Non-uniform array wake-up system and method for fan blade acoustic fingerprint monitoring

Abstract

The present application relates to the technical field of wind power monitoring, and particularly relates to a non-uniform array wake-up system and method for wind turbine blade acoustic monitoring. The non-linear non-uniform spacing arrangement of the acoustic fingerprint collection array simultaneously considers the large array aperture required for low-frequency aerodynamic noise monitoring and the high spatial resolution required for high-frequency damage characteristic sound monitoring within a limited aperture, thereby improving the collection quality of full-band acoustic fingerprint signals. The signal processing module differentiates and processes low-frequency components and high-frequency components according to the spatial positions of the acoustic sensors, realizes frequency-space mapping, and further enhances the processing specificity and signal-to-noise ratio of signals of different frequency bands. At the same time, the wake-up trigger module arranged in the hub or the front end of the nacelle continuously monitors the blade rotation characteristic signal, and only wakes up the main system to enter the working state when an effective trigger condition is identified, thereby avoiding data redundancy and power waste caused by continuous collection, and overcoming the defect that key signals are easily missed in timed collection.

Description

Technical Field

[0001] This invention relates to the field of wind power monitoring technology, and in particular to a non-uniform array wake-up system and method for monitoring the acoustic signature of wind turbine blades. Background Technology

[0002] Wind turbine blades are the core components of wind turbine units that capture wind energy. Operating under harsh environments such as alternating loads, wind and sand erosion, lightning strikes, and extreme temperatures for extended periods, they are prone to damage such as cracks, delamination, pinholes, leading-edge corrosion, and even breakage, severely impacting the safety and power generation efficiency of the unit. Currently, wind turbine blade damage detection technologies mainly include drone inspections, infrared thermal imaging, manual inspection, and acoustic emission testing. Drone inspections capture images of the blades using drones equipped with high-definition cameras, and then use deep learning algorithms to identify surface defects. However, during turbine operation, the high blade tip rotation speed prevents drones from approaching, necessitating shutdown for operation, and these methods can only detect surface damage. Infrared thermal imaging converts infrared radiation from the blade surface into thermal images, using surface temperature differences to determine internal defects. However, its detection results are significantly affected by factors such as lighting angle, ambient temperature, and small temperature differences on the blade surface. Manual inspection mainly includes visual inspection and basket inspection, which have blind spots, require shutdown, and are time-consuming and labor-intensive. Acoustic emission detection uses high-sensitivity sensors to capture stress waves spontaneously released by defects under stress, enabling dynamic monitoring of crack initiation and propagation. However, it is difficult to detect existing static defects, is sensitive to environmental noise, and requires complex systems with long-distance sensor deployment. All of these technologies suffer from problems such as requiring shutdown for inspection, only being able to detect surface defects, being highly susceptible to environmental factors, or having complex systems, making it difficult to meet the practical needs of wind turbines for uninterrupted, low-cost, and 24 / 7 online monitoring.

[0003] Acoustic fingerprint-based monitoring technology, as a non-contact, continuously operating method for monitoring the health status of wind turbine blades, has gradually gained attention in recent years. This technology uses microphone arrays placed in the wind turbine nacelle or tower to collect aerodynamic noise and damage characteristic sound signals generated during blade operation, thereby achieving blade condition assessment and fault early warning. However, existing acoustic fingerprint acquisition devices have significant shortcomings in their placement schemes: Firstly, microphones are mostly arranged in a uniform, fixed array, which cannot be optimized according to the frequency distribution characteristics of the blade sound source. The low-frequency aerodynamic noise generated by blade rotation and the mid-to-high-frequency damage characteristic sounds differ significantly in their spatial propagation characteristics. A uniform array cannot simultaneously meet the requirements of large aperture for low frequencies and high spatial resolution for high frequencies, resulting in poor signal acquisition quality. Secondly, online monitoring devices currently employ two operating modes: continuous acquisition and timed acquisition. Continuous acquisition generates a large amount of invalid data, causing excessive data storage and computational pressure, while timed acquisition carries the risk of missing key damage signals. Therefore, there is an urgent need for an acoustic fingerprint monitoring device placement scheme that can adapt to the frequency distribution characteristics of wind turbine blade sound sources, improve signal acquisition quality, and reduce system power consumption. Summary of the Invention

[0004] The technical problem to be solved by the present invention is to provide a non-uniform array wake-up system and method for monitoring the acoustic signature of wind turbine blades, so as to solve the above-mentioned technical problem.

