuation characterizes the consistency of the local correlation state over time. Compared to comparing temperature peaks or vibration root mean square values ​​alone, this correlation feature can reflect the dynamic coupling relationship between thermal effects and vibration during friction, and can determine whether there are local friction state changes in the brake during the stable friction phase. Attached Figure Description

[0012] 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.

[0013] Figure 1 This is a flowchart of a dynamic testing method for multi-parameter performance of low-altitude unmanned aerial vehicle (UAV) brakes; Figure 2 This is a flowchart of the grey relational analysis of the phase difference between braking hydraulic pressure and piston displacement, vibration energy, and temperature slope. Figure 3 This is a flowchart for generating dynamic test curves of multi-parameter performance of brakes; Schematic diagram; Figure 4 It is a sequence curve of the local correlation coefficient between vibration energy and temperature slope during the stable friction stage; Figure 5 This is a graph showing the relationship between the dynamic performance characteristics of low-altitude UAV brakes and the duration of the stable friction phase. Figure 6 This is the dynamic test curve of the maximum phase difference during multiple braking processes of a low-altitude unmanned aerial vehicle (UAV) brake. Detailed Implementation

[0014] 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.

[0015] See Figure 1 This invention provides a multi-parameter dynamic testing method for low-altitude unmanned aerial vehicle (UAV) brakes, including multi-parameter synchronous acquisition, dynamic segmentation of the braking process, parameter interaction feature extraction, and dynamic performance evaluation. In the multi-parameter synchronous acquisition step, the arrival time of the braking control command of the low-altitude UAV brake is used as the time reference, and time-series data of brake piston displacement, braking hydraulic pressure, friction interface temperature, shell vibration acceleration, and motor drive current are acquired synchronously. In the dynamic segmentation step of the braking process, the braking process is divided into a pre-pressure stage, a pressure-building stage, a stable friction stage, and a release stage based on the rising edge stagnation point and falling edge inflection point of the piston displacement time sequence. In the parameter interaction feature extraction step, the phase difference sequence between the braking hydraulic pressure and piston displacement is extracted during the pressure-building stage, and the local correlation coefficient sequence between the shell vibration acceleration envelope and the slope of the friction interface temperature is extracted during the stable friction stage. In the dynamic performance evaluation step, the dynamic performance test results of the brake are generated based on the maximum phase difference in the phase difference sequence and the average correlation coefficient in the local correlation coefficient sequence.

[0016] In practical implementation, the multi-parameter synchronous acquisition step in the dynamic test method for multi-parameter performance of low-altitude UAV brakes is implemented in the following manner.

[0017] A pressure sensor is installed in the brake cylinder, a displacement sensor is installed in the piston rod, a thermocouple is embedded on the back of the friction pads, an acceleration sensor is installed in the brake housing, and a current clamp is installed on the drive motor cable. The pressure sensor detects the brake hydraulic pressure, the displacement sensor detects the brake piston displacement, the thermocouple detects the friction interface temperature, the acceleration sensor detects the housing vibration acceleration, and the current clamp detects the motor drive current.

[0018] The field-programmable logic device (FPGA) receives braking control commands, with the arrival time of the commands serving as the time reference. The FPGA triggers a sampling pulse output on the rising edge of the braking control command, which is then sent to a multi-channel synchronous acquisition card. Driven by the sampling pulse, the multi-channel synchronous acquisition card simultaneously acquires the braking hydraulic pressure signal from the pressure sensor, the piston displacement signal from the displacement sensor, the friction interface temperature signal from the thermocouple, the housing vibration acceleration signal from the accelerometer, and the motor drive current signal from the current clamp. These five signal types correspond to five time series.

[0019] Clock correction and resampling are performed on five sets of time series. In the specific implementation, each channel of the multi-channel synchronous acquisition card has an independent sampling timestamp record. The sampling timestamp of each channel is compared with the reference time count output by the field-programmable logic device (FPGA). The reference time count output by the FPGA increments from the arrival time of the braking control command. The difference between the sampling timestamp of each channel and the reference time count is used as the clock offset for each channel.

[0020] In some embodiments, the preset offset threshold is set to 5 microseconds. The basis for setting the preset offset threshold to 5 microseconds is that the sampling clock crystal oscillator deviation range of each channel of the multi-channel synchronous acquisition card does not exceed ±3 microseconds. A 2-microsecond margin is retained on top of ±3 microseconds to form the 5-microsecond preset offset threshold. For channels whose clock offset exceeds the preset offset threshold, the sampled values ​​are corrected to a unified time grid using linear interpolation. The unified time grid is generated based on the reference time count output by the field-programmable logic device, and the interval between adjacent unified time grid points is equal to the target sampling period.

[0021] Optionally, linear interpolation correction is performed using the following formula: in, This represents the corrected signal value. This represents the time corresponding to a uniform time grid point. This represents an original sampled value in the channel to be corrected that is located before the unified time grid point. This represents an original sampled value in the channel to be corrected, located after a uniform time grid point. This represents the time corresponding to an original sample value located before a uniform time grid point. This represents the time corresponding to an original sample value located after a unified time grid point.

