Adaptive launch vehicle high-frequency vibration data acquisition method and system based on FFT

Abstract

The application provides an FFT-based adaptive launch vehicle high-frequency vibration data acquisition method and system, comprising: acquiring frequency domain data of initial acquisition signals; performing robust standardization processing on the frequency domain data to obtain standardized frequency domain data; acquiring frequency domain weights based on at least one of a current acquisition period and a state triggering event and dynamically weighting to obtain weighted frequency domain data; determining a cutoff frequency based on frequency domain energy accumulation of the weighted frequency domain data; determining a target sampling rate based on the cutoff frequency and performing secondary acquisition on the initial acquisition signals at the target sampling rate. The application determines the cutoff frequency through robust standardization and dynamic weighting, and adaptively adjusts the sampling rate, eliminates steady-state redundant data under the premise of retaining effective information, and improves telemetry code rate utilization and signal fidelity.

Description

Technical Field

[0001] This invention relates to the field of signal processing and data acquisition technology, specifically to an adaptive high-frequency vibration data acquisition method and system for launch vehicles based on FFT. Background Technology

[0002] In fields such as launch vehicle telemetry, high-frequency vibration parameters are crucial for assessing the structural health and engine operating status of the rocket. Traditional designs require high-frequency vibration data collection to consistently maintain the highest sampling frequency at each measurement point throughout the flight, resulting in massive amounts of vibration data per flight. For example, with a certain type of launch vehicle, high-frequency vibration parameter data accounts for approximately 21.8% of all telemetry data. However, these parameters remain relatively stable for most of the flight, with a large amount of steady-state redundant data consuming valuable telemetry bitrate resources, preventing the adequate transmission of more valuable state parameter information. Therefore, significantly reducing the amount of high-frequency vibration data while ensuring its validity is a pressing technical challenge.

[0003] Currently, various adaptive acquisition or data compression schemes exist, but none are suitable for scenarios where high-frequency vibration data from launch vehicles changes rapidly and has large amplitude variations. For example, high-frequency telemetry data compression methods based on polynomial fitting are post-processing compression methods, where the acquisition end still uses a fixed high sampling rate throughout the entire process, without reducing the amount of data buffered on the rocket, and the fitting error is large for non-stationary and severe vibration signals. Adaptive sampling frequency tracking methods based on power spectrum energy accumulation are designed for simple mechanical vibration scenarios, lack a weighting mechanism related to flight time sequence, and do not perform robust preprocessing on frequency domain data. When outliers or impulse noise are present, the normalization result is easily distorted by extreme values, leading to inaccurate cutoff frequency determination. Adaptive frequency conversion acquisition strategies based on the rotating door algorithm are suitable for slowly varying parameters but not for high-frequency vibration data. Wireless sensor network adaptive acquisition schemes suffer from large acquisition distortion and insufficient real-time performance, failing to meet the requirements of high-fidelity and low-latency acquisition.

[0004] In summary, existing adaptive acquisition methods are not suitable for engineering scenarios where high-frequency vibration data of launch vehicles change rapidly, have large amplitude variations, and have relatively fixed transmission frame formats. There is an urgent need for a high-frequency vibration adaptive acquisition method that can perform frequency domain analysis in real time on the rocket, is robust, correlates with flight timing, and is equipped with improved telemetry frame formats. Summary of the Invention

[0005] To address the shortcomings of existing technologies, the purpose of this invention is to provide an adaptive high-frequency vibration data acquisition method and system for launch vehicles based on FFT.

[0006] An adaptive acquisition method for high-frequency vibration data of a launch vehicle based on FFT is provided by the present invention, the method comprising: Acquire the frequency domain data of the initial acquired signal; The frequency domain data is subjected to robust normalization to obtain normalized frequency domain data; Based on at least one of the current acquisition period and state triggering events, obtain frequency domain weights, and dynamically weight the standardized frequency domain data based on the frequency domain weights to obtain weighted frequency domain data; The cutoff frequency is determined based on the frequency domain energy accumulation of the weighted frequency domain data. The target sampling rate is determined based on the cutoff frequency, and the initial acquisition signal is acquired a second time according to the target sampling rate.

