A Noise Prediction Method and System for Unmanned Aerial Vehicles Based on Flight Parameter Sensing

By deploying a microphone array and flight parameter sensing module on the drone, noise signals are collected and processed in real time, solving the problems of real-time and accuracy in drone noise prediction, adapting to urban environments, and supporting active noise reduction and flight path optimization.

CN123111454APending Publication Date: 2026-11-17BEIJING TUSHENG TIANDI TECH CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202611335809.9
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-08-31
Publication Date
2026-11-17

AI Technical Summary

Technical Problem

Existing technologies lack real-time online prediction capabilities, making it impossible to accurately predict noise impact based on the real-time flight status of drones. Furthermore, noise propagation models are not precise enough in urban environments, hindering the effective coupling of flight parameters and noise prediction.

Method used

By deploying a microphone array on the drone to collect near-field noise signals in real time, and combining it with a flight parameter sensing module, the noise source intensity is extracted in real time. A preset propagation attenuation model is used to predict the noise impact, including filtering out wind noise and pseudo-sound components of aircraft vibration, and performing background noise separation and sound source identification.

Benefits of technology

It achieves high-precision noise impact prediction with a time delay of up to the second during UAV flight, adapts to complex urban environments, provides active noise reduction and route optimization decision support, and improves prediction accuracy and applicability.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN123111454A_ABST
    Figure CN123111454A_ABST
Patent Text Reader

Abstract

This invention discloses a method and system for predicting drone noise based on flight parameter perception, comprising: acquiring drone flight status data and raw sound pressure signals synchronously collected by an onboard microphone array; preprocessing the raw sound pressure signals to filter out wind noise components and compensate for body vibration pseudo-sound; calculating the blade passing frequency based on the real-time rotor speed, extracting rotor noise harmonic components and performing background noise separation to obtain the self-noise spectrum; converting the sound power level of the drone; and calculating the predicted sound pressure level at the target receiving point using a propagation attenuation model. This invention achieves real-time online perception of noise source intensity during flight by deploying a microphone array on the drone, and combines flight parameters to achieve background noise separation and self-noise extraction, significantly improving the real-time performance and accuracy of urban low-altitude drone noise prediction.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention relates to the field of environmental noise prediction and control technology, specifically to a method and system for real-time prediction of noise impact in low-altitude civil unmanned aerial vehicles (hereinafter referred to as "UAVs") operation scenarios, which is particularly suitable for noise impact assessment and proactive management in urban low-altitude logistics distribution, inspection operations, and passenger transportation operation scenarios. Background Technology

[0002] With the rapid development of the low-altitude economy, the frequency and density of civilian drone operations in urban environments have increased dramatically. The noise generated by drones during flight has become a significant factor affecting the quality of life for urban residents and is also one of the key bottlenecks restricting the large-scale development of low-altitude operations.

[0003] From the perspective of sound source mechanism, drone noise mainly includes the following two categories:

[0004] (1) Rotor noise: including rotational noise (discrete sounds at the blade passing frequency and its harmonics) and broadband noise (turbulent intake noise, blade self-noise, etc.). The intensity and spectral characteristics of rotor noise directly depend on flight parameters such as rotor speed, blade pitch angle, and incoming flow velocity, and show significant differences under different flight conditions such as hovering, climbing, and forward flight.

[0005] (2) Fuselage aerodynamic noise: broadband noise generated by the interaction between the turbulent boundary layer on the fuselage surface and the fuselage structure, as well as flow separation noise caused by protrusions such as landing gear and load racks. Fuselage aerodynamic noise is closely related to flight speed and fuselage attitude angle.

[0006] In existing technologies, the assessment of the noise impact of drones mainly adopts the following methods: Method 1: Post-event measurement. Noise monitoring stations are deployed on the ground to measure and record the sound pressure level as the drone flies overhead. This method only acquires data on noise events that have already occurred; it lacks predictive capabilities and cannot determine the range and extent of noise impact before or during flight.

[0007] Method 2: Offline Simulation. This method utilizes computational fluid dynamics (CFD) and computational aeroacoustics (CAA) to numerically simulate the noise field of a UAV under specific flight conditions. However, this method is computationally intensive and time-consuming, failing to meet real-time requirements, and is typically only used for acoustic performance evaluation during the design phase.

[0008] Method 3: Simplified assessment method based on statistical models. This method uses empirical formulas or statistical regression models to estimate ground noise levels based on coarse-grained parameters such as aircraft type and flight altitude, as exemplified by the simplified noise exposure calculation used by some UAV operation management platforms. This type of method does not consider the dynamic impact of real-time operating parameters such as flight attitude and rotor speed on noise source intensity, resulting in lower prediction accuracy.

[0009] In summary, the existing technology has the following main drawbacks: First, there is a lack of real-time online prediction capabilities. Existing methods either rely on post-flight measurements, offline simulations, or static queries of pre-built databases, none of which can predict the impact of noise on the ground based on real-time flight status during UAV flight. This makes it impossible to provide timely feedforward information for dynamic flight path adjustments and proactive noise reduction decisions.

[0010] Second, the method of obtaining sound source intensity deviates from real-world operating conditions. Existing simplified assessment methods primarily rely on prior wind tunnel calibration experiments or limited measured data to form a fixed noise database for the specific aircraft model. However, the noise radiation characteristics of UAVs in actual operation are affected by various factors such as payload weight, environmental weather, equipment status, and flight attitude, resulting in significant deviations from the calibration conditions. This "calibrate first, then query" approach fails to reflect the dynamic changes of sound sources during actual flight, leading to predictions that are severely inconsistent with reality.

[0011] Third, noise propagation models are not sufficiently adaptable to urban built environments. Existing technologies typically only use geometrical divergence attenuation and simple atmospheric absorption correction in noise propagation calculations, failing to fully consider the blocking, diffraction, and reflection effects of buildings on sound wave propagation in urban environments, and also failing to accurately account for the impact of ground type (hard ground, soft ground, mixed ground) on sound propagation, resulting in extremely large prediction errors in densely built-up urban areas.

[0012] Fourth, there is a lack of online coupling capability between flight parameter sensing and noise prediction. In existing technologies, the acquisition of flight parameters and the calculation of noise propagation are two separate processes, failing to achieve organic online coupling. Flight parameters are only used to provide the spatial location of the sound source (i.e., flight trajectory), but fail to drive the sound source intensity model to dynamically adjust according to real-time operating conditions, nor can attitude information be used for real-time correction of sound radiation directivity, thus failing to leverage the proper role of flight parameters in noise prediction.

[0013] Therefore, how to provide a method and system that can extract the noise source intensity in real time during flight and improve the accuracy of ground noise impact prediction is a problem that urgently needs to be solved by those skilled in the art. Summary of the Invention

[0014] In view of the above problems, this invention is proposed to provide a method and system for predicting UAV noise based on flight parameter perception, which overcomes or at least partially solves the above problems. By deploying a microphone array on the UAV to collect near-field noise signals in real time and coupling it online with a flight parameter perception module, the invention achieves real-time extraction of noise source intensity and high-precision prediction of ground noise impact during flight, providing technical support for noise control and flight path optimization of urban low-altitude UAVs.