[0022] The above linear interpolation correction is performed on each uniform time grid point of each channel whose clock offset exceeds the preset offset threshold. Channels whose clock offset does not exceed the preset offset threshold are not subject to linear interpolation correction, and the original sampled values ​​are directly retained.

[0023] After calibration, each channel sequence is resampled to the same sampling frequency. In practice, the same sampling frequency is set to 20 kHz. The basis for setting the same sampling frequency to 20 kHz is that the main frequency components of the piston displacement signal and brake hydraulic signal during the braking process of the low-altitude UAV brake do not exceed 2 kHz, and 20 kHz is selected as the same sampling frequency according to the principle of not less than ten times the main frequency components. After resampling, each channel sequence has the same time interval, and the first sampling point of each channel sequence is aligned with the arrival time of the braking control command. The five sets of time series after clock calibration and resampling are used for subsequent processing of the multi-parameter performance dynamic test of the low-altitude UAV brake.

[0024] It is understandable that the sampling pulses output by the field-programmable logic device are synchronized with the rising edge of the braking control command, enabling the sampling times of each channel of the multi-channel synchronous acquisition card to be aligned to the same time starting point. Clock offset calculation and linear interpolation correction eliminate the deviation of sampling clocks from different channels, and resampling ensures that the five time series have a unified sampling frequency and time starting point.

[0025] In practice, the piston displacement time series is low-pass filtered to obtain a smoothed displacement sequence. A second-order Butterworth low-pass filter is used, with a cutoff frequency set to 50 Hz. The reason for setting the cutoff frequency to 50 Hz is that the effective motion components related to the braking process segments in the piston displacement time series are concentrated below 50 Hz, while components above 50 Hz are mainly sensor noise and high-frequency mechanical disturbances. After filtering, a smoothed displacement sequence is obtained, where each sampling point corresponds to a time point and a displacement value.

[0026] The piston velocity sequence is obtained by taking the first derivative of the smoothed displacement sequence. In practice, the central difference method is used to take the first derivative of the smoothed displacement sequence. For each sampling point in the smoothed displacement sequence except for the first and last sampling points, the piston velocity is calculated according to the following formula: in, The sequence number in the piston motion velocity sequence is... The piston velocity corresponding to the sampling point. This indicates that the sequence number in the smoothed displacement sequence is... The displacement value corresponding to the sampling point. This indicates that the sequence number in the smoothed displacement sequence is... The displacement value corresponding to the sampling point. This represents the sampling period of the smoothed displacement sequence. This represents the sampling point number in the smoothed displacement sequence. The piston velocity is calculated using forward differencing for the first sampling point of the smoothed displacement sequence and backward differencing for the last sampling point. The piston velocity sequence is formed by the piston velocities of all sampling points.

[0027] The process involves finding the moment in the piston's velocity sequence where the velocity changes from zero to a positive value, and using this moment as the starting point of the braking process. In practice, this is done by iterating through the piston's velocity sequence until the sequence number is... The sampling point corresponds to a piston velocity less than or equal to zero, and the sequence number is... When the piston velocity corresponding to the sampling point is greater than zero, the sequence number is... The sampling point corresponds to the moment when the velocity changes from zero to a positive value, and the moment when the velocity changes from zero to a positive value is taken as the starting moment of the braking process.

[0028] In the piston motion velocity sequence, the maximum velocity value is found. Then, within the time interval preceding the moment corresponding to the maximum velocity value, the nearest local minimum is found. Similarly, within the time interval following the moment corresponding to the maximum velocity value, the nearest local minimum is found. A local minimum is defined as a point in the piston motion velocity sequence where the velocity value corresponding to a sampling point is less than the velocity values ​​corresponding to the two adjacent sampling points. The moment corresponding to the maximum velocity value is taken as the peak moment, the moment corresponding to the local minimum before the peak moment is taken as the front boundary, and the moment corresponding to the local minimum after the peak moment is taken as the back boundary.

[0029] The velocity descent slope is calculated between the peak moment and the front boundary. In practice, the first-order forward difference is calculated point-by-point for the piston motion velocity sequence between the peak moment and the front boundary. The result of the first-order forward difference is the velocity descent slope sequence. The velocity descent slope with the largest absolute value in the velocity descent slope sequence is found, and the sampling point corresponding to the velocity descent slope with the largest absolute value is taken as the time point with the largest velocity descent slope. The time point with the largest velocity descent slope is taken as the end time of the preload stage.

[0030] The process involves searching the piston velocity sequence for the moment when the velocity decreases from a positive value to zero while the displacement remains stable. In practice, this is done by iterating through the piston velocity sequence until the value at index 1 is found. The sampling point corresponds to a piston velocity greater than zero, and the sequence number is When the piston velocity at the sampling point is zero, the record number is... The sampling point corresponds to the time. At the sequence number... Ten consecutive sampling points are selected after the time corresponding to the sampling point. If the piston velocity corresponding to these 10 sampling points is zero, and the maximum change in displacement value in the smooth displacement sequence corresponding to these 10 sampling points is less than the preset displacement stability threshold, then the sequence number is... The sampling point corresponding to the time is taken as the starting point of the stable friction stage. The preset displacement stability threshold is set to 0.05 mm. The basis for setting the preset displacement stability threshold to 0.05 mm is that the displacement change of the piston due to the wear and elastic deformation of the friction material during the stable friction stage is less than 0.05 mm.