[0007] Preferably, the robust normalization process performed on the frequency domain data to obtain normalized frequency domain data includes: Determine the median and interquartile range of the frequency domain data; Based on the median and the interquartile range, the amplitude of each frequency point in the frequency domain data is scaled and normalized to obtain the standardized frequency domain data.

[0008] Preferably, the step of obtaining frequency domain weights based on at least one of the current acquisition period and state trigger events, and dynamically weighting the standardized frequency domain data based on the frequency domain weights to obtain weighted frequency domain data, includes: Using at least one of the current collection period and the state trigger event as index conditions, the weight coefficients corresponding to each frequency interval are obtained from a preset weight lookup table; The weighted frequency domain data is obtained by multiplying the energy value of each frequency point in the standardized frequency domain data with the corresponding weighting coefficient point by point.

[0009] Preferably, the weight coefficients in the weight lookup table are calculated using a nonlinear weighting model, the expression of which is:

[0010] in, Current data collection period The weighting coefficients, Current flight time; This is the default value for the weights; For relative to the first The relative time of each state-triggered event The weight correction function, The distance from the current time to the first The time difference between the occurrence of each event; For the first Frequency weighting correction coefficient for each state-triggered event.

[0011] Preferably, determining the cutoff frequency based on the frequency domain energy accumulation of the weighted frequency domain data includes: The energy values ​​of each frequency point in the weighted frequency domain data are accumulated point by point in ascending order of frequency to obtain the accumulated energy. When the proportion of the accumulated energy to the total energy reaches a preset threshold factor for the first time, the frequency corresponding to the current frequency point is determined as the cutoff frequency.

[0012] Preferably, the step of determining the target sampling rate based on the cutoff frequency and performing a second sampling of the initial acquisition signal according to the target sampling rate includes: The initial sampling rate is determined by a preset multiple of the cutoff frequency, and the initial sampling rate is aligned to an integer multiple of the transmission frame frequency to obtain the target sampling rate. Sampling points are extracted from the initial acquisition signal according to the target sampling rate, and the extracted sampling points are framed and transmitted together with the state variable representing the target sampling rate.

[0013] Preferably, acquiring the frequency domain data of the initial acquired signal includes: Data is collected at equal intervals using a fixed sampling rate, and the collected time-domain data is cached in a sliding window. The time-domain data within the sliding window is subjected to a Fast Fourier Transform to obtain the frequency-domain data; The method is executed in parallel in a pipelined manner within a field-programmable gate array.

[0014] An adaptive acquisition system for high-frequency vibration data of a launch vehicle based on FFT, provided by the present invention, includes: The frequency domain acquisition module is used to acquire the frequency domain data of the initial acquired signal; A robust normalization module is used to perform robust normalization processing on the frequency domain data to obtain normalized frequency domain data. The dynamic weighting module is used to obtain frequency domain weights based on at least one of the current acquisition period and state trigger events, and to dynamically weight the standardized frequency domain data based on the frequency domain weights to obtain weighted frequency domain data. The frequency determination module is used to determine the cutoff frequency based on the frequency domain energy accumulation of the weighted frequency domain data; The secondary acquisition module is used to determine the target sampling rate based on the cutoff frequency, and to perform secondary acquisition on the initial acquisition signal according to the target sampling rate.

[0015] According to the present invention, a computer-readable storage medium is provided thereon storing a computer program, which, when executed by a processor, implements the aforementioned FFT-based adaptive acquisition method for high-frequency vibration data of launch vehicles.

[0016] According to the present invention, an aircraft includes a vibration sensor, an analog-to-digital converter, a core processing unit, and a telemetry transmitter. The vibration sensor is used to collect high-frequency vibration signals; The analog-to-digital converter is used to sample and quantize the high-frequency vibration signal and output an initial acquisition signal; The core processing unit is configured to execute the FFT-based adaptive acquisition method for high-frequency vibration data of launch vehicles. The telemetry transmitter is used to frame and transmit the data obtained from the secondary acquisition.