[0015] To achieve the above objectives, the present invention adopts the following technical solution: In a first aspect, embodiments of the present invention provide a method for predicting UAV noise based on flight parameter perception, comprising the following steps: S1: Acquire flight status data and raw sound pressure signals synchronously collected by the airborne microphone array during the flight of the UAV. The flight status data includes at least the real-time rotation speed and spatial position of each rotor. S2: Preprocess the original sound pressure signal, including at least filtering out wind noise components and compensating for pseudo-sound components introduced by body vibration; S3: Calculate the blade passing frequency based on the real-time rotation speed of each rotor, extract the rotor noise harmonic component with the blade passing frequency as the fundamental frequency from the spectrum of the preprocessed original sound pressure signal, and perform background noise attenuation on the spectrum after deducting the rotor noise harmonic component to obtain the self-noise spectrum of the UAV under the current flight conditions. S4: Calculate the sound source power level of the UAV based on the self-noise spectrum; S5: Based on the spatial location and the sound power level of the sound source, and combined with the preset propagation attenuation model, calculate the predicted sound pressure level of the UAV at the target receiving point.

[0016] Preferably, in step S2, the step of compensating for the pseudo-sound component introduced by body vibration includes: Acquire triaxial vibration acceleration signals synchronously collected by accelerometers installed near each microphone; The synthesized signal of the triaxial vibration acceleration signal is used as the reference input of the adaptive filter, and the original sound pressure signal collected by the microphone is used as the desired signal. The pseudo-sound component related to the synthesized signal is subtracted from the original sound pressure signal by the adaptive filtering algorithm, and the adaptive filter outputs the vibration-compensated sound pressure signal.

[0017] Preferably, the background noise in step S3 includes ambient background noise, and its extraction steps include: For the remaining spectrum after deducting rotor noise harmonic components, the amplitude square coherence function between different microphones is calculated based on the microphone self-power spectrum and interaction rate spectrum. The frequency band energy with a coherence function value greater than a preset coherence threshold is determined as the broadband component of the UAV's self-noise and retained, while the frequency band energy with a coherence function value less than or equal to the preset coherence threshold is determined as environmental background noise.

[0018] Preferably, the background noise in step S3 includes far-field background noise, and its extraction steps include: The beam output power of each frequency component of the preprocessed original sound pressure signal in each direction is calculated based on the beamforming spatial positioning method. Based on the spatial location of the UAV, the direction with the highest beam output power within the preset spatial range of the UAV is determined as the direction of the sound source. When the angle between the direction of the sound source and the direction of the UAV's center of mass is less than a preset angle threshold, the frequency component is determined to be the UAV's self-noise and is retained; otherwise, it is determined to be environmental background noise.

[0019] Preferably, the step S4 of calculating the sound source power level of the UAV based on the self-noise spectrum includes: Calculate the sound pressure level at each microphone measuring point based on the self-noise spectrum; Based on the distance parameters from each microphone to the acoustic center of the UAV and the directivity correction function value of each microphone in the direction of sound radiation corresponding to its installation position, the sound pressure level at each microphone measuring point is converted into the sound power level corresponding to that microphone. The average of the sound power levels calculated from each microphone is used to obtain the total sound source power level spectrum of the UAV under the current flight conditions.

[0020] Preferably, the value of the directional correction function is determined as follows: When spatially locating each frequency component using beamforming, the beam output power is calculated in each direction of the scanning space to form a beam power pattern. After normalizing the beam power pattern, take the logarithm to obtain the directivity correction function values ​​of the UAV in each direction.

[0021] Preferably, the propagation attenuation model includes one or more of the following: geometric divergence attenuation term, atmospheric absorption attenuation term, ground effect attenuation term, and building obstruction attenuation term; in step S5, the method for calculating the predicted sound pressure level of the UAV at the target receiving point includes one or more of the following: The straight-line distance between the spatial location and the target receiving point is calculated to determine the geometric divergence attenuation term; The ambient temperature and relative humidity are obtained, and the atmospheric absorption coefficient of each frequency band is calculated according to the atmospheric sound absorption standard to determine the atmospheric absorption attenuation term. The ground factor is determined based on the ground type, and the ground effect attenuation term is calculated according to the outdoor sound propagation attenuation standard. Acquire the building geometry data between the UAV and the target receiving point, perform ray tracking, and if there is building obstruction, calculate the diffraction path difference. Based on the diffraction path difference, use the diffraction attenuation formula to determine the building obstruction attenuation term. Subtract one or more of the geometric divergence attenuation term, atmospheric absorption attenuation term, ground effect attenuation term, and building obstruction attenuation term from the sound source power level to obtain the predicted sound pressure level of each frequency band at the target receiving point.

[0022] Preferably, the calculation of the predicted sound pressure level of the UAV at the target receiving point in step S5 further includes: The predicted sound pressure levels of each frequency band are superimposed after applying A-weighting correction to synthesize the total A-weighted sound pressure level at the target receiving point.

[0023] Preferably, in step S5, the calculation step for the predicted sound pressure level further includes at least one of the following: The ground area below the UAV's flight path is divided into grids, and the A-weighted total sound pressure level is calculated for each grid point to generate a two-dimensional noise impact heat map. Calculate the equivalent continuous A-weighted sound level based on the time series of the A-weighted total sound pressure level within a preset time period: The predicted sound pressure level is compared with the preset standard limit, and the compliance assessment result and the value exceeding the limit are output. When the predicted sound pressure level exceeds the noise limit constraint, the system searches in reverse for flight status data that meets the noise limit constraint as a flight parameter adjustment scheme based on the mapping relationship between the current flight status data and the sound pressure level, and outputs route optimization suggestions.

[0024] In a second aspect, embodiments of the present invention provide a UAV noise prediction system based on flight parameter perception, used to implement the UAV noise prediction method based on flight parameter perception described in the first aspect, comprising: The flight parameter sensing module is used to acquire flight status data during the flight of the UAV, and the flight status data includes at least the real-time rotation speed and spatial position of each rotor; The airborne noise monitoring module includes a microphone array deployed on the UAV body for synchronously acquiring raw sound pressure signals; The sound source intensity calculation module is used to preprocess the original sound pressure signal to filter out wind noise components and compensate for pseudo-sound components introduced by body vibration. It calculates the blade passing frequency based on the real-time rotation speed of each rotor and extracts rotor noise harmonic components from the spectrum of the preprocessed sound pressure signal. It performs background noise separation on the spectrum after deducting the rotor noise harmonic components to obtain the self-noise spectrum of the UAV under the current flight conditions, and calculates the sound source power level of the UAV based on the self-noise spectrum. The noise propagation calculation module is used to calculate the predicted sound pressure level of the UAV at the target receiving point based on the spatial location and the sound power level of the sound source, combined with a preset propagation attenuation model. The noise impact prediction output module is used to output the predicted sound pressure level.