[0031] The process involves finding the moment in the smoothed displacement sequence and the piston velocity sequence when the displacement begins to decrease and the velocity changes from zero to a negative value. In practice, this is done by iterating through the smoothed displacement sequence until the value at index 1... The displacement value corresponding to the sampling point is greater than the sequence number. The displacement value corresponding to the sampling point, and the sequence number is The piston velocity corresponding to the sampling point is zero, and the serial number is When the piston velocity corresponding to the sampling point is less than zero, the sequence number is... The sampling point corresponds to the time as the starting point of the release phase.

[0032] In practice, the braking process is divided into four stages: pre-compression stage, pressure-building stage, stable friction stage, and release stage, based on the start time of the braking process, the end time of the pre-compression stage, the start time of the stable friction stage, and the start time of the release stage. Specifically, the period from the start time of the braking process to the end time of the pre-compression stage is the pre-compression stage; the period from the end time of the pre-compression stage to the start time of the stable friction stage is the pressure-building stage; the period from the start time of the stable friction stage to the start time of the release stage is the stable friction stage; and the period after the start time of the release stage is the release stage.

[0033] Understandably, low-pass filtering removes high-frequency noise from the piston displacement time series, making the rising edge stagnation point and falling edge inflection point positions in the smooth displacement series more stable. The piston motion velocity sequence obtained by central difference differentiation can reflect the change in piston motion direction. The moment when the velocity changes from zero to a positive value corresponds to the moment when the piston begins to move. The moment when the velocity decreases with the largest slope corresponds to the moment when the piston motion velocity begins to decrease rapidly at the end of the pre-compression stage. The moment when the velocity decreases from a positive value to zero and the displacement remains stable corresponds to the moment when the stable friction stage begins. The moment when the displacement begins to decrease and the velocity changes from zero to a negative value corresponds to the moment when the release stage begins.

[0034] In specific implementation, please refer to Figure 2During the pressure build-up phase, bandpass filtering is applied to both the brake hydraulic time series and the piston displacement time series. First, the peak values ​​of the brake hydraulic time series during the pressure build-up phase are obtained. A Fast Fourier Transform (FFT) is then performed on the brake hydraulic time series to obtain its amplitude spectrum. The frequency corresponding to the maximum amplitude value is then identified and used as the center frequency. The lower passband boundary is defined as 0.8 times the center frequency, and the upper passband boundary is defined as 1.2 times the center frequency. These two boundaries constitute the passband boundaries.

[0035] A finite-length unit impulse response (FIR) filter, designed using a Kaiser window, is used to filter the brake hydraulic time series and the piston displacement time series. In the Kaiser window parameters, the shape parameter is set to 5.0, and the filter order is set to 128. The shape parameter of 5.0 is chosen to achieve a balance between passband ripple and stopband attenuation, and the filter order of 128 is chosen to ensure that the filter transition band width is less than 0.05 times the center frequency. The impulse response of the ideal bandpass filter is determined based on the passband boundary and the filter order. The impulse response of the ideal bandpass filter is then multiplied point-by-point by the Kaiser window function to obtain the FIR filter coefficients. Convolution filtering of the brake hydraulic time series using the FIR filter coefficients yields the preliminary filtered sequence of the brake hydraulic time series. Similarly, convolution filtering of the piston displacement time series using the FIR filter coefficients yields the preliminary filtered sequence of the piston displacement time series.

[0036] Zero-phase round-trip filtering is applied to both the initial filtered sequences of the brake hydraulic time series and the initial filtered sequences of the piston displacement time series. The initial filtered sequence of the brake hydraulic time series is then inverted to obtain the inverted filtered sequence. This inverted filtered sequence is then passed through the same finite-length unit impulse response filter to obtain the secondary filtered sequence. This secondary filtered sequence is then inverted to obtain the zero-phase filtered brake hydraulic time series. The same inversion, secondary filtering, and inversion operations are then performed on the initial filtered sequence of the piston displacement time series to obtain the zero-phase filtered piston displacement time series.

[0037] Using the zero-phase filtered piston displacement time series as a reference series, the phase difference between the zero-phase filtered brake hydraulic pressure and the zero-phase filtered piston displacement within each window is obtained through sliding window cross-correlation peak detection, forming a phase difference sequence. The sliding window length is set to 200 sampling points, and the sliding step size is set to 10 sampling points. For each sliding window, a segment of the brake hydraulic pressure window sequence is extracted from the zero-phase filtered brake hydraulic pressure time series, and a segment of the piston displacement window sequence within the same time period is extracted from the zero-phase filtered piston displacement time series. The cross-correlation function between the brake hydraulic pressure window sequence and the piston displacement window sequence is calculated, and the cross-correlation function value varies with the number of delay points. The delay point number corresponding to the maximum cross-correlation function value is found in the cross-correlation function, and this number is taken as the peak delay point number. The peak delay point number is divided by the sampling frequency to obtain the time delay value. The time delay value is multiplied by twice pi, and then multiplied by the center frequency to obtain the phase difference within that sliding window. Perform the above operation on all sliding windows to obtain the phase difference within each sliding window. The phase differences within all sliding windows are arranged in the order of the sliding windows to form a phase difference sequence.