[0017] Compared with the prior art, the present invention has the following beneficial effects: 1. This invention performs robust normalization processing on frequency domain data based on median and interquartile range, and uses a statistically robust estimator to replace the traditional mean and standard deviation for normalization scaling. This suppresses the distortion effect of outliers and impulse noise on the frequency domain energy distribution, and improves the robustness of cutoff frequency determination under strong noise interference environment.

[0018] 2. This invention obtains frequency domain weights based on the current acquisition period and state trigger events, and dynamically weights the standardized frequency domain data. It maps the prior working state of the signal source and the time sequence stage to a nonlinear frequency weight correction coefficient, so that the frequency domain energy accumulation process can adaptively match the time-varying characteristics of non-stationary signals in different flight stages, and realizes the accurate tracking of the cutoff frequency to the real vibration characteristics.

[0019] 3. This invention determines the cutoff frequency through the synergistic effect of robust standardization and dynamic weighting, and adjusts the target sampling rate and frame downlink strategy for secondary acquisition accordingly. While ensuring the complete preservation of effective signal information, it significantly eliminates steady-state redundant data, achieving a significant improvement in telemetry code rate resource utilization and an optimal balance between signal fidelity. Attached Figure Description

[0020] Other features, objects, and advantages of the present invention will become more apparent from the following detailed description of non-limiting embodiments with reference to the accompanying drawings: Figure 1 This is a flowchart of the signal adaptive acquisition method according to an embodiment of this application; Figure 2 This is a hardware system architecture diagram of the adaptive signal acquisition system according to an embodiment of this application. Detailed Implementation

[0021] The present invention will now be described in detail with reference to specific embodiments. These embodiments will help those skilled in the art to further understand the present invention, but do not limit the invention in any way. It should be noted that those skilled in the art can make several changes and improvements without departing from the concept of the present invention. These all fall within the protection scope of the present invention.

[0022] like Figure 1 and Figure 2 As shown, this invention provides an adaptive acquisition method for high-frequency vibration data of launch vehicles based on FFT (Fast Fourier Transform, hereinafter referred to as FFT). It is applied to the high-frequency vibration data acquisition scenario of launch vehicles and solves the problems that the cutoff frequency determination is easily distorted by extreme values ​​and cannot adapt to the time-varying characteristics of the signal under strong noise and non-stationary signal environments.

[0023] like Figure 1 As shown, the method of this invention is divided into two stages: front-end data acquisition and processing, and back-end data acquisition and processing. The front-end data acquisition and processing stage includes, in sequence: fixed-frequency data acquisition, i.e., acquiring the vibration sensor output signal at equal intervals with a fixed sampling rate; establishing a sliding window, i.e., establishing a sliding window of length N on the acquired time-domain data; FFT of the dataset within the sliding window, i.e., performing a Fast Fourier Transform on the time-domain dataset within the sliding window to obtain frequency-domain data; robust standardization of frequency amplitude, i.e., performing robust standardization on the frequency amplitude of the frequency-domain data to eliminate the influence of extreme values; setting nonlinear weights based on time and event triggers, i.e., obtaining the nonlinear weight coefficients of each frequency point through a weight lookup table based on the current acquisition period and state-triggered events; and adaptively adjusting the cutoff frequency using an energy accumulation method, i.e., adaptively determining the cutoff frequency based on the frequency-domain energy accumulation of the weighted frequency-domain data. The backend data acquisition and processing stages include, in sequence: setting the frequency of secondary acquisition, i.e. determining the target sampling rate based on the cutoff frequency; secondary acquisition of vibration data, i.e., extracting sampling points from the initial acquisition signal according to the target sampling rate for secondary acquisition; and vibration data framing and downlinking, i.e., framing the secondary acquired vibration data according to the improved frame format and downlinking it to the telemetry system.

[0024] This embodiment also provides a hardware execution environment for the above method, which includes a vibration sensor, an analog-to-digital converter (ADC), a field-programmable gate array (FPGA) as the core processing unit, and a telemetry transmitter.