[0025] The specific beneficial effects include at least the following: This invention enables real-time prediction of noise impact during flight. By using onboard noise monitoring equipment to collect data in real time and online processing by a sound source intensity calculation module, this invention can predict the noise impact of the current flight conditions on various locations on the ground in real time, with a time delay of only seconds, based on actual noise measurement data. This completely changes the existing working mode of "fly first, then measure" or "offline simulation" or "pre-set database query," making real-time prediction of noise impact possible.

[0026] This invention significantly improves the accuracy of noise prediction under various flight conditions. It directly acquires actual near-field noise signals during flight using the UAV's built-in microphone array, and combines this with background noise separation and sound source identification aided by flight parameters. This accurately reflects the actual noise source intensity changes of the UAV under different flight conditions such as hovering, climbing, cruising, turning, and descending. Compared to methods that rely on interpolation queries from a pre-set database, this method is directly based on measured data, fundamentally avoiding the deviation caused by inconsistencies between database calibration conditions and actual flight conditions, and effectively capturing noise peaks during maneuvering flight.

[0027] This invention is adapted to noise propagation calculations in complex urban environments. By integrating urban 3D geographic information data and building occlusion and diffraction calculation functions, this invention can accurately assess the occlusion and reflection effects of buildings on noise propagation in urban built environments, avoiding the significant errors produced by simplified propagation models in densely built-up areas, and making the prediction results closer to reality.

[0028] This invention provides decision support for active noise reduction and flight path optimization. The real-time noise prediction results output by this invention can be directly used as constraints for UAV flight path planning and dynamic adjustment. Combined with the mapping relationship between flight parameters and noise, optimization suggestions are generated in reverse. This provides operators with a complete technical link from feedforward prediction to decision output for noise-aware flight path optimization and active noise reduction control. It helps to effectively control noise impact while ensuring operational efficiency, and promotes the coordinated development of low-altitude economy and urban living environment.

[0029] This invention features a highly scalable system suitable for various aircraft models and scenarios. The airborne noise monitoring equipment employs a modular design, allowing the microphone array's installation location and quantity to be flexibly adjusted according to the structural characteristics of different aircraft models. The flight parameter sensing module is compatible with mainstream UAV flight control system data interface standards (such as the MAVLink protocol), and the noise propagation calculation module is compatible with GIS data formats from different cities. The system exhibits excellent versatility and scalability. New aircraft models can be integrated into the system simply by adapting and installing the microphone array on the airframe, eliminating the need for expensive acoustic wind tunnel calibration. Attached Figure Description

[0030] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.

[0031] Figure 1 A flowchart illustrating the UAV noise prediction method based on flight parameter perception provided in an embodiment of the present invention; Figure 2 This is a flowchart of the sound source power level calculation provided in an embodiment of the present invention; Figure 3 This is a diagram illustrating the architecture of a UAV noise prediction system based on flight parameter perception, provided in an embodiment of the present invention. Figure 4 An example diagram of noise impact prediction output provided for an embodiment of the present invention.

[0032] Figure 5 This is a schematic diagram of the propagation path between the sound source (drone) and the ground receiving point, provided in an embodiment of the present invention. Detailed Implementation

[0033] 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, and 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.

[0034] This invention specifically discloses a method for predicting UAV noise based on flight parameter perception, such as... Figure 1 As shown, it includes the following steps: S1: Acquire flight status data and raw sound pressure signals synchronously collected by the airborne microphone array during the flight of the UAV. The flight status data includes at least the real-time rotation speed and spatial position of each rotor.

[0035] In one embodiment, the following flight parameters are acquired in real time through an onboard sensor system and flight control data bus deployed on the drone: Flight attitude parameters: including pitch angle ( ), roll angle ( ), yaw angle ( The sampling frequency is not less than 10Hz, provided by the inertial measurement unit (IMU); Rotor speed parameters: including the real-time speed of each rotor motor (or engine) ( i =1,2,...,N, where N is the number of rotors), obtained by electronic speed controller (ESC) or speed sensor, with a sampling frequency of not less than 20Hz; Flight speed parameters: including ground speed V g airspeed V a and three-dimensional velocity components (V) x V y V z (), provided jointly by GNSS module and pitot tube; Spatial location parameters include longitude, latitude, and altitude H (or altitude above ground AGL), provided jointly by a GNSS / RTK module and a barometric altimeter, with a positioning accuracy better than 1 meter; Airborne noise monitoring data includes raw near-field sound pressure signals collected in real time by a microphone array deployed on the UAV. The microphone array consists of no fewer than four microphones, installed on the UAV's arms, fuselage underside, and tail, with a sampling frequency of at least 20kHz, covering a frequency range of 20Hz to 10kHz. Each microphone synchronously collects the time-domain sound pressure signal, which is then aggregated by the airborne data acquisition unit and transmitted synchronously with other flight parameters. Environmental parameters: including ambient temperature (T), relative humidity (RH), and atmospheric pressure. P atm The data is provided by airborne environmental sensors and is used for the accurate calculation of the atmospheric sound absorption coefficient.

[0036] The above parameters and noise monitoring data are aggregated by the airborne data acquisition unit and then transmitted in real time to the ground computing platform (or processed locally by the airborne edge computing unit) via the airborne communication link (such as 4G / 5G mobile communication or data link) as input for the calculation of sound source intensity, i.e. sound source power level.

[0037] S2: Preprocess the original sound pressure signal, including at least filtering out wind noise components and compensating for pseudo-sound components introduced by body vibration.

[0038] In one embodiment, after each microphone signal is passed through an airborne preamplifier and an anti-aliasing filter, it is converted and packaged by a multi-channel synchronous data acquisition unit at a sampling rate of not less than 20kHz, and output synchronously with other flight parameter data frames. The output signal participates in the preprocessing operation in step S2.

[0039] In one embodiment, the raw sound pressure signal acquired by the airborne microphone ( i (where M is the number of microphones, and the input is 1, 2, ..., M) The following preprocessing steps are required before entering the sound source intensity extraction stage: Wind noise filtering steps: Due to the impact of incoming airflow on the microphone during drone flight, the collected signal contains low-frequency wind noise artifacts. A high-pass filter with a cutoff frequency of 50Hz is used to filter out the low-frequency components dominated by wind noise.

[0040] The steps for compensating for pseudo-sound components introduced by airframe vibration: Vibrations of the UAV's airframe structure are coupled to the acoustic measurement channel via the microphone mounting bracket, introducing pseudo-sound components related to vibration acceleration into the acquired signal. This invention employs a vibration compensation method based on Normalized Least Mean Square (NLMS) adaptive filtering, with the specific steps as follows: S21: Reference signal acquisition: Using miniature accelerometers (triaxial MEMS type, mounted on the body side of the microphone vibration isolation bracket) installed near each microphone, triaxial vibration acceleration signals are acquired synchronously. , , The sampling rate is consistent with that of the microphone (not less than 20kHz).