[0005] The technical solution of this invention to solve the above-mentioned technical problems is as follows: A non-uniform array wake-up system for wind turbine blade acoustic signature monitoring, comprising: an acoustic signature acquisition array, a signal processing module, and a wake-up triggering module; the acoustic signature acquisition array includes multiple acoustic sensors arranged along a preset direction, the adjacent spacing of the multiple acoustic sensors varying non-linearly along the preset direction, so as to form a low-frequency sensing area with relatively large adjacent spacing and a high-frequency sensing area with relatively small adjacent spacing in the acoustic signature acquisition array along the preset direction, the low-frequency sensing area being used to acquire low-frequency acoustic signals generated by blade rotation, and the high-frequency sensing area being used to acquire high-frequency acoustic signals generated by blade damage; the signal processing module... The processing module is connected to the acoustic fingerprint acquisition array and is used to extract low-frequency components from the signals collected by the acoustic sensors in the low-frequency sensing area and process them, and to extract high-frequency components from the signals collected by the acoustic sensors in the high-frequency sensing area, based on the spatial position of each acoustic sensor in the acoustic fingerprint acquisition array. The wake-up trigger module is located in the hub of the wind turbine or at the front end of the nacelle and is used to continuously monitor the characteristic signals generated by the rotation of the blades and generate a wake-up signal when the characteristic signals meet the preset trigger conditions. The acoustic fingerprint acquisition array and / or the signal processing module are configured to switch from a sleep state to a working state after receiving the wake-up signal.

[0006] The beneficial effects of this invention are as follows: By employing a nonlinear, non-uniform spacing arrangement of the acoustic signature acquisition array, both the large array aperture required for low-frequency aerodynamic noise monitoring and the high spatial resolution required for high-frequency damage characteristic acoustic monitoring are simultaneously achieved within a limited aperture range, thereby improving the acquisition quality of acoustic signature signals across the entire frequency band. The signal processing module differentiates and processes low-frequency and high-frequency components based on the spatial location of the acoustic sensor, achieving frequency-space mapping and further enhancing the processing specificity and signal-to-noise ratio of different frequency bands. Simultaneously, a wake-up trigger module located within the hub or at the front of the nacelle continuously monitors the blade rotation characteristic signals, waking up the main system only when a valid trigger condition is detected. This avoids data redundancy and power consumption waste associated with continuous acquisition and overcomes the drawback of missed key signals in timed acquisition, achieving low-power operation and high-reliability triggering of the acoustic signature monitoring system.

[0007] Based on the above technical solution, the present invention can be further improved as follows.

[0008] Furthermore, the position coordinates of the i-th acoustic sensor in the acoustic signature acquisition array Defined by the following log-interval distribution function: ; Where L is the total length of the acoustic signature array, and M is the total number of acoustic sensors. To control the nonlinear coefficient of the spacing change rate, and .

[0009] Furthermore, the wake-up trigger module includes: The signal acquisition unit is used to continuously acquire the characteristic signals generated by the rotation of the blades. The signal conditioning unit is used to filter and amplify the acquired feature signals. The feature extraction unit is used to extract the amplitude features of the conditioned feature signal; The feature recognition unit is used to determine whether the preset triggering condition is met based on the amplitude feature, and outputs the wake-up signal when the condition is met.

[0010] Furthermore, the preset triggering condition includes: the amplitude of the feature signal exceeds a preset amplitude threshold, and the duration of the amplitude exceeding the preset amplitude threshold is within a preset time window.

[0011] Furthermore, the preset triggering conditions also include: within a preset observation period, detecting a feature signal that continuously meets the condition that the amplitude exceeds a preset amplitude threshold multiple times and the duration of the amplitude exceeding the preset amplitude threshold is within a preset time window, and the error between the interval of adjacent feature signals and the theoretical blade passage period is less than a preset value.

[0012] Furthermore, the signal processing module includes a multi-rate filter bank, which is used to divide the time-domain signals collected by each acoustic sensor into multiple sub-bands, and to use delay summation beamforming for low-frequency sub-bands with frequencies lower than a first preset frequency, and to use minimum variance distortion-free response beamforming for high-frequency sub-bands with frequencies higher than a second preset frequency.