[0025] The FFT-based adaptive acquisition method for high-frequency vibration data of launch vehicles provided in this embodiment includes acquiring the frequency domain data of the initial acquisition signal. This step is achieved by combining fixed-interval initial acquisition with a sliding window buffer.

[0026] The system first collects high-frequency vibration data from the launch vehicle at fixed sampling rates at equal intervals to obtain time-domain vibration data sequences. To ensure the real-time performance and continuity of subsequent frequency domain analysis, a sliding window of length N is established for the collected time-domain data. The window slides one sampling point for each new data point received. In this embodiment, the length N of the sliding window is preferably an integer power of 2, such as N=1024, to facilitate subsequent FFT operations. In terms of hardware implementation, the FPGA's on-chip Block RAM is configured as a dual-port RAM, using a ping-pong operation. One portion is used to store the latest N time-domain sampling points continuously acquired by the ADC, while the other portion is used for backend processing, thereby achieving parallel pipelined execution of data acquisition and processing and avoiding data throughput bottlenecks.

[0027] After acquiring the time-domain data, an FFT is performed on the time-domain data within the sliding window to obtain the frequency-domain data. The Discrete Fourier Transform (DFT) is the fundamental method for converting time-domain data into frequency-domain data. For a sliding window of length... Time domain data Its DFT can be expressed as:

[0028] in Frequency point number The time-domain sampling point number, For the first Complex spectral values ​​corresponding to each frequency point, rotation factor , It is the imaginary unit.

[0029] Directly calculating the DFT requires Multiplication of complex numbers and Complex additions are computationally intensive when N is large. The FFT, leveraging the periodicity and symmetry of the DFT, uses a butterfly algorithm to decompose a long DFT into multiple shorter DFTs, dividing the N points into... odd points and For an even number of points, the computational complexity is reduced by nearly half after one decomposition, and this process is repeated recursively. For example, the number of DFT operations is approximately 204 times that of FFT operations.

[0030] In terms of hardware implementation, this embodiment employs a radix-2 time-domain decimation algorithm to decompose the N-point DFT into multiple butterfly operation units. Inside the FPGA, a fully pipelined architecture is constructed using a hardware description language, containing 10 butterfly operation units. Each stage performs parallel computation in time via a pipelined approach, and complex multiplication operations are implemented using on-chip DSP slice resources. Actual measurements show that the time to complete a 1024-point FFT calculation is less than 10μs, far less than the update cycle of the sliding window (e.g., 204.8ms), ensuring the system can respond to signal changes in real time. The specific number of points and pipeline stages of the FFT can be adjusted according to actual hardware resources and signal bandwidth requirements, and is not limited to the 1024-point and 10-stage structure described in this embodiment.

[0031] After obtaining the frequency domain data, robust normalization is performed to obtain normalized frequency domain data. This step is used to suppress the distortion of frequency domain energy distribution by outliers and impulse noise. Robust normalization involves determining the median and interquartile range of the frequency domain data, and scaling and normalizing the amplitude of each frequency point in the frequency domain data based on the median and interquartile range. In this embodiment, the Robust normalization method is used, and its calculation formula is as follows:

[0032] in, This represents the data at the k-th frequency point after Robust normalization. This represents the original frequency domain amplitude data. This represents the median of the frequency domain dataset. This represents the interquartile range, and .in, The first quartile is the value at the 25th percentile after the data is arranged in ascending order; It is the third quartile, which is the value at the 75th percentile after the data is arranged in ascending order.

[0033] Compared to traditional normalization methods based on mean and standard deviation, Robust standardization utilizes statistically robust estimators, is insensitive to extreme outliers, and effectively preserves the overall trend and distribution characteristics of the data. Experimental results show that after Robust standardization preprocessing, the amount of data generated can be reduced by approximately 0.2%, compared to the case without this processing, demonstrating its effectiveness in removing invalid noise and redundancy.