[0041] S22: NLMS Adaptive Filter Construction: Synthesizing Signals from Triaxial Accelerometer Signals As the reference input of the adaptive filter The raw sound pressure signal collected by the microphone As expected signal The adaptive filter output is an estimate of the vibration pseudo-sound. ,in for The weight coefficient vector of the first-order FIR filter ( Take 64~128). The reference input vector is [reference input vector]. The error signal is [error signal]. That is, the sound pressure signal after vibration compensation.

[0042] S23: Iterative Update of Weight Coefficients: The filter weight coefficients are updated online using the NLMS algorithm. in, This is the step size factor (ranging from 0.01 to 0.1). Small positive number (take) To prevent the denominator from being zero, the weight coefficients are updated iteratively for each frame of data (1024 points), with a convergence time of approximately 0.5 to 1 second.

[0043] Verify the effect of compensation: Online verification in flight state - compare the output of the same microphone with vibration compensation on / off, record the microphone output in pure vibration environment, the residual signal after NLMS compensation should be lower than the microphone's noise floor (<30 dBA), verifying that vibration artifacts have been fully eliminated.

[0044] After pseudosound component compensation is completed, the preprocessing operation also includes a spectral transformation step: converting the preprocessed time-domain signal into a time-spectrum using a short-time Fourier transform (STFT). .in, For frequency, This is the center moment of the time window. The STFT parameters are: window length 1024 points (approximately 51.2 ms), overlap rate 50%, and Hanning window function. Time-spectrum signal. Participate in step S3 calculation.

[0045] S3: Calculate the blade passing frequency based on the real-time rotation speed of each rotor, extract the rotor noise harmonic components with the blade passing frequency as the fundamental frequency from the spectrum of the preprocessed original sound pressure signal, and perform background noise attenuation on the spectrum after deducting the rotor noise harmonic components to obtain the self-noise spectrum of the UAV under the current flight conditions.

[0046] In one embodiment, step S3, the step of subtracting rotor noise harmonic components, includes: S31: The spectral characteristics of rotor noise are based on the blade passing frequency (BPF= ,in B A discrete harmonic series with the number of blades as the fundamental frequency. The real-time rotor speed is obtained. The positions of the BPF and its harmonic frequencies are precisely marked on the spectrum. Narrowband energy located within ±5% of these frequencies is extracted as rotor noise components, and these rotor noise components are then subtracted.

[0047] In one embodiment, after deducting the harmonic components of rotor noise, the broadband components in the remaining spectrum include fuselage aerodynamic noise and ambient background noise. Specifically, due to the presence of ambient background noise (such as traffic noise, wind noise, and noise from other aircraft) in the UAV's flight environment, the time-spectrum signal collected by the airborne microphone... The noise includes not only the radiated noise of the drone itself, but also the aforementioned environmental background noise components. Separation is achieved by utilizing the spatial coherence differences between multiple microphones installed at different locations on the drone—the drone's self-noise exhibits high coherence between different microphones (the microphone spacing is much smaller than the sound wave wavelength), while the far-field environmental background noise has lower coherence between different microphones. Step S3 also includes step S32, which separates the noise by calculating the Magnitude Squared Coherence (MSC) function of each microphone pair, as detailed below: S321: Cross-power spectrum and self-power spectrum calculation: For microphone pairs Cross-power spectrum calculated using Welch average periodogram method and their respective self-power spectra , Take 20 consecutive STFT data segments (1024 points per frame), with a frequency resolution of approximately 19.5 Hz.

[0048] S322: Coherence Function Calculation: The squared amplitude coherence function is defined as: in, The range of values ​​is The closer the value is to 1, the stronger the signal coherence between the two microphones at that frequency.

[0049] S322: Coherence Threshold Determination: Set a coherence threshold. For each frequency band If at least one pair of microphones has a coherent function value Then the energy in that frequency band is determined to belong to the self-noise of the UAV (the near-field sound source is highly correlated at different microphones); if all microphone pairs If the frequency band energy is determined to be ambient background noise (low coherence of far-field sound sources), then it is determined to be ambient background noise.

[0050] S323: For frequency components identified as ambient background noise, post-filtering based on beamforming is used for attenuation.

[0051] In one embodiment, step S3 further includes using beamforming technology of the microphone array to spatially locate each frequency component, distinguish between near-field self noise and far-field background noise, and perform directional attenuation of the background noise component. The specific steps are as follows: S331: Delayed Summation Beamforming: For microphone arrays A microphone, in a spherical coordinate system, with azimuth angle and elevation angle For each scan direction, the scan variable is defined. Calculate the beam output power of the array: in, For the first i A microphone at frequency f , No. t Complex sound pressure at the center of each STFT time window (obtained via STFT). For the first Each microphone is positioned relative to a reference microphone (at the geometric center of the array) in the direction of... The theoretical arrival time difference is determined by calculations based on the microphone installation coordinates and the speed of sound. ,in For the first The position vector of each microphone It is a unit vector of direction. Let be the speed of sound. The above formula is calculated within each STFT time window. This indicates the beam output power for the current time window.

[0052] S332: Automatic near / far field criterion: in each direction of the scan space Simultaneously, the distance estimate of the sound source in that direction is calculated. Utilizing the focusing characteristics of near-field beamforming—when the scanning direction points to the vicinity of the UAV's body position (known three-dimensional spatial coordinates) (a sphere with a radius of 1 meter centered on the UAV's center of mass), the direction in which the beam output power reaches its maximum value is determined as the direction of near-field self-noise; other directions are determined as the direction of far-field background noise. The automatic criterion is: in, The direction with the maximum beam output power. The direction vector pointing from the center of the array to the center of mass of the UAV. The angle threshold is set to 15°. It represents the angle between two direction vectors.

[0053] S333: Background Noise Directional Attenuation: For frequency components identified as background noise, a beamforming-based post-filter technique is used for attenuation. The Wiener filter gain function is constructed using the beam output from the noise direction as the target signal estimate. in, The beam output power is the beam output power from the direction of near-field self-noise. This represents the average beam output power in the direction of the far-field background noise. Applying this gain function to the frequency domain signal of each microphone yields the purified signal after background noise attenuation: .

[0054] S334: Correlation with Directivity Correction: The beamforming analysis described above, while locating the direction of the sound source, outputs a beam power pattern. This directly reflects the directivity characteristics of the UAV's sound radiation in different directions. Taking the logarithm after normalizing this radiation pattern yields the directivity correction function required by the sound source intensity calculation module. This is used in step S4 to convert near-field sound pressure to sound source intensity and in step S5 to correct the directivity of the far-field receiving point.

[0055] S335: Integrated Output: Combines the microphone signals attenuated by background noise within each time window. The purified drone self-noise spectrum is obtained by averaging multiple channels. .

[0056] like Figure 2 As shown, by combining the above three steps S31 to S33, the self-noise spectrum of the UAV under the current flight conditions is extracted from the airborne microphone signal. .

[0057] S4: Calculate the sound power level of the UAV's sound source based on the self-noise spectrum.