[0013] Furthermore, the system also includes a power management module, which provides independent power to the wake-up trigger module and controls the power domain switching of the voiceprint acquisition array and the signal processing module.

[0014] Furthermore, after the acoustic fingerprint acquisition array or the signal processing module is woken up, it enters the working state. When the time without a valid blade acoustic fingerprint signal exceeds the preset sleep time, it automatically returns to the sleep state. When the wake-up pulse interval generated by the wake-up trigger module is less than the preset time multiple times in a row, the acoustic fingerprint acquisition array or the signal processing module is in a cyclic sleep-wake-up mode.

[0015] Furthermore, the acoustic signature acquisition array is mounted on a bracket at the connection between the wind turbine nacelle and the tower. The bracket is a ring bracket or a semi-ring bracket, and the orientation of each acoustic sensor is perpendicular to the blade rotation plane.

[0016] To address the aforementioned technical problems, this invention also provides a non-uniform array wake-up method for wind turbine blade acoustic signature monitoring, comprising: continuously monitoring characteristic signals generated by blade rotation through a wake-up trigger module disposed within the hub or at the front end of the nacelle of the wind turbine, and generating a wake-up signal when the characteristic signal is found to meet a preset trigger condition; sending the wake-up signal to an acoustic signature acquisition array, controlling the acoustic signature acquisition array to switch from a dormant state to an operating state; acquiring blade acoustic signature signals through multiple acoustic sensors arranged along a preset direction in the acoustic signature acquisition array, wherein the adjacent spacing of the multiple acoustic sensors varies non-linearly along the preset direction, so as to form a low-frequency sensing area with relatively large adjacent spacing and a high-frequency sensing area with relatively small adjacent spacing along the preset direction in the acoustic signature acquisition array; extracting and processing low-frequency components from the signals acquired by the acoustic sensors in the low-frequency sensing area, and extracting and processing high-frequency components from the signals acquired by the acoustic sensors in the high-frequency sensing area, based on the spatial position of each acoustic sensor in the acoustic signature acquisition array. Attached Figure Description

[0017] Figure 1 This is a schematic diagram of the non-uniform array wake-up system for monitoring the acoustic signature of wind turbine blades according to the present invention; Figure 2 This is a graph showing the spacing distribution of the biomimetic non-uniform array of the present invention. Figure 3 This is a signal timing diagram of the intermittent wake-up energy-saving arrangement of the present invention; Figure 4 This is a flowchart of the non-uniform array wake-up method for wind turbine blade acoustic signature monitoring according to the present invention.

[0018] The attached diagram lists the components represented by each number as follows: 1. Acoustic fingerprint array; 2. Blade; 3. Hub; 4. Nacelle; 5. Tower. Detailed Implementation

[0019] The principles and features of the present invention are described below. The examples given are only for explaining the present invention and are not intended to limit the scope of the present invention.

[0020] Example 1 like Figure 1As shown, this embodiment provides a non-uniform array wake-up system for wind turbine blade acoustic signature monitoring, including: an acoustic signature acquisition array 1, a signal processing module, and a wake-up trigger module; the acoustic signature acquisition array 1 includes multiple acoustic sensors arranged along a preset direction, the adjacent spacing of the multiple acoustic sensors changing non-linearly along the preset direction, so as to form a low-frequency sensing area with relatively large adjacent spacing and a high-frequency sensing area with relatively small adjacent spacing in the acoustic signature acquisition array 1 along the preset direction. The low-frequency sensing area is used to collect low-frequency acoustic signals generated by the rotation of the blade 2, and the high-frequency sensing area is used to collect high-frequency acoustic signals generated by damage to the blade 2; the signal processing module is connected to the acoustic signature acquisition array 1, and is used to extract low-frequency components from the signals collected by the acoustic sensors in the low-frequency sensing area and process them, and extract high-frequency components from the signals collected by the acoustic sensors in the high-frequency sensing area, according to the spatial position of each acoustic sensor in the acoustic signature acquisition array 1; the wake-up trigger module is set in the hub 3 or the front end of the nacelle 4 of the wind turbine, and is used to continuously monitor the characteristic signals generated by the rotation of the blade 2, and generate a wake-up signal when the characteristic signal meets the preset trigger conditions; the acoustic signature acquisition array 1 and / or the signal processing module are configured to switch from a dormant state to a working state after receiving the wake-up signal.