[0038] During the stable friction phase, envelope detection is performed on the shell vibration acceleration time series to obtain the vibration envelope curve. A Hilbert transform is then performed on the shell vibration acceleration time series to obtain the analytic signal. The amplitude of the analytic signal is then calculated to obtain the vibration envelope curve. Each sampling point in the vibration envelope curve corresponds to a time point and a vibration envelope amplitude.

[0039] The vibration envelope curve is divided into multiple integration windows with a fixed window length. The fixed window length is set to 100 sampling points, and there is no overlap between adjacent integration windows. The number of integration windows is obtained by dividing the total number of sampling points of the vibration envelope curve by the fixed window length and rounding down. The integration window numbers start from 1 and increase sequentially up to the number of integration windows.

[0040] Calculate the vibration energy value within each integration window. For the q-th integration window, the vibration energy value is calculated using the following formula: in, Indicates the first Vibration energy value within an integration window Indicates the integration window number. Indicates the length of the integration window. This indicates the sampling point number within the integration window. The vibration envelope curve represents the first... Within the first integration window The amplitude of each sampling point is calculated. The above calculation is performed for each integration window to obtain the vibration energy value within each integration window. The vibration energy values ​​within all integration windows constitute a vibration energy value sequence.

[0041] Simultaneously, the slope value of the friction interface temperature time series within each integration window is calculated. For the q-th integration window, a temperature window sequence corresponding to that integration window time period is extracted from the friction interface temperature time series. A least-squares linear fit is performed on the temperature window sequence to obtain a temperature fitting line, and the slope of this line is used as the temperature slope value within the q-th integration window. The above extraction and linear fitting operations are performed on each integration window to obtain the temperature slope value within each integration window. The temperature slope values ​​within all integration windows constitute a temperature slope value sequence.

[0042] The vibration energy value sequence and the temperature slope value sequence are aligned by amplitude normalization. Each vibration energy value in the vibration energy value sequence is divided by the maximum value in the sequence to obtain the normalized vibration energy value sequence. Similarly, each temperature slope value in the temperature slope value sequence is divided by the absolute value of the temperature slope value with the largest absolute value to obtain the normalized temperature slope value sequence. The normalized vibration energy value sequence and the normalized temperature slope value sequence have the same number of sampling points, and each point corresponds to the same integration window.

[0043] The local correlation coefficients between the normalized vibration energy value sequence and the normalized temperature slope value sequence are calculated window by window, forming a local correlation coefficient sequence. For the q-th integration window, the q-th value in the normalized vibration energy value sequence is used as the reference point, and the q-th value in the normalized temperature slope value sequence is used as the comparison point. The grey correlation coefficient between the reference point and the comparison point is calculated, with the resolution coefficient set to 0.5. The reason for setting the resolution coefficient to 0.5 is that a resolution coefficient of 0.5 in the grey correlation coefficient calculation can clearly distinguish the degree of correlation between different windows, while avoiding excessive changes in the correlation coefficient caused by small fluctuations. Grey correlation coefficient calculation is performed for each integration window to obtain the local correlation coefficient within each integration window. The local correlation coefficients within all integration windows are arranged in the order of the integration windows to form a local correlation coefficient sequence.

[0044] It is understandable that bandpass filtering of the brake hydraulic fluid time series and piston displacement time series can remove low-frequency trend components and high-frequency noise during the pressure build-up phase, while retaining the pulsation components related to the pressure build-up process. Sliding window cross-correlation peak detection using the piston displacement time series as a reference series can obtain the phase difference change of brake hydraulic fluid relative to piston displacement at different time positions. Envelope detection of the shell vibration acceleration time series followed by dividing it into integration windows to calculate vibration energy values ​​can reflect the vibration intensity at different time intervals during the stable friction phase. Fitting the slope of the friction interface temperature time series within the same integration window can obtain the temperature rise rate within each integration window. After normalizing and aligning the amplitudes of the vibration energy value series and the temperature slope value series, calculating the grey correlation coefficient window by window can obtain the degree of local correlation between vibration energy and temperature slope, and each value in the local correlation coefficient series corresponds to a fixed integration window position.

[0045] See Figure 4 In the figure, the horizontal axis represents the integration window number, and the vertical axis represents the normalized amplitude and local correlation coefficient, reflecting the dynamic changes in the vibration energy value sequence of the low-altitude UAV brake housing, the temperature slope value sequence of the friction interface, and the local correlation coefficient sequence of the two during the stable friction stage. The blue solid line represents the normalized vibration energy value sequence, the orange dashed line represents the normalized temperature slope value sequence, and the green dotted line represents the local correlation coefficient sequence.