[0058] In one embodiment, step S4, which calculates the sound source power level of the UAV based on the self-noise spectrum, includes: The extracted near-field spectrum of the drone's self-noise (Unit: Pa / √Hz) Convert to sound source power level spectrum. First, calculate the sound pressure level at each microphone measuring point: in, Pa is the reference sound pressure level (human hearing threshold). Let be the amplitude of the self-noise spectrum. Then, the... i Sound pressure level at each microphone measuring point Sound power level of drone sound source The relationship between them is: in, For the first i The distance from the microphone to the drone acoustics center

[0059] Directivity correction is applied to the microphone's orientation (determined by both the microphone's installation location and the UAV's acoustic radiation directionality). The average of conversion results from multiple microphones is used to improve the robustness of the sound source intensity estimation. The result obtained after the above conversion This is the total sound source power level spectrum of the UAV under the current flight conditions, which is directly output to the noise propagation calculation module. The sound source is simplified in space as a single point sound source located at the center of mass of the UAV (far-field approximation).

[0060] S5: Based on the spatial location and sound source power level, and combined with the preset propagation attenuation model, calculate the predicted sound pressure level of the UAV at the target receiving point.

[0061] In one embodiment, the method is extended and adapted based on the standard method of "Outdoor Sound Propagation Attenuation Calculation".

[0062] For those located in spatial position The sound source, at the ground receiving point First Sound pressure level in each frequency band Calculate using the following formula: The meanings of each symbol are as follows: : Receiving point Each frequency band (center frequency) The sound pressure level (unit: dB) of 1000 saturated bulbs. : Time drone in The total sound power level of the sound source in each frequency band (unit: dB) is output by the sound source intensity calculation module; Directivity index of the sound source in the direction of the receiving point (unit: dB), obtained online by the multi-microphone beamforming analysis in the sound source intensity calculation module; Geometric divergence attenuation (unit: dB); Atmospheric absorption attenuation (unit: dB); Ground effect attenuation (unit: dB); Barrier / building shading attenuation (unit: dB).

[0063] The physical meaning and calculation method of each attenuation term are as follows: Directional correction item To consider the directional characteristics of sound source radiation, for multi-rotor UAVs, the directivity of sound radiation is mainly determined by the normal direction of the rotor disk plane and the fuselage attitude. The directivity index of the sound source in the direction of the receiving point is obtained online through multi-microphone beamforming analysis in the sound source intensity calculation module and determined using the azimuth and elevation angles in the sound source coordinate system as indices.

[0064] Geometric divergence decay For the propagation of a point sound source in a semi-free field (ground half-space), Calculate using the following formula: in, The straight-line distance (in meters) from the sound source to the receiver is calculated from the real-time location coordinates: in, The three-dimensional spatial coordinates of the sound source (drone) These are the three-dimensional spatial coordinates of the receiving point, all in meters.

[0065] Atmospheric absorption attenuation Atmosphere's absorption coefficient of sound wave energy With respect to ambient temperature T, relative humidity RH and atmospheric pressure Relatedly, the attenuation coefficients for each frequency band are calculated according to the ISO 9613-1 standard: in, For the first Center frequency of each frequency band The atmospheric sound absorption coefficient (unit: dB / km) at a given location is determined by the ambient temperature. (Unit: °C), Relative Humidity (Unit: %) and atmospheric pressure (Unit: kPa) Calculated according to ISO 9613-1 standard. When airborne environmental sensors are available, , , Use real-time measurements; otherwise, use the standard reference values ​​from ground meteorological stations.

[0066] Ground effect attenuation Consider the impact of ground acoustic impedance on sound propagation. For UAV flight scenarios, the ground type is typically hard ground (concrete, asphalt pavement) or mixed ground. Ground attenuation. It is determined according to the general calculation method specified in ISO 9613-2, and is related to the ground factor G (0~1, 0 for hard ground, 1 for soft ground), the height of the sound source, the height of the receiving point, and the frequency.

[0067] Barrier / Building shading attenuation In urban low-altitude operation scenarios, the blocking effect of buildings on sound propagation is crucial. An integrated sub-model for calculating building blocking attenuation is provided. Building geometry model: Obtain 3D geometric data (including building outline, height, and roof shape) of buildings below and surrounding the flight path from a city 3D Geographic Information System (3D GIS) or Building Information Modeling (BIM). Figure 5 As shown; Sound ray tracing: Sound ray tracing is performed between the sound source and the receiver to determine if there are buildings obstructing the sound. If the sound ray is blocked by a building, the diffraction path difference is calculated. (That is, the difference between the length of the diffraction path and the length of the direct path). Single-screen diffraction attenuation: For a single occlusion, Calculate using the Maekawa formula or its improved form: in, For the first Wavelength of sound waves in each frequency band (unit: meters). It is the speed of sound (approximately 340 m / s). The diffraction path difference (the difference between the diffraction path length and the direct path length, in meters). This is an empirical constant (usually taken as 20). For multiple occlusions, the diffraction attenuation is accumulated based on the contributions of each barrier, but does not exceed the upper limit (usually 20~25 dB).

[0068] In one embodiment, step S5 calculates the predicted sound pressure level of the UAV at the target receiving point and generates outputs tailored to different application requirements, including single-point noise prediction values: For a user-specified ground point of interest (such as the location of noise-sensitive buildings like residential buildings, schools, or hospitals), output the A-weighted sound pressure level at that point at the current moment. (Sound pressure levels of each frequency band are weighted and synthesized using an A-weighting network): A-weighted networks simulate the human ear's varying sensitivity to different sound frequencies (with greater attenuation at low and high frequencies, and sensitivity to mid-frequency ranges of 1–4 kHz). The center frequencies of each 1 / 3 octave band are... Corresponding A-weighted adjustment value Calculate using the following formula: After applying A-weighting to the sound pressure levels of each frequency band, the energy is superimposed to synthesize the A-weighted total sound pressure level: in, The total number of frequency bands (24 bands in total, ranging from 50 Hz to 10 kHz in 1 / 3 octave bands). The output of the noise propagation calculation step Each frequency band sound pressure level.

[0069] In addition to outputting the sound pressure level spectrum for each frequency band, it can also output the maximum instantaneous sound pressure level. Specifically, Defined as the maximum A-weighted instantaneous sound pressure level within the observation period, it is calculated using the following formula: in, The instantaneous sound pressure level (in Pa) is obtained by A-weighting filtering of the sound pressure levels in each frequency band output by the noise propagation calculation module. Pa. The sampling interval is taken as the microphone's original sampling period (0.05 ms) to capture noise peak events.

[0070] In one embodiment, step S5, the calculation step for predicting the sound pressure level, further includes at least one of the following: Noise impact heatmap: The ground area below the flight path is divided into grids (grid resolution is configurable, typically 2m×2m to 10m×10m), and the noise value is calculated for each grid point. A two-dimensional noise impact heatmap is generated, which visually displays the spatial distribution characteristics of noise using color mapping. The heatmap is updated in real time as the drone flies.

[0071] Time-integral noise index: Calculates the equivalent continuous A-weighted sound level by integrating the noise exposure over a specified time period. : in, for A-weighted sound pressure level at any given time (unit: dBA). The length of the integration time period (unit: seconds). Indicates the time period The stable sound level equivalent to the energy of internal fluctuation noise.