[0021] This system improves the acquisition quality of acoustic signature signals across the entire frequency band by employing a non-linear, non-uniform spacing arrangement of the acoustic signature acquisition array 1, which simultaneously accommodates the large array aperture required for low-frequency aerodynamic noise monitoring and the high spatial resolution required for high-frequency damage characteristic acoustic monitoring within a limited aperture. The signal processing module differentiates and processes low-frequency and high-frequency components based on the spatial location of the acoustic sensors, achieving frequency-space mapping and further enhancing the processing specificity and signal-to-noise ratio of different frequency bands. Simultaneously, a wake-up trigger module located within the hub 3 or at the front of the nacelle 4 continuously monitors the rotational characteristic signals of the blades 2, waking up the main system only when a valid trigger condition is detected. This avoids data redundancy and power consumption waste associated with continuous acquisition and overcomes the drawback of missed key signals in timed acquisition, achieving low-power operation and high-reliability triggering of the acoustic signature monitoring system.

[0022] Specifically, the acoustic signature acquisition array 1 is mounted on a bracket at the connection between the wind turbine nacelle 4 and the tower 5, and is used to acquire acoustic signature signals generated during the rotation of the blades 2. The signal processing module is electrically connected to the acoustic signature acquisition array 1 and is used to perform filtering, beamforming, and feature extraction on the acquired multi-channel acoustic signature signals. The wake-up trigger module is located inside the wind turbine hub 3 or at the front end of the nacelle 4, and is used to continuously monitor the characteristic signals generated by the rotation of the blades 2, and generate a wake-up signal when a preset trigger condition is met, which is then sent to the acoustic signature acquisition array 1 and / or the signal processing module.

[0023] The acoustic signature acquisition array 1 and the signal processing module are in sleep mode by default, i.e., low-power mode. When the wake-up trigger module recognizes a valid blade 2 rotation characteristic signal and issues a wake-up signal, the acoustic signature acquisition array 1 and / or the signal processing module switch from sleep mode to working mode and begin high-speed acquisition and processing of blade 2 acoustic signature signals. If there is no valid blade 2 acoustic signature signal for more than the preset sleep duration, the acoustic signature acquisition array 1 and / or the signal processing module automatically return to sleep mode.

[0024] Optionally, in this embodiment, the acoustic signature acquisition array 1 is mounted on a bracket at the connection between the wind turbine nacelle 4 and the tower 5. The bracket is a ring bracket or a semi-ring bracket, and the orientation of each acoustic sensor is perpendicular to the plane of rotation of the blade 2.

[0025] In this embodiment, the acoustic signature acquisition array 1 includes multiple acoustic sensors, such as microphones, arranged along a preset direction. The array is mounted on a bracket at the connection between the wind turbine nacelle 4 and the tower 5. The bracket can be a ring bracket (suitable for 360° omnidirectional monitoring, capable of distinguishing different blades 2) or a semi-ring bracket (120° to 180°, suitable for saving channel count and monitoring only the acoustic signature when the blade 2 sweeps across the front of the tower 5). The array is installed at a height of 1m to 2m behind the center of the hub 3, and the array is oriented perpendicular to the plane of rotation of the blade 2, i.e., pointing towards the wind turbine. The orientation of each acoustic sensor is also perpendicular to the plane of rotation of the blade 2.

[0026] The spacing between adjacent acoustic sensors varies non-linearly along a preset direction. Specifically, along the array direction, the spacing between acoustic sensors gradually decreases as the frequency increases, thus forming a low-frequency sensing area with relatively large adjacent spacing and a high-frequency sensing area with relatively small adjacent spacing in the acoustic signature acquisition array 1 along the preset direction. The low-frequency sensing area is located in the near-field region of the blade 2 near the hub 3, and is used to collect low-frequency acoustic signals (center frequency 20Hz to 500Hz) generated by the rotation of the blade 2; the high-frequency sensing area is distributed with gradually varying density along the axial direction of the tower 5, and is used to collect high-frequency acoustic signals (center frequency 2kHz to 20kHz) generated by damage to the blade 2.

[0027] In one specific implementation, the voiceprint acquisition array 1 includes M sound sensors (M=8, 12, 16 or 24), arranged at equal intervals along a horizontal straight line or arc (meaning equal intervals in the mapping dimension, not physical equal intervals).