[0046] As shown in the figure, the normalized vibration energy value sequence exhibits a trend of first increasing and then decreasing within the integration window range of approximately 0 to 170, with the peak occurring around approximately 120, indicating that the shell vibration intensity underwent a process of enhancement and attenuation during this stage. The normalized temperature slope value sequence shows a similar overall trend to the vibration energy value within the same range, starting near zero, rapidly rising to a peak, and then gradually decreasing, indicating that the rate of temperature increase at the friction interface first increases and then slows down. The two curves show high synchronicity in this range, verifying the correlation between vibration energy and temperature slope.

[0047] The variation range of the local correlation coefficient sequence is relatively stable, generally remaining between 0.6 and 0.9, indicating a strong and continuous local correlation between the shell vibration envelope and the temperature slope of the friction interface during the stable friction phase. The slight fluctuations in the local correlation coefficient reflect the time-varying characteristics of this correlation, consistent with the resolution coefficient of 0.5 used in grey relational coefficient calculations, which effectively distinguishes correlation changes between different integration windows.

[0048] After the integration window number exceeds approximately 170, both the normalized vibration energy value sequence and the normalized temperature slope value sequence show a continuous downward trend, approaching zero, indicating that the shell vibration and temperature rise rate are significantly weakened in the later stage of stable friction. The local correlation coefficient sequence oscillates between 0.6 and 0.8 in this range, maintaining a moderate correlation level, reflecting that although the correlation between vibration and temperature slope has weakened, it still exists.

[0049] In specific implementation, please refer to Figure 3 The maximum phase difference and its occurrence time are extracted from the phase difference sequence. Each phase difference value in the sequence is traversed, and the largest value is recorded as the maximum phase difference value. The index of the maximum phase difference value in the sequence is also recorded. The occurrence time of the maximum phase difference is calculated based on the sliding window parameters corresponding to the phase difference sequence. Each phase difference value in the sequence corresponds to a sliding window, and the starting time of the sliding window is obtained by adding the product of the starting time of the pressure build-up phase and the sliding window index and the sliding step size. The occurrence time of the maximum phase difference is equal to the starting time of the pressure build-up phase plus the product of the sliding window index corresponding to the maximum phase difference value and the sliding step size. The maximum phase difference value and its occurrence time are used together as a braking response delay feature.

[0050] The average correlation coefficient and the variance of correlation coefficient fluctuation are extracted from the local correlation coefficient sequence. The average correlation coefficient is obtained by summing all local correlation coefficients in the sequence and dividing the sum by the number of local correlation coefficients. The difference between each local correlation coefficient and the average correlation coefficient is calculated, and the squares of each difference are summed. The variance of correlation coefficient fluctuation is obtained by dividing the sum of squared differences by the number of local correlation coefficients. The average correlation coefficient and the variance of correlation coefficient fluctuation are used together as characteristics of braking friction stability.

[0051] Calculate the durations of the pre-compression stage, the pressure-building stage, and the stable friction stage. The duration of the pre-compression stage equals the end time of the pre-compression stage minus the start time of the braking process. The duration of the pressure-building stage equals the start time of the stable friction stage minus the end time of the pre-compression stage. The duration of the stable friction stage equals the start time of the release stage minus the start time of the stable friction stage. The durations of the pre-compression stage, the pressure-building stage, and the stable friction stage are collectively used as the stage duration characteristics.

[0052] The braking response delay feature, braking friction stability feature, and stage duration feature are combined and encoded. The braking response delay feature value is used as the abscissa component, and the maximum phase difference value is taken as the braking response delay feature value. The braking friction stability feature value is used as the ordinate component, and the average correlation coefficient is taken as the braking friction stability feature value. Dynamic performance feature points are generated in a two-dimensional coordinate system. The abscissa component of the dynamic performance feature points is the maximum phase difference value, and the ordinate component of the dynamic performance feature points is the average correlation coefficient.

[0053] The three durations of the phase duration feature are concatenated sequentially into an attribute byte segment. In specific implementation, the durations of the pre-pressure phase, pressure build-up phase, and stable friction phase are converted to milliseconds. These millisecond values ​​are then written into a byte array in the order of pre-pressure phase duration, pressure build-up phase duration, and stable friction phase duration. This byte array serves as the attribute byte segment. The attribute byte segment is appended to the dynamic performance feature point to form the dynamic performance test result of this braking process. The dynamic performance test result of this braking process includes the abscissa component, ordinate component, and attribute byte segment of the dynamic performance feature point.

[0054] The dynamic performance test results of multiple braking processes are connected in chronological order to obtain a multi-parameter dynamic performance test curve for the brake. In practice, a dynamic performance test result is generated for each braking process, and multiple dynamic performance test results are arranged in the order in which the braking processes occur. The dynamic performance test results corresponding to two adjacent braking processes are connected sequentially to form a complete dynamic performance test curve.