[0034] Next, based on at least one of the current acquisition period and state trigger events, frequency domain weights are obtained, and the standardized frequency domain data is dynamically weighted based on these weights to obtain weighted frequency domain data. This step enables the frequency domain energy accumulation process to adaptively match the time-varying characteristics of non-stationary signals at different flight stages. Using at least one of the current acquisition period and state trigger events as index conditions, the weight coefficients corresponding to each frequency interval are obtained from a preset weight lookup table, and the energy value of each frequency point in the standardized frequency domain data is multiplied point by point with the corresponding weight coefficient. A "state trigger event" refers to the prior operating state or timing stage of the signal source, such as engine ignition or stage separation of a launch vehicle, rather than a simple timed interruption. The weight lookup table is stored in the FPGA's on-chip ROM, recording the spectral characteristic patterns under different time periods and events summarized from past flight experiences. High-frequency vibration data from different launches of the same model exhibit a certain correlation, and the high-frequency vibration data near characteristic events show similar trends. For example, data from past flights of a certain type of launch vehicle shows that the high-frequency vibration measurement points on the second-stage liquid engine support exhibit significantly increased vibration amplitude and frequency during the operation of the second-stage engine, while the data remain stable during other periods.

[0035] The weight coefficients in the weight lookup table are calculated using a nonlinear weighting model, the expression of which is:

[0036] in, Current data collection period The weighting coefficients, Current flight time; This is the default weight, representing the basic weight under steady-state conditions; For relative to the first The relative time of each state-triggered event The weight correction function, The distance from the current time to the first The time difference between the occurrence of each event is used to describe the dynamic decay or change of the weight over time after the event occurs. For the first The frequency weight correction coefficients for each state-triggered event reflect the excitation intensity of a specific event on vibrations in a specific frequency band. Through this time- and event-triggered nonlinear weighting mechanism, the system can automatically increase the weight of the corresponding frequency band during periods of severe vibration, such as engine operation, making the cutoff frequency determination more closely match the actual vibration characteristics and overcoming the shortcomings of fixed-weight methods that cannot adapt to complex excitation environment changes.

[0037] Subsequently, the cutoff frequency is determined based on the frequency domain energy accumulation of the weighted frequency domain data. The energy density at each frequency point is...

[0038] in, For the first Weighted energy density at each frequency point For Robust's standardized version Individual frequency data, For the first The weighting coefficients corresponding to each frequency point. The total energy of each frequency point is... ,in This is the sum of the energy densities at all frequencies. The energy density at each frequency is accumulated point-by-point in ascending order of frequency. The point is reached when the proportion of accumulated energy to total energy first reaches a preset threshold factor. At that time, the corresponding frequency point number The corresponding frequency is the cutoff frequency, which can be expressed as: ,in This is the mapping function from frequency point number to actual frequency.

[0039] The selection of the threshold factor directly determines the balance between data compression ratio and signal fidelity. In this embodiment, the threshold factor... The preferred setting is 0.9999. Comparative testing has verified that when… At that time, although all information was retained, the data compression rate was only 70.45%, and a large amount of noise was not filtered out; when At that time, although the compression ratio dropped to 0.23%, high-frequency effective signals began to be lost; and when At that time, the data compression rate was 12.91%, which, while removing zero-frequency repetitive data and most noise, completely preserved the effective vibration information, achieving an optimal balance. Threshold factor The specific value can be adaptively adjusted according to the different requirements of accuracy and bandwidth in the actual application scenario, and is usually selected in the range of 0.9 to 0.9999.

[0040] After determining the cutoff frequency, a target sampling rate is determined based on the cutoff frequency, and the initially acquired signal is sampled a second time according to the target sampling rate. A preset multiple of the cutoff frequency is used as the initial sampling rate, and this initial sampling rate is aligned to an integer multiple of the transmission frame frequency to obtain the target sampling rate. Based on the Nyquist sampling theorem and engineering experience, the preset multiple is typically set to 2.5 times or higher to ensure distortion-free signal reconstruction. Simultaneously, to adapt to the fixed frame format constraints of the telemetry system, the calculated sampling rate needs to be aligned to the telemetry frame frequency (e.g., 40Hz). times ( =0,1,2..., (is a non-negative integer).