[0028] Optionally, in the embodiment, the position coordinates of the i-th acoustic sensor in the voiceprint acquisition array 1 Defined by the following log-interval distribution function: ; Where L is the total length (array aperture) of the acoustic signature acquisition array 1, typically ranging from 0.8m to 2.0m, and determined based on the space in cabin 4; M is the total number of acoustic sensors; To control the nonlinear coefficient of the spacing change rate, Preferably, The value ranges from 2 to 5 to achieve a good balance between aperture gain in the low-frequency sensing region and spatial resolution in the high-frequency sensing region.

[0029] According to the above formula, the spacing between adjacent acoustic sensors for: .

[0030] when At time, spacing The spacing decreases as i increases, meaning the larger spacing at the front end of the array (closer to hub 3) corresponds to the low-frequency sensing region; the smaller spacing at the rear end of the array (farthest from hub 3) corresponds to the high-frequency sensing region, such as... Figure 2 As shown. This arrangement mimics the unique frequency-position mapping characteristics of the human cochlea, mapping different frequency components generated by the two sound sources on the blades to different spatial positions of the array: high-frequency signals are perceived by microphone pairs with small spacing (high spatial sampling rate), and low-frequency signals are perceived by microphone pairs with large spacing (low spatial sampling rate, but large aperture). The array as a whole exhibits a distribution characteristic in which the spacing decreases exponentially with increasing frequency.

[0031] The signal processing module is connected to the voiceprint acquisition array 1 and is used to perform digital processing and feature extraction on multi-channel voiceprint signals.

[0032] Optionally, in an embodiment, the signal processing module includes a multi-rate filter bank, used to divide the time-domain signals acquired by each acoustic sensor into multiple sub-bands, and to use delay summation beamforming for low-frequency sub-bands with frequencies lower than a first preset frequency, and to use minimum variance distortion-free response beamforming for high-frequency sub-bands with frequencies higher than a second preset frequency.

[0033] The signal processing module first divides the time-domain signals acquired by each acoustic sensor into multiple sub-bands using a multi-rate filter bank. In one specific implementation, the number of sub-bands K≥5; for example, when K=5, it can be divided into five frequency bands: ultra-low frequency band, low frequency band, mid frequency band, high frequency band, and ultra-high frequency band.

[0034] The signal processing module employs different beamforming strategies for signals in different frequency bands based on the spatial position of each acoustic sensor within the acoustic signature acquisition array 1. Specifically, for the low-frequency sub-band with frequencies below a first preset frequency (e.g., 500Hz), the signal processing module uses only the signals from the first 1 to M / 3 channels, i.e., the acoustic sensor signals in the low-frequency sensing area, and employs delayed summation beamforming. Using a large-aperture array in the low-frequency band can achieve narrower directivity, effectively extracting the low-frequency aerodynamic noise signals generated by the rotation of blade 2.

[0035] For high-frequency sub-bands with frequencies higher than the second preset frequency (e.g., 2kHz), the signal processing module only uses the signals from the last M / 2 to M channels, i.e., the acoustic sensor signals in the high-frequency sensing area, employing minimum variance distortion-free response (MVDR) beamforming. Using a small-pitch array in the high-frequency band avoids spatial aliasing, while MVDR beamforming effectively suppresses environmental noise and improves the spatial resolution of the damage characteristic acoustic signals.

[0036] For the intermediate frequency subband, the signal processing module uses signals from all channels and employs adaptive beamforming.

[0037] In addition, the signal processing module multiplies the signal of each channel by a weighting coefficient w. i The weighting coefficients for the low-frequency channels are set relatively large to highlight the monitoring of aerodynamic noise; the weighting coefficients for the high-frequency channels are adaptively adjusted according to the signal-to-noise ratio to optimize the extraction of damage feature sounds.

[0038] The signal processing module also utilizes the passing time of blade 2 provided by the wind turbine speed encoder or the wake-up trigger module to perform synchronization time window truncation and overall averaging for each blade 2, thereby improving the signal-to-noise ratio. Specifically, the following synchronization averaging formula is used: ; in, This is the voiceprint signal output after synchronization and averaging. The original time-domain signal is collected by the acoustic sensor. T is the passing period of blade 2 (single blade 2), and N is the number of rotations involved in the averaging, which is recommended to be between 10 and 100.