[0055] It can be understood that the maximum phase difference reflects the maximum delay of the braking hydraulic pressure relative to the piston displacement during the pressure build-up phase, and the moment of occurrence of the maximum phase difference reflects the position of the maximum delay on the time axis. The average correlation coefficient reflects the average correlation level between the housing vibration envelope and the friction interface temperature slope during the stable friction phase, and the variance of the correlation coefficient fluctuation reflects the degree of fluctuation in the correlation between the housing vibration envelope and the friction interface temperature slope. The stage duration feature reflects the duration distribution of different stages of the braking process. By using the braking response delay feature and the braking friction stability feature as two-dimensional coordinate components, and the stage duration feature as an additional attribute byte segment, dynamic performance test results containing multi-parameter performance information can be generated.

[0056] See Figure 5The horizontal axis of the graph represents the maximum extracted phase difference (in radians), and the vertical axis represents the corresponding average correlation coefficient, reflecting the relationship between braking response delay characteristics and braking friction stability characteristics. Each point in the graph represents a dynamic performance characteristic point of a single braking process. The color of the points is mapped according to the numerical duration of the stable friction phase. The color bars show a duration range of approximately 40 milliseconds to 160 milliseconds, with the color gradually changing from dark purple to light yellow, reflecting the distribution of different stable friction phase durations.

[0057] The figure shows an overall negative correlation between the maximum phase difference and the average correlation coefficient. That is, as the maximum phase difference increases, the average correlation coefficient gradually decreases. This indicates that the greater the maximum delay of the brake hydraulic pressure relative to the piston displacement, the lower the correlation between the housing vibration envelope and the slope of the friction interface temperature, reflecting a decrease in brake friction stability. The linear fitting trend line, represented by the red dashed line, clearly depicts this negative correlation. The fitting line extends from the upper left to the lower right, verifying the linear characteristic of the aforementioned trend.

[0058] As can be seen from the color mapping, data points with a shorter duration of the stable friction phase (darker purple) are mainly concentrated in the region with a smaller maximum phase difference and a higher average correlation coefficient, indicating that a shorter duration of the stable friction phase results in a smaller braking response delay and better friction stability. Conversely, data points with a longer duration of the stable friction phase (lighter yellow) are distributed in the region with a larger maximum phase difference and a lower average correlation coefficient, indicating that as the duration of the stable friction phase increases, the braking response delay increases and the friction stability decreases.

[0059] Furthermore, although the data points in the figure are somewhat dispersed, the overall trend is clear, which is consistent with the technical solution of dynamic performance evaluation by extracting multiple parameters such as the maximum phase difference, average correlation coefficient, and duration of the stable friction phase. This figure demonstrates how combining braking response delay characteristics and braking friction stability characteristics, supplemented by the duration attribute of the stable friction phase, achieves a multi-dimensional comprehensive characterization of the dynamic performance of low-altitude UAV brakes.

[0060] In practice, at the end of each braking process, the dynamic performance test results corresponding to that braking process are acquired. Braking response delay features are extracted from these results. These features include the maximum phase difference value and the time when the maximum phase difference occurs. The maximum phase difference value from the braking response delay features is used as the ordinate value of the curve point corresponding to that test.

[0061] The cumulative test time is used as the x-axis value of the curve point corresponding to this test. The cumulative test time is calculated from the start time of the first braking process, with the start time of the first braking process as the zero point of the cumulative test time. At the end of each braking process, the time elapsed from the zero point of the cumulative test time to the end time of this braking process is recorded, and the recorded time length is used as the x-axis value corresponding to this test.

[0062] A point on a two-dimensional plane is determined by the x-coordinate and y-coordinate values ​​corresponding to this test. This point on the two-dimensional plane is then used as the curve point for this test. A curve point is generated for each braking process, and all curve points are arranged in ascending order of cumulative test time.

[0063] In some embodiments, the curve points corresponding to two adjacent braking processes are connected by a straight line. The curve points corresponding to two adjacent braking processes refer to the curve points corresponding to two braking processes with consecutive cumulative test times. When connecting with a straight line, a straight line segment is drawn between the two endpoints, using the curve points corresponding to two adjacent braking processes as endpoints. All curve points corresponding to two adjacent braking processes are connected sequentially to form a connected polyline.

[0064] A moving average filter is applied at the connection point. In practice, moving average filtering is performed on each curve point on the broken line after the connection. The moving average filter uses a moving average window with a radius of 2. The reason for setting the moving average window radius to 2 is that the ordinate values ​​of the curve points near the connection point change continuously within a local range. A moving average window radius of 2 can cover two adjacent curve points on each side of the connection point, smoothing out local fluctuations while preserving the overall trend of the brake dynamic performance test curve. The moving average filter uses the following formula: in, The sequence number after moving average filtering is... The ordinate values ​​of the curve points; This indicates the index of the curve point on the connected polyline before and after the moving average filtering process; Represents the moving average window radius. The value is 2; This represents the offset within the moving average window. The range of values ​​is from arrive Integers; The sequence number on the line connecting the points before and after the moving average filtering process is... The ordinate values ​​of the curve points; This represents the total number of curve points within the moving average window.