[0072] Based on continuous A-weighted sound level Statistical sound level indicators (such as) , , wait).

[0073] Compliance assessment results: Predicted noise values The system compares the limits with relevant standards (such as the "Environmental Noise Quality Standard" and the "Airport Surrounding Aircraft Noise Environmental Standard"), outputs compliance / non-compliance labels and exceedance values, and provides a basis for operational decisions.

[0074] Optimization suggestion output: Predicted noise value When the preset threshold is exceeded, the system can combine the mapping relationship between flight parameters and sound source intensity to search in reverse for flight parameter adjustment schemes that meet the noise limit constraints (such as adjusting flight altitude, reducing flight speed, adjusting rotor speed or yaw path), generate optimization suggestions and output them to the UAV flight control system or operation management platform.

[0075] Based on the same inventive concept, combined with Figure 3 As shown, this embodiment of the invention also provides a UAV noise prediction system based on a flight parameter perception-based UAV noise prediction method. The technical features of this system are also applicable to the execution steps of the flight parameter perception-based UAV noise prediction method. The system includes: The flight parameter sensing module is used to acquire flight status data during the flight of the UAV. The flight status data includes at least the real-time rotation speed and spatial position of each rotor. The airborne noise monitoring module includes a microphone array deployed on the UAV body for synchronously acquiring raw sound pressure signals; The sound source intensity calculation module is used to preprocess the original sound pressure signal to filter out wind noise components and compensate for pseudo-sound components introduced by body vibration. It calculates the blade passing frequency based on the real-time rotation speed of each rotor and extracts rotor noise harmonic components from the spectrum of the preprocessed sound pressure signal. It performs background noise separation on the spectrum after deducting rotor noise harmonic components to obtain the self-noise spectrum of the UAV under the current flight conditions, and calculates the sound source power level of the UAV based on the self-noise spectrum. The noise propagation calculation module is used to calculate the predicted sound pressure level of the UAV at the target receiving point based on the spatial location and the sound power level of the sound source, combined with a preset propagation attenuation model. The noise impact prediction output module is used to output the predicted sound pressure level.

[0076] In one embodiment, the airborne noise monitoring module comprises a microphone array deployed on the UAV fuselage. The microphone array includes at least four microphones, installed in the following locations: the front end of the arm (near the underside of each rotor to capture near-field rotor noise), the center of the fuselage belly (to capture overall noise from below the fuselage), and the tail position (to capture tail aerodynamic noise). Each microphone should have a frequency response range of 20Hz to 10kHz and a dynamic range of at least 94dB, with a sampling frequency of at least 20kHz.

[0077] The microphone is mounted on a vibration-damping bracket to reduce the interference of the machine structure vibration on the acoustic measurement, and a wind shield is installed at the front end of the microphone to reduce false sounds caused by airflow.

[0078] The specific implementation of the present invention will be described in detail below through specific examples.

[0079] Example 1: Real-time prediction of flight path noise for urban logistics drones.

[0080] System Deployment: This embodiment uses a quadcopter logistics drone (maximum takeoff weight 25kg, rotor diameter 0.7m, cruising speed 15m / s) in an urban environment as an application scenario.

[0081] The system deployment architecture is an airborne perception combined with ground computing mode: Onboard: An additional sensor system is installed on the UAV platform or existing sensor systems are utilized, including an IMU (for acquiring pitch, roll, and yaw angles), an electronic speed controller speed feedback interface (for acquiring the real-time speed of the four motors), a GNSS / RTK module (for acquiring three-dimensional position and ground speed), an airspeed tube (for acquiring airspeed), a temperature and humidity sensor, and an onboard noise monitoring array consisting of six microphones (four mounted under the rotor at the front of each arm, one mounted in the center of the fuselage belly, and one mounted in the tail). The onboard data acquisition unit summarizes flight parameters at a frequency of 20Hz and simultaneously acquires six channels of noise signals at a sampling rate of 20kHz. After being timestamped, the two types of data are transmitted in real time to the ground computing platform via a 4G / 5G communication module.

[0082] Ground-based: The ground computing platform deploys modules for sound source intensity calculation, noise propagation calculation, and noise impact prediction output. The platform connects to a city-wide 3D GIS database (containing the outline, height, and location data of buildings along the flight path) and a meteorological data interface (to obtain ground temperature, humidity, and air pressure), as well as a UAV model parameter library (containing the coordinates of the microphone installation location). and the directivity correction function of each microphone ).

[0083] Airborne microphone array calibration and parameter configuration: Before the system is put into operation, the airborne microphone array deployed on the UAV needs to be calibrated with installation parameters and transfer function once. (1) Microphone installation position calibration: Use three-dimensional coordinate measuring equipment (such as laser tracker or photogrammetry system) to accurately determine the three-dimensional coordinates of the acoustic center of each microphone in the UAV body coordinate system. And calculate the distance from each microphone to the acoustic center (geometric center or centroid) of the UAV. .

[0084] (2) Microphone sensitivity calibration: In an anechoic chamber or semi-anechoic chamber environment, use a standard sound source (such as a B&K 4231 sound calibrator, 94dB@1kHz) to calibrate the sensitivity of each microphone and record the sensitivity correction value of each microphone. .

[0085] (3) Determination of Directivity Correction Function: In an anechoic chamber, the UAV was placed on a turntable, and the radiation directivity of the UAV in each direction was measured using a far-field standard microphone array (a hemispherical array with a radius of not less than 3m). Simultaneously, the synchronous measurement data of the airborne microphone and the far-field microphone were recorded, and the directivity correction function of each airborne microphone was established. Lookup table (with 1 / 3 octave frequency and 5° azimuth / elevation angle as resolution).

[0086] (4) Parameter storage: Store the above calibration parameters (microphone coordinates) Sensitivity correction directional correction The parameters are stored in the UAV model parameter library as fixed configuration parameters for the sound source intensity calculation module. The above calibration is a one-time operation and does not need to be repeated before each flight.

[0087] Real-time prediction process: The real-time prediction process during system operation is as follows: Step S1: The flight parameter sensing module collects the current pitch angle of the UAV at a frequency of 20Hz. Roll angle Yaw angle The four rotor speeds ~ 3D position coordinates (X s ,Y s H), three-dimensional velocity components (V) x V y V z ), ambient temperature T and relative humidity RH.

[0088] Step S2: After receiving the 6-channel noise signal and synchronous flight parameters transmitted from the airborne source, the sound source intensity calculation module performs the following real-time processing flow: (a) Time-domain signal of 6-channel microphone Perform wind noise high-pass filtering (50Hz cutoff) and LMS adaptive vibration compensation based on accelerometer signals.

[0089] (b) Perform a short-time Fourier transform with a 1024-point Hanning window and a 50% overlap rate to obtain the 6-channel time spectrum. .

[0090] (c) Utilizing real-time rotor speed Calculate the blade's passing frequency BPF and its harmonics. Identify and extract rotor noise harmonic components from the spectrum.