[0039] Optionally, in this embodiment, the wake-up trigger module includes: The signal acquisition unit is used to continuously acquire the characteristic signals generated by the rotation of blade 2; The signal conditioning unit is used to filter and amplify the acquired feature signals. The feature extraction unit is used to extract the amplitude features of the conditioned feature signal; The feature recognition unit is used to determine whether the preset triggering conditions are met based on the amplitude characteristics, and outputs a wake-up signal when the conditions are met.

[0040] The wake-up trigger module includes a signal acquisition unit, a signal conditioning unit, a feature extraction unit, and a feature recognition unit.

[0041] In one specific implementation, the signal acquisition unit employs a low-power pre-electret condenser microphone or a piezoelectric pressure sensor, and is located inside the hub 3 of the wind turbine or at the front end of the nacelle 4. This unit operates continuously, monitoring the signals generated by the rotation of the blades 2 in real time, without needing to wake up the main system.

[0042] When the wind turbine is running, as each blade 2 rotates past the front of the tower 5, a low-pressure zone is generated on the surface of the blade 2 due to the obstruction effect of the tower 5 on the airflow. This low-pressure zone sweeps across a fixed position (such as the front of the nacelle 4 or the side of the hub 3) along with the blade 2, thereby generating a periodic air pressure pulsation signal at a fixed point in space. The pulse width corresponds to the time it takes for the blade 2 to pass through the obstruction zone of the tower 5. The frequency range of this low-frequency air pressure pulsation signal is 0.5Hz to 5Hz, and it is the core triggering characteristic signal of this scheme.

[0043] The signal conditioning unit is used to filter and amplify the acquired characteristic signals. Specifically, the signal conditioning unit includes a low-pass filter (cutoff frequency, for example, 10Hz) and a two-stage amplifier circuit. The low-pass filter is used to filter out high-frequency noise components and retain low-frequency air pressure pulsation signals; the two-stage amplifier circuit is used to amplify the weak sensor signals to an amplitude range that can be processed by subsequent circuits.

[0044] The feature extraction unit is used to extract the amplitude features of the conditioned feature signal. Specifically, the feature extraction unit includes an envelope detector, used to extract the time-domain amplitude envelope, reflecting the amplitude variation trend of the feature signal.

[0045] The feature recognition unit is used to determine whether the preset triggering conditions are met based on the amplitude characteristics, and outputs a wake-up signal when the conditions are met.

[0046] Optionally, in an embodiment, the preset triggering condition includes: the amplitude of the feature signal exceeds a preset amplitude threshold, and the duration of the amplitude exceeding the preset amplitude threshold is within a preset time window.

[0047] The preset amplitude threshold can be adjusted according to the wind turbine model calibration, and the preset time window range can be determined based on the theoretical minimum and maximum values ​​of the pulse width of blade 2, calculated in real time or pre-stored according to the current rotational speed. This forms a dual-threshold time window verification method: the amplitude threshold is used to eliminate small-amplitude noise, and the time window is used to eliminate atypical pulse signals such as transient spikes and slow drifts. Specifically, the threshold comparison can be achieved through a comparator.

[0048] In this embodiment, the discrimination logic of blade 2 using the double threshold time window verification method through the feature identifier (BTI) is as follows: Level 1 threshold (amplitude threshold): The preset amplitude threshold is 5 Pa. When the envelope amplitude is greater than 5 Pa, the time counter is started.

[0049] Second-level threshold (duration verification): Requires the duration Δt of the amplitude exceeding 5 Pa to meet the following conditions: 0.1Tpulse-min<Δt<1.5Tpulse-max; Among them, Tpulse-min and Tpulse-max are the theoretical minimum and maximum values ​​of the pulse width through which blade 2 passes.

[0050] It should be noted that 0.1 is used to eliminate transient spike noise such as electrical noise or lightning strikes. Tpulse-min is generally between 0.2s and 0.8s, while the duration of most transient impact noises is between 10ms and 20ms. 1.5 is used to eliminate slowly varying pressure drift caused by sudden gusts, changes in turbulence scale, or slow fluctuations in atmospheric pressure. The fluctuation rate of wind turbine speed in a short period of time is generally within ±20% to ±30%, and can reach ±40% under extreme conditions, so 1.5 is used.

[0051] Optionally, in the embodiment, the preset triggering condition further includes: within a preset observation period, detecting a feature signal that continuously meets the requirement that the amplitude exceeds a preset amplitude threshold and the duration of the amplitude exceeding the preset amplitude threshold is within a preset time window, and the error between the interval of adjacent feature signals and the theoretical blade 2 passing period is less than a preset value.