[0041] During the secondary acquisition process, sampling points are extracted from the time-domain data buffered in the sliding window according to the target sampling rate. The extracted sampling points are then framed and transmitted along with the state variables representing the target sampling rate. The insertion of the state variables enables the ground receiver to reverse-analyze and reconstruct the vibration data of each measurement point based on the state information of each frame, thereby achieving compatible transmission of variable sampling rate data without changing the overall architecture of the telemetry system.

[0042] Corresponding to the above method, this embodiment also provides an adaptive acquisition system for high-frequency vibration data of launch vehicles based on FFT. This system includes a frequency domain acquisition module, a robust normalization module, a dynamic weighting module, a frequency determination module, and a secondary acquisition module. Each module is implemented in hardware logic within an FPGA and works collaboratively through a pipeline.

[0043] The FPGA chip serves as the core processing unit of the system and performs the following functions: The FFT processing engine consists of a pipelined structure (butterfly pipeline) composed of cascaded multi-stage butterfly operation units (i.e., 16 DSP slices). It employs a radix-2 time-domain decimation algorithm and utilizes the on-chip FPGA DSP slices to perform complex multiplication. It performs FFT transformation on N points of time-domain data within a sliding window. Under the fully pipelined architecture, the 10 stages of butterfly units operate in parallel, with a single 1024-point FFT operation taking no more than 10μs. The on-chip Block RAM (BRAM) is configured as a dual-port RAM. One part forms the sliding window data buffer, storing the latest N time-domain sampling points continuously acquired by the ADC; the other part forms the frequency domain data buffer, storing the frequency domain data output by the FFT. The dual-port structure supports ping-pong operation, enabling parallel acquisition and processing. The Robust normalization module consists of sorting and statistical circuits and operation units, calculating the median and IQR of the frequency domain data in real time and performing normalization calculations. The weighted energy accumulation module (i.e., weighted + energy accumulation) includes a ROM-based weight lookup table that stores the weight coefficients for each frequency interval associated with flight time and events; a multiplier array performs point-by-point multiplication of frequency domain energy and weights; and an accumulator accumulates the weighted energy in ascending frequency order and compares it with a threshold, outputting the cutoff frequency. The secondary sampling control module includes an address generator, which operates based on the cutoff frequency. Generate a secondary sampling address sequence, with a sampling rate of [value missing]. And align the time-domain data to an integer multiple of the telemetry frame rate, extracting time-domain data from BlockRAM. The telemetry framing module consists of parallel-to-serial conversion logic, state variable insertion logic, and FIFO, and performs secondary sampling (corresponding to the attached...) Figure 2The data and status variables of the three sub-modules (parallel-to-serial conversion, state variable insertion, and FIFO) under the telemetry framing module are packaged according to the improved frame format and output as a serial telemetry code stream. The control state machine acts as a global timing control unit, generating enable, handshake, and reset signals (i.e., clock reset) for each module and managing pipeline synchronization.

[0044] The onboard computer sends flight time reference signals and event trigger signals (such as interstage separation and engine ignition) to the FPGA via the onboard bus, which serve as address indexes for the weighted lookup table. The telemetry transmitter receives the telemetry bitstream output from the FPGA, modulates and amplifies it, and then transmits it to the ground via an antenna. All algorithm functions in this hardware architecture are implemented in hardware using parallel pipelined logic on the FPGA chip, without the need for an embedded processor. The total processing delay from data acquisition to cutoff frequency determination for a single window is no more than 20μs, far less than the sliding window update cycle of 204.8ms, meeting the engineering requirements for real-time adaptive acquisition of high-frequency vibration data during launch vehicle flight.

[0045] This embodiment also provides a computer-readable storage medium storing a computer program thereon. When executed by a processor, this computer program implements the FFT-based adaptive acquisition method for high-frequency vibration data of a launch vehicle as described in any of the preceding embodiments. The technical solution of this invention can be implemented not only in pure hardware logic but also as a software program product deployed in various general-purpose or special-purpose processors, thus possessing wide applicability.