[0065] In practical implementation, when performing moving average filtering on the curve points on the connected broken line, the sequence number... The curve points within the range of the moving average window are sequentially selected. For curve points at both ends of the connected polyline that are not covered by the moving average window, no moving average filtering is performed; the ordinate values ​​before moving average filtering are directly retained. After moving average filtering, the local angles at the connection points of the connected polyline are smoothed.

[0066] The complete curve formed after straight-line connection and moving average filtering is used as the dynamic test curve of the brake's multi-parameter performance. The dynamic test curve of the brake's multi-parameter performance is plotted with the cumulative test time on the x-axis and the maximum phase difference corresponding to each braking process on the y-axis, reflecting the change of the braking response delay characteristics with the cumulative test time during multiple braking processes.

[0067] It is understandable that by using the braking response delay characteristic as the ordinate and the cumulative test time as the abscissa to generate curve points at the end of each braking process, a time series of braking response delay characteristics between multiple braking processes can be established. Connecting the curve points corresponding to two adjacent braking processes with a straight line can form a continuous broken line. Moving average filtering can locally smooth the connection point, making the dynamic test curve of the brake's multi-parameter performance continuous at the connection point.

[0068] See Figure 6 In the graph, the horizontal axis represents the cumulative test time in seconds, starting from the beginning of the first braking process. The vertical axis represents the maximum phase difference corresponding to this braking process in radians. In the legend, orange dots indicate the curve points generated in each braking process, blue thin lines connect the curve points of two adjacent braking processes, and green dashed lines represent the smoothed curve after the connected broken line has been processed by moving average filtering.

[0069] As shown in the figure, the maximum phase difference exhibits a periodic fluctuation trend with the cumulative test time, ranging from approximately 0.15 to 0.7 radians. In the early test phase, the maximum phase difference fluctuates more drastically, with several peaks and troughs appearing. As the cumulative test time increases, the maximum phase difference generally shows an upward trend, and the fluctuation amplitude decreases relatively, indicating that the braking response delay characteristics gradually stabilize with repeated braking processes.

[0070] The broken line shape of the connecting line reflects the continuity of the maximum phase difference between adjacent braking processes, but there are sharp local changes. After the moving average filtering process, the abrupt change at the connection of the green dashed curve is smoothed, effectively reducing local fluctuations and highlighting the overall trend of the braking response delay characteristics.

[0071] The cumulative test time in the figure exceeds 2000 seconds, with a large number of test points covering multiple complete braking processes, reflecting the long-term dynamic changes in the braking response delay characteristics of low-altitude UAV brakes. The periodic fluctuations in the overall curve may be related to the periodic changes in the brake's working environment or operating status, but the specific reasons require further analysis in conjunction with other parameters.

[0072] 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 dynamic test method for multi-parameter performance of low-altitude unmanned aerial vehicle (UAV) brakes, characterized in that, Includes the following steps: Multi-parameter synchronous acquisition: Using the arrival time of the braking control command of the low-altitude UAV brake as the time reference, time series data of brake piston displacement, braking hydraulic pressure, friction interface temperature, shell vibration acceleration and motor drive current are synchronously acquired. Dynamic segmentation of the braking process: Based on the rising edge stagnation point and falling edge inflection point of the piston displacement time sequence, the braking process is divided into the pre-compression stage, the pressure building stage, the stable friction stage, and the release stage. Parameter interaction feature extraction: Extract the phase difference sequence between braking hydraulic pressure and piston displacement during the pressure build-up stage, and extract the local correlation coefficient sequence between the shell vibration acceleration envelope and the friction interface temperature slope during the stable friction stage; Dynamic performance evaluation: Based on the maximum phase difference in the phase difference sequence and the average correlation coefficient in the local correlation coefficient sequence, the dynamic performance test results of the brake are generated.

2. The method for dynamic testing of multi-parameter performance of low-altitude unmanned aerial vehicle (UAV) brakes according to claim 1, characterized in that, The multi-parameter synchronous acquisition step includes: A pressure sensor is installed in the brake cylinder, a displacement sensor is installed in the piston rod, a thermocouple is embedded on the back of the friction pad, an acceleration sensor is installed in the brake housing, and a current clamp is installed on the drive motor cable. After receiving the braking control command, the field-programmable logic device outputs a sampling pulse that is synchronized with the rising edge of the command, driving the multi-channel synchronous acquisition card to simultaneously acquire five types of signals from the pressure sensor, displacement sensor, thermocouple, accelerometer, and current clamp. Clock correction and resampling are performed on five time series corresponding to five types of signals to make the five time series have the same time interval.

3. The method for dynamic testing of multi-parameter performance of low-altitude unmanned aerial vehicle (UAV) brakes according to claim 1, characterized in that, The dynamic segmentation steps of the braking process include: A smoothed displacement sequence is obtained by low-pass filtering the piston displacement time series. The piston velocity sequence is obtained by taking the first derivative of the smooth displacement sequence; The moment when the velocity changes from zero to a positive value in the piston motion velocity sequence is taken as the starting moment of the braking process; The moment when the speed first reaches its peak and then begins to decline is taken as the end of the pre-compression phase. The moment when the velocity decreases from a positive value to zero while the displacement remains stable is taken as the starting point of the stable friction stage. The moment when the displacement begins to decrease and the velocity changes from zero to a negative value is taken as the starting point of the release phase.