[0091] (d) Calculate the spatial coherence function for each microphone pair. The broadband component of the UAV's self-noise and the ambient background noise were separated using a coherence threshold of 0.6.

[0092] (e) Perform beamforming localization analysis on each frequency component, retain the acoustic energy incident on the UAV body and attenuate the acoustic energy incident on distant objects, to obtain the purified UAV self-noise spectrum. .

[0093] (f) Using pre-stored microphone distance parameters and directional correction function According to the formula:

[0094] Converting multi-channel near-field sound pressure levels into the spectrum of sound source sound power levels .

[0095] Step S3: The noise propagation calculation module calculates the sound propagation process grid by grid based on the UAV's current 3D position coordinates and the preset ground receiving point grid (or user-specified points of interest). For each receiving point: (a) Calculate the straight-line distance from the sound source to the receiver. d Determine the geometric divergence attenuation .

[0096] (b) Calculate the atmospheric absorption coefficient for each frequency band based on ambient temperature T, relative humidity RH, and atmospheric pressure, in accordance with ISO 9613-1 standard. ,get .

[0097] (c) Determine the ground factor G and calculate the ground effect attenuation according to ISO 9613-2. .

[0098] (d) Load building data between the sound source and receiver from the 3D GIS database and perform ray tracing. If the sound ray intersects with a building, calculate the diffraction path difference. The building shading attenuation was calculated using the diffraction formula. .

[0099] (e) Determine the directivity index based on the azimuth and elevation angles of the receiving point relative to the UAV, using the directivity information obtained online by the sound source intensity calculation module. .

[0100] (f) According to the formula:

[0101] Calculate the sound pressure level for each frequency band.

[0102] (g) Apply A-weighting to the sound pressure levels of each frequency band and sum them to obtain the A-weighted sound pressure level at the receiving point. .

[0103] Step S4: The noise impact prediction output module comprehensively processes and presents the calculation results: (a) All grid points The values ​​are rendered as a two-dimensional noise heatmap using color mapping and displayed overlaid on the geographic base map.

[0104] (b) Marking noise-sensitive areas (such as residential areas, schools) The value is compared with the noise limit for the corresponding functional area in the relevant standard, and the compliance status is indicated by color (green / yellow / red).

[0105] (c) Records Time series, calculated at 1-second intervals within the runtime segment This is used to submit a noise compliance report after the task is completed.

[0106] (d) When a certain noise-sensitive point When the predicted value exceeds the limit, the system automatically triggers an optimization search: searching the flight parameter space for the parameter combination that satisfies the noise limit and minimizes the flight performance cost (e.g., increasing the flight altitude to...). (Lower to the minimum altitude below the limit) to generate route adjustment suggestions.

[0107] like Figure 4 As shown, Figure 4 The interface should display a comprehensive view of the noise impact prediction output, with the core being a two-dimensional noise heatmap overlaid on a geographic base map, where the grid point colors correspond to their respective locations. Values; noise-sensitive points (residential areas, schools, etc.) and their values ​​should be marked on the map. Values ​​are displayed, with green / yellow / red indicating compliance status; receiving points can be used as supplementary indicators. Time series and statistical results In scenarios exceeding limits, optimization suggestions should be provided (such as increasing flight altitude or adjusting flight path).

[0108] To verify the technical effect of the present invention, a comparative test was conducted on a low-altitude logistics test route in a city (approximately 3 km long, with residential buildings, schools and other noise-sensitive structures along the route).

[0109] Comparison with Option 1 (existing simplified method): The ground noise value is estimated using a simple geometric attenuation formula (considering only distance attenuation and ignoring atmospheric absorption, ground effect and building obstruction) based solely on the flight altitude and the aircraft's nominal noise value.

[0110] Comparative Solution 2 (Method of the Invention): The complete technical solution proposed in this invention is adopted to sense flight parameters in real time, calculate noise source intensity online, and include all attenuation terms.

[0111] Simultaneous field measurements were conducted at eight ground-level noise monitoring points (covering both open and building-shaded areas), and the results showed that: In the open region, the root mean square error (RMSE) between the predicted and measured values ​​of the method of the present invention is 1.8 dBA, while the RMSE of the existing simplified method is 4.5 dBA. In areas obstructed by buildings, the RMSE of the predicted and measured values ​​using the method of this invention is 2.3 dBA, while the RMSE of the existing simplified method is as high as 8.1 dBA (the simplified method cannot reflect the building obstruction effect and systematically overestimates the noise value). When the UAV is performing rapid pitch maneuvers (such as the rapid climb phase), the method of this invention can accurately capture the transient increase in noise source intensity (about 3~5 dBA), while the simplified method, because it does not perceive attitude changes, significantly underestimates this phase.

[0112] Example 2: Airborne edge computing deployment mode.

[0113] This embodiment provides a deployment implementation method for the system on an airborne edge computing platform of an unmanned aerial vehicle (UAV), which is suitable for scenarios where the communication link is unreliable or where real-time requirements are extremely high.

[0114] The main difference from Embodiment 1 is that the sound source intensity calculation module, noise propagation calculation module, and noise impact prediction output module are all deployed on an airborne edge computing unit (such as NVIDIA Jetson Orin or an embedded AI computing platform with equivalent performance). The UAV model parameter library (containing microphone calibration parameters) and the urban 3D GIS database (or a subset thereof) are pre-stored in the airborne storage. The airborne microphone array and flight parameter sensing module communicate directly with the computing module via the airborne bus, without needing to transmit data to the ground.

[0115] In this deployment mode, noise prediction results can be utilized in the following ways: (1) Data is transmitted via airborne data link to ground monitoring stations for display and recording; (2) It is directly fed back to the airborne flight control system as a constraint input for real-time route optimization; (3) In the event of a communication interruption, the airborne terminal can autonomously make noise compliance judgments and route adjustment decisions.

[0116] Example 3: Multi-machine cooperative noise prediction.

[0117] When multiple UAVs operate simultaneously in the airspace, the system can be expanded to support multi-UAV cooperative noise prediction. For any ground receiving point, each UAV is treated as an independent sound source. After calculating the propagation attenuation of each sound source to the receiving point, the total noise value is synthesized by incoherent superposition (the phases of each sound source are random and incoherent): in, The number of drones operating simultaneously. For the first The A-weighted sound pressure level (dBA) generated by the drone at the receiving point. for The total A-weighted sound pressure level at that receiving point at any given time. Multi-machine collaborative prediction results can be used for airspace capacity assessment and collaborative noise reduction scheduling.

[0118] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on its differences from other embodiments. Similar or identical parts between embodiments can be referred to interchangeably. For the apparatus disclosed in the embodiments, since they correspond to the methods disclosed in the embodiments, the description is relatively simple; relevant parts can be referred to the method section.