[0046] This embodiment also provides an aircraft, specifically a launch vehicle. The aircraft includes a vibration sensor, an analog-to-digital converter (ADC), a core processing unit, and a telemetry transmitter. The vibration sensor is installed at key parts of the rocket body, with a measurement range covering 20Hz~2000Hz, used to collect high-frequency vibration signals. The analog input of the ADC is connected to the vibration sensor, and the analog signal is sampled and quantized at equal intervals with a fixed sampling rate, outputting 16-bit parallel digital data to the core processing unit. The core processing unit (such as an FPGA) is configured to execute the aforementioned FFT-based adaptive acquisition method for high-frequency vibration data of the launch vehicle. Its on-chip implementation uses a hardware description language to implement functional modules such as an FFT processing engine, on-chip Block RAM, Robust normalization module, weighted and energy accumulation module, secondary sampling control module, telemetry framing module, and control state machine. The telemetry transmitter is used to frame and transmit the data obtained from the secondary acquisition.

[0047] Verified by actual flight data of a certain type of launch vehicle, the proposed solution reduces the amount of secondary data acquisition from 26 high-frequency vibration measurement points to only 7.84% of the primary acquisition amount, achieving an overall data compression rate of 92.16%. This saves 0.6–1.29 Mbps of telemetry code rate for transmitting other high-value parameters. The effective time-domain values ​​of the secondary acquisition data are essentially equal to those of the original data (e.g., the effective value of a measurement point changes from 127.9895 to 127.9467), the Pearson correlation coefficient is 1, and the global mean absolute error (GMAE) is only 0.4413, far smaller than the data range [0, 255]. Furthermore, the single-window data processing delay is less than 20 μs. These data fully demonstrate that the present invention significantly improves data transmission efficiency while maintaining high signal fidelity, meeting the engineering requirements of launch vehicles for real-time and accurate acquisition of high-frequency vibration data.

[0048] Those skilled in the art will understand that, besides implementing the system and its various devices, modules, and units provided by this invention in the form of purely computer-readable program code, the same functions can be achieved entirely through logical programming of the method steps, making the system and its various devices, modules, and units of this invention function in the form of logic gates, switches, application-specific integrated circuits, programmable logic controllers, and embedded microcontrollers. Therefore, the system and its various devices, modules, and units provided by this invention can be considered as a hardware component, and the devices, modules, and units included therein for implementing various functions can also be considered as structures within the hardware component; alternatively, the devices, modules, and units for implementing various functions can be considered as both software modules implementing the method and structures within the hardware component.

[0049] Specific embodiments of the present invention have been described above. It should be understood that the present invention is not limited to the specific embodiments described above, and those skilled in the art can make various changes or modifications within the scope of the claims, which do not affect the essence of the present invention. Unless otherwise specified, the embodiments and features described in this application can be arbitrarily combined with each other.

Claims

1. An adaptive acquisition method for high-frequency vibration data of launch vehicles based on FFT, characterized in that, The method includes: Acquire the frequency domain data of the initially acquired signal; The frequency domain data is subjected to robust normalization to obtain normalized frequency domain data; Based on at least one of the current acquisition period and state triggering events, obtain frequency domain weights, and dynamically weight the standardized frequency domain data based on the frequency domain weights to obtain weighted frequency domain data; The cutoff frequency is determined based on the frequency domain energy accumulation of the weighted frequency domain data. The target sampling rate is determined based on the cutoff frequency, and the initial acquisition signal is acquired a second time according to the target sampling rate.

2. The adaptive acquisition method for high-frequency vibration data of launch vehicles based on FFT according to claim 1, characterized in that, The process of robustly normalizing the frequency domain data to obtain standardized frequency domain data includes: Determine the median and interquartile range of the frequency domain data; Based on the median and the interquartile range, the amplitude of each frequency point in the frequency domain data is scaled and normalized to obtain the standardized frequency domain data.