[0052] Furthermore, in order to prevent false triggering caused by non-periodic disturbances such as sudden wind pressure changes and gust impacts, the feature recognition unit also adopts periodic verification logic: within a preset observation period, if a feature signal is detected multiple times (e.g., more than twice) that meets the above-mentioned amplitude exceeding the preset amplitude threshold and the duration is within the preset time window range, and the error between the interval of adjacent feature signals and the theoretical blade 2 passing period is less than a preset value (e.g., less than 20%), it is confirmed as a valid blade 2 rotation signal, and a wake-up signal is output.

[0053] Only after successful verification will the feature recognition unit output a high-level active wake-up pulse signal, lasting for, for example, 10 to 50 ms. Figure 3 As shown.

[0054] Optionally, in an embodiment, the system further includes a power management module, which provides independent power to the wake-up trigger module and controls the power domain switching of the voiceprint acquisition array 1 and the signal processing module.

[0055] The power management module includes an independent low-noise power supply unit and a power domain switching unit. The independent low-noise power supply unit provides a continuous and stable low-noise power supply to the wake-up trigger module, ensuring that the wake-up trigger module can operate reliably for a long time. The power domain switching unit receives the wake-up pulse signal from the wake-up trigger module and switches the power domain of the acoustic pattern acquisition array 1 and the signal processing module from 0V to the operating voltage through a MOSFET switch.

[0056] Optionally, in the embodiment, the acoustic fingerprint acquisition array 1 or the signal processing module enters the working state after being woken up. When the time without a valid acoustic fingerprint signal from the blade 2 exceeds the preset sleep time, it automatically returns to the sleep state. When the wake-up pulse interval generated by the wake-up trigger module is less than the preset time multiple times in a row, the acoustic fingerprint acquisition array 1 or the signal processing module is in a cyclic sleep-wake-up mode.

[0057] The system's workflow is as follows: Standby monitoring phase: The signal acquisition unit of the wake-up trigger module continues to work, acquiring the low-frequency air pressure pulsation characteristic signal generated by the rotation of blade 2 in real time. The signal conditioning unit performs low-pass filtering and two-stage amplification on the acquired signal, and the feature extraction unit extracts the signal amplitude envelope through an envelope detector. At this time, the acoustic signature acquisition array 1 and the signal processing module are in sleep mode with extremely low power consumption.

[0058] Triggering the recognition phase: The feature recognition unit judges the extracted amplitude envelope. First, a dual-threshold time window verification is performed. When the amplitude exceeds the preset amplitude threshold, a time counter is started. If the duration of the amplitude exceeding the preset amplitude threshold meets the preset time window range, then the single pulse verification is passed. Then, periodic verification (optional enhancement logic) is performed: If more than two consecutive pulses that pass the single verification are detected within the preset observation period, and the error between the interval of adjacent pulses and the theoretical blade 2 passing period is less than 20%, then it is confirmed as a valid blade 2 rotation signal.

[0059] Wake-up Phase: After successful verification, the feature recognition unit outputs a high-level wake-up pulse signal (lasting 10 to 50 ms). Upon receiving the wake-up pulse, the power domain switching unit of the power management module switches the power domain of the voiceprint acquisition array 1 and the signal processing module from 0V to the operating voltage via a MOSFET switch, and the system enters the working state.

[0060] Operational monitoring phase: After the acoustic signature acquisition array 1 is activated, it begins high-speed sampling to acquire the acoustic signature signal of blade 2. The signal processing module performs filtering, beamforming, and feature extraction on the acquired multi-channel signals to achieve blade 2 status monitoring and fault early warning.

[0061] Sleep Control Phase: After the voiceprint acquisition array 1 and the signal processing module are awakened, they enter the working state and continuously monitor signal activity. Continuous working conditions are: the detected envelope amplitude is greater than a threshold (which can be set to half of a preset amplitude threshold) or the high-frequency characteristic band energy is more than 5dB higher than the background noise. When the continuous working conditions are met, the system remains in working state; when the continuous working conditions are not met, the system enters a cyclic sleep state.

[0062] When the wake-up pulse interval generated by the wake-up trigger module is consistently less than 5 seconds, the fan is in normal operation and the system enters a cyclic sleep-wake-up mode. That is, it is woken up each time blade 2 passes to collect acoustic fingerprint signals, and automatically returns to sleep mode after collection is completed. When there is no valid acoustic fingerprint signal from blade 2 for more than the preset sleep time (e.g., 30 seconds), the system determines that the fan is shut down and enters a deep sleep mode.