4. The method for dynamic testing of multi-parameter performance of low-altitude unmanned aerial vehicle (UAV) brakes according to claim 1, characterized in that, The parameter interaction feature extraction step includes: During the pressure build-up phase, the brake hydraulic fluid time series and piston displacement time series are bandpass filtered respectively. Then, using the piston displacement time series as a reference series, the phase difference between the brake hydraulic fluid and the piston displacement in each window is obtained by sliding window cross-correlation peak detection to form a phase difference sequence. During the stable friction phase, the vibration envelope curve is obtained by envelope detection of the time series of shell vibration acceleration. The vibration envelope curve is divided into multiple integration windows with a fixed window length, and the vibration energy value in each integration window is calculated. Simultaneously, the slope value of the friction interface temperature time series within each integration window is calculated. After aligning the vibration energy value and the temperature slope value by amplitude normalization, the local correlation coefficient between the two is calculated window by window to form a local correlation coefficient sequence.

5. The method for dynamic testing of multi-parameter performance of low-altitude unmanned aerial vehicle (UAV) brakes according to claim 1, characterized in that, The dynamic performance evaluation steps include: The maximum phase difference and its occurrence time are extracted from the phase difference sequence and used as braking response delay features. The average correlation coefficient and the variance of correlation coefficient fluctuation are extracted from the local correlation coefficient sequence as characteristics of braking friction stability. The duration of the pre-compression stage, the pressure-building stage, and the stable friction stage is calculated, and the duration of these three stages is used as the stage duration characteristics. The braking response delay characteristics, braking friction stability characteristics, and stage duration characteristics are combined and encoded to obtain the dynamic performance test results of the braking process. By connecting the dynamic performance test results of multiple braking processes in chronological order, a multi-parameter performance dynamic test curve of the brake is obtained.

6. The method for dynamic testing of multi-parameter performance of low-altitude unmanned aerial vehicle (UAV) brakes according to claim 5, characterized in that, The step of combining and encoding the braking response delay characteristics, braking friction stability characteristics, and stage duration characteristics to obtain the dynamic performance test results of the braking process includes: The braking response delay characteristic value is used as the abscissa component, and the braking friction stability characteristic value is used as the ordinate component to generate dynamic performance characteristic points in a two-dimensional coordinate system. Concatenate the three durations in the stage duration feature into an attribute byte segment in sequence; The attribute byte segment is appended to the dynamic performance feature point to form the dynamic performance test result of this braking process.

7. The method for dynamic testing of multi-parameter performance of low-altitude unmanned aerial vehicle (UAV) brakes according to claim 3, characterized in that, The step of taking the moment when the speed first reaches its peak and then begins to decline as the end time of the pre-compression phase includes: Find the maximum velocity in the piston motion velocity sequence, and take a local minimum point before and after the maximum velocity; The moment corresponding to the maximum velocity is taken as the peak moment, and the moments corresponding to the next local minimum on both sides of the peak moment are taken as the front boundary and the back boundary, respectively. Find the point between the peak value and the front boundary where the velocity decreases at the maximum rate, and use this point as the end of the pre-compression phase.

8. The method for dynamic testing of multi-parameter performance of low-altitude unmanned aerial vehicle (UAV) brakes according to claim 2, characterized in that, The steps of clock correction and resampling for the five time series corresponding to the five types of signals include: The sampling timestamp of each channel is compared with the reference time count output by the field programmable logic device to calculate the clock offset of each channel; For channels whose clock offset exceeds the preset offset threshold, linear interpolation is used to correct the sampled values ​​to a uniform time grid. After calibration, each channel sequence is resampled to the same sampling frequency, and the first sampling point of each channel sequence is aligned with the arrival time of the braking control command.

9. The method for dynamic testing of multi-parameter performance of low-altitude unmanned aerial vehicle (UAV) brakes according to claim 4, characterized in that, The steps of performing bandpass filtering on the brake hydraulic time series and piston displacement time series respectively include: The center frequency is determined based on the peak value of the brake hydraulic time series during the pressure build-up phase, and the passband boundary is defined as 0.8 to 1.2 times the center frequency. A finite-length unit impulse response filter with a Kaiser window design is used to filter the brake hydraulic time series and piston displacement time series; After filtering, the two sequences are subjected to zero-phase round-trip filtering.

10. The method for dynamic testing of multi-parameter performance of low-altitude unmanned aerial vehicle (UAV) brakes according to claim 5, characterized in that, The step of connecting the dynamic performance test results of multiple braking processes in chronological order to obtain the dynamic test curve of the brake's multi-parameter performance includes: At the end of each braking process, the braking response delay feature in the dynamic performance test results is used as the vertical axis and the cumulative test time is used as the horizontal axis to generate a curve point corresponding to this test. Connect the curve points corresponding to two adjacent braking processes with a straight line, and perform moving average filtering at the connection point; The complete curve formed after connection is used as the dynamic test curve for the multi-parameter performance of the brake.

Images