[0119] The above description of the disclosed embodiments enables those skilled in the art to make or use the invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the invention. Therefore, the invention is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for predicting UAV noise based on flight parameter perception, characterized in that, Includes the following steps: S1: Acquire flight status data and raw sound pressure signals synchronously collected by the airborne microphone array during the flight of the UAV. The flight status data includes at least the real-time rotation speed and spatial position of each rotor. S2: Preprocess the original sound pressure signal, including at least filtering out wind noise components and compensating for pseudo-sound components introduced by body vibration; S3: Calculate the blade passing frequency based on the real-time rotation speed of each rotor, extract the rotor noise harmonic component with the blade passing frequency as the fundamental frequency from the spectrum of the preprocessed original sound pressure signal, and perform background noise attenuation on the spectrum after deducting the rotor noise harmonic component to obtain the self-noise spectrum of the UAV under the current flight conditions. S4: Calculate the sound source power level of the UAV based on the self-noise spectrum; S5: Based on the spatial location and the sound power level of the sound source, and combined with the preset propagation attenuation model, calculate the predicted sound pressure level of the UAV at the target receiving point.

2. The UAV noise prediction method based on flight parameter perception as described in claim 1, characterized in that, In step S2, the step of compensating for the pseudo-sound component introduced by the vibration of the machine body includes: Acquire triaxial vibration acceleration signals synchronously collected by accelerometers installed near each microphone; The synthesized signal of the triaxial vibration acceleration signal is used as the reference input of the adaptive filter, and the original sound pressure signal collected by the microphone is used as the desired signal. The pseudo-sound component related to the synthesized signal is subtracted from the original sound pressure signal by the adaptive filtering algorithm, and the adaptive filter outputs the vibration-compensated sound pressure signal.

3. The UAV noise prediction method based on flight parameter perception as described in claim 1, characterized in that, The background noise mentioned in step S3 includes ambient background noise, and its extraction steps include: For the remaining spectrum after deducting rotor noise harmonic components, the amplitude square coherence function between different microphones is calculated based on the microphone self-power spectrum and interaction rate spectrum. The frequency band energy with a coherence function value greater than a preset coherence threshold is determined as the broadband component of the UAV's self-noise and retained, while the frequency band energy with a coherence function value less than or equal to the preset coherence threshold is determined as environmental background noise.

4. The UAV noise prediction method based on flight parameter perception as described in claim 1, characterized in that, The background noise mentioned in step S3 includes far-field background noise, and its extraction steps include: The beam output power of each frequency component of the preprocessed original sound pressure signal in each direction is calculated based on the beamforming spatial positioning method. Based on the spatial location of the UAV, the direction with the highest beam output power within the preset spatial range of the UAV is determined as the direction of the sound source. When the angle between the direction of the sound source and the direction of the UAV's center of mass is less than a preset angle threshold, the frequency component is determined to be the UAV's self-noise and is retained; otherwise, it is determined to be environmental background noise.

5. The UAV noise prediction method based on flight parameter perception as described in claim 1, characterized in that, Step S4, which involves calculating the sound source power level of the UAV based on the self-noise spectrum, includes: Calculate the sound pressure level at each microphone measuring point based on the self-noise spectrum; Based on the distance parameters from each microphone to the acoustic center of the UAV and the directivity correction function value of each microphone in the direction of sound radiation corresponding to its installation position, the sound pressure level at each microphone measuring point is converted into the sound power level corresponding to that microphone. The average of the sound power levels calculated from each microphone is used to obtain the total sound source power level spectrum of the UAV under the current flight conditions.

6. The UAV noise prediction method based on flight parameter perception as described in claim 5, characterized in that, The value of the directional correction function is determined as follows: When spatially locating each frequency component using beamforming, the beam output power is calculated in each direction of the scanning space to form a beam power pattern. After normalizing the beam power pattern, take the logarithm to obtain the directivity correction function values ​​of the UAV in each direction.

7. The UAV noise prediction method based on flight parameter perception as described in claim 1, characterized in that, The propagation attenuation model includes one or more of the following: geometric divergence attenuation term, atmospheric absorption attenuation term, ground effect attenuation term, and building obstruction attenuation term; in step S5, the method for calculating the predicted sound pressure level of the UAV at the target receiving point includes one or more of the following: The straight-line distance between the spatial location and the target receiving point is calculated to determine the geometric divergence attenuation term; The ambient temperature and relative humidity are obtained, and the atmospheric absorption coefficient of each frequency band is calculated according to the atmospheric sound absorption standard to determine the atmospheric absorption attenuation term. The ground factor is determined based on the ground type, and the ground effect attenuation term is calculated according to the outdoor sound propagation attenuation standard. Acquire the building geometry data between the UAV and the target receiving point, perform ray tracking, and if there is building obstruction, calculate the diffraction path difference. Based on the diffraction path difference, use the diffraction attenuation formula to determine the building obstruction attenuation term. Subtract one or more of the geometric divergence attenuation term, atmospheric absorption attenuation term, ground effect attenuation term, and building obstruction attenuation term from the sound source power level to obtain the predicted sound pressure level of each frequency band at the target receiving point.

8. The UAV noise prediction method based on flight parameter perception as described in claim 7, characterized in that, Step S5, which involves calculating the predicted sound pressure level of the UAV at the target receiving point, further includes: The predicted sound pressure levels of each frequency band are superimposed after applying A-weighting correction to synthesize the total A-weighted sound pressure level at the target receiving point.

9. The UAV noise prediction method based on flight parameter perception as described in claim 8, characterized in that, In step S5, the calculation step for the predicted sound pressure level further includes at least one of the following: The ground area below the UAV's flight path is divided into grids, and the A-weighted total sound pressure level is calculated for each grid point to generate a two-dimensional noise impact heat map. Calculate the equivalent continuous A-weighted sound level based on the time series of the A-weighted total sound pressure level within a preset time period: The predicted sound pressure level is compared with the preset standard limit, and the compliance assessment result and the value exceeding the limit are output. When the predicted sound pressure level exceeds the noise limit constraint, the system searches in reverse for flight status data that meets the noise limit constraint as a flight parameter adjustment scheme based on the mapping relationship between the current flight status data and the sound pressure level, and outputs route optimization suggestions.

10. A UAV noise prediction system based on flight parameter perception, used to implement the UAV noise prediction method based on flight parameter perception as described in any one of claims 1 to 9, characterized in that, include: The flight parameter sensing module is used to acquire flight status data during the flight of the UAV, and the flight status data includes at least the real-time rotation speed and spatial position of each rotor; The airborne noise monitoring module includes a microphone array deployed on the UAV body for synchronously acquiring raw sound pressure signals; The sound source intensity calculation module is used to preprocess the original sound pressure signal to filter out wind noise components and compensate for pseudo-sound components introduced by body vibration. It calculates the blade passing frequency based on the real-time rotation speed of each rotor and extracts rotor noise harmonic components from the spectrum of the preprocessed sound pressure signal. It performs background noise separation on the spectrum after deducting the rotor noise harmonic components to obtain the self-noise spectrum of the UAV under the current flight conditions, and calculates the sound source power level of the UAV based on the self-noise spectrum. The noise propagation calculation module is used to calculate the predicted sound pressure level of the UAV at the target receiving point based on the spatial location and the sound power level of the sound source, combined with a preset propagation attenuation model. The noise impact prediction output module is used to output the predicted sound pressure level.