3. The adaptive acquisition method for high-frequency vibration data of launch vehicles based on FFT according to claim 1, characterized in that, The process of obtaining frequency domain weights based on at least one of the current acquisition period and state-triggered events, and then dynamically weighting the standardized frequency domain data based on the frequency domain weights to obtain weighted frequency domain data, includes: Using at least one of the current collection period and the state trigger event as index conditions, the weight coefficients corresponding to each frequency interval are obtained from a preset weight lookup table; The weighted frequency domain data is obtained by multiplying the energy value of each frequency point in the standardized frequency domain data with the corresponding weighting coefficient point by point.

4. The adaptive acquisition method for high-frequency vibration data of launch vehicles based on FFT according to claim 3, characterized in that, The weight coefficients in the weight lookup table are calculated using a nonlinear weight model, the expression of which is: in, Current data collection period The weighting coefficients, Current flight time; This is the default value for the weights; For relative to the first The relative time of each state-triggered event The weight correction function, The distance from the current time to the first The time difference between the occurrence of each event; For the first Frequency weighting correction coefficient for each state-triggered event.

5. The adaptive acquisition method for high-frequency vibration data of launch vehicles based on FFT according to claim 1, characterized in that, The determination of the cutoff frequency based on the frequency domain energy accumulation of the weighted frequency domain data includes: The energy values ​​of each frequency point in the weighted frequency domain data are accumulated point by point in ascending order of frequency to obtain the accumulated energy. When the proportion of the accumulated energy to the total energy reaches a preset threshold factor for the first time, the frequency corresponding to the current frequency point is determined as the cutoff frequency.

6. The adaptive acquisition method for high-frequency vibration data of launch vehicles based on FFT according to claim 5, characterized in that, The step of determining the target sampling rate based on the cutoff frequency and performing a second sampling of the initial acquisition signal according to the target sampling rate includes: The initial sampling rate is determined by a preset multiple of the cutoff frequency, and the initial sampling rate is aligned to an integer multiple of the transmission frame frequency to obtain the target sampling rate. Sampling points are extracted from the initial acquisition signal according to the target sampling rate, and the extracted sampling points are framed and transmitted together with the state variable representing the target sampling rate.

7. The adaptive acquisition method for high-frequency vibration data of launch vehicles based on FFT according to claim 1, characterized in that, The acquisition of frequency domain data of the initial acquired signal includes: Data is collected at equal intervals using a fixed sampling rate, and the collected time-domain data is cached in a sliding window. The time-domain data within the sliding window is subjected to a Fast Fourier Transform to obtain the frequency-domain data; The method is executed in parallel in a pipelined manner within a field-programmable gate array.

8. An adaptive acquisition system for high-frequency vibration data of a launch vehicle based on FFT, characterized in that, include: The frequency domain acquisition module is used to acquire the frequency domain data of the initial acquired signal; A robust normalization module is used to perform robust normalization processing on the frequency domain data to obtain normalized frequency domain data. The dynamic weighting module is used to obtain frequency domain weights based on at least one of the current acquisition period and state trigger events, and to dynamically weight the standardized frequency domain data based on the frequency domain weights to obtain weighted frequency domain data. The frequency determination module is used to determine the cutoff frequency based on the frequency domain energy accumulation of the weighted frequency domain data; The secondary acquisition module is used to determine the target sampling rate based on the cutoff frequency, and to perform secondary acquisition on the initial acquisition signal according to the target sampling rate.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the FFT-based adaptive acquisition method for high-frequency vibration data of launch vehicles as described in any one of claims 1 to 7.

10. An aircraft, characterized in that, The aircraft includes a vibration sensor, an analog-to-digital converter, a core processing unit, and a telemetry transmitter; The vibration sensor is used to collect high-frequency vibration signals; The analog-to-digital converter is used to sample and quantize the high-frequency vibration signal and output an initial acquisition signal; The core processing unit is configured to execute the FFT-based adaptive acquisition method for high-frequency vibration data of launch vehicles as described in any one of claims 1 to 7; The telemetry transmitter is used to frame and transmit the data obtained from the secondary acquisition.

Images