[0063] In addition, the system is also equipped with a forced timeout protection function: if the continuous working time after a single wake-up exceeds the preset timeout threshold (e.g., 10 minutes), the system will automatically enter a hibernation state to prevent abnormal freezing.

[0064] The following example illustrates the energy-saving effect of this system: Assuming a 2.5MW wind turbine operates at 12RPM and has 3 blades, then the frequency f passing through blade 2 is... blade =3×12 / 60=0.6Hz, which means it passes through once every 1.67 seconds, and the pulse width is about 0.3 seconds.

[0065] In traditional continuous acquisition schemes, the main array power consumption is 5W (typical value), and the daily energy consumption is 5×24=120Wh.

[0066] The automatic wake-up scheme of this system is adopted as follows: the power consumption of the pre-triggered unit is 0.5mW, and the daily energy consumption is 0.012Wh; the main array only wakes up and works for 0.6 seconds after each passage of blade 2 (the first 0.3 seconds monitor the pulse itself, and the last 0.3 seconds additionally sample the acoustic fingerprint of blade 2), the total daily working time = 0.6×(24×3600 / 1.67)≈31042 seconds≈8.62 hours; the daily energy consumption of the main array is 5×8.62=43.1Wh; the total energy consumption is 0.012+43.1≈43.112Wh; the energy saving ratio is 120 / 43.112≈2.78 times, that is, the power consumption is reduced to 35.9% of the traditional scheme.

[0067] It is evident that this system significantly reduces system power consumption while ensuring monitoring reliability through an intermittent wake-up mechanism.

[0068] Example 2 like Figure 4 As shown, this embodiment provides a non-uniform array wake-up method for acoustic signature monitoring of wind turbine blades, including: S101. The wake-up trigger module installed in the hub 3 of the wind turbine or at the front end of the nacelle 4 continuously monitors the characteristic signals generated by the rotation of the blades 2, and generates a wake-up signal when the characteristic signals meet the preset trigger conditions. S102. Send a wake-up signal to the voiceprint acquisition array 1 to control the voiceprint acquisition array 1 to switch from sleep state to working state; S103. Acoustic signals of blade 2 are collected by multiple acoustic sensors arranged along a preset direction in acoustic pattern acquisition array 1. The adjacent spacing of the multiple acoustic sensors changes non-linearly along the preset direction, so as to form a low-frequency sensing area with relatively large adjacent spacing and a high-frequency sensing area with relatively small adjacent spacing in acoustic pattern acquisition array 1 along the preset direction. S104. Based on the spatial position of each acoustic sensor in the acoustic pattern acquisition array 1, extract the low-frequency component from the signal acquired by the acoustic sensor in the low-frequency sensing area and process it, and extract the high-frequency component from the signal acquired by the acoustic sensor in the high-frequency sensing area and process it.

[0069] The above description is merely a preferred embodiment of the present invention and an explanation of the technical principles employed. Those skilled in the art should understand that the scope of disclosure in this invention is not limited to technical solutions formed by specific combinations of the above-described technical features, but should also cover other technical solutions formed by arbitrary combinations of the above-described technical features or their equivalents without departing from the above-disclosed concept. For example, technical solutions formed by substituting the above features with (but not limited to) technical features with similar functions disclosed in this invention.

[0070] It should be noted that the terms "first," "second," etc., used in the specification and claims of this application are used to distinguish similar objects and do not imply a specific order or sequence. Where appropriate, the order of use for similar objects can be interchanged so that the embodiments of this application described herein can be implemented in an order other than that shown or described.

[0071] Those skilled in the art will recognize that this invention can be implemented as a system, method, or computer program product. Therefore, this invention can be specifically implemented in the following forms: it can be entirely hardware, entirely software (including firmware, resident software, microcode, etc.), or a combination of hardware and software, generally referred to herein as a "circuit," "module," or "system." Furthermore, in some embodiments, this invention can also be implemented as a computer program product contained in one or more computer-readable media, which includes computer-readable program code.

[0072] Although embodiments of the present invention have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting the present invention. Those skilled in the art can make changes, modifications, substitutions and variations to the above embodiments within the scope of the present invention.

Claims

1. A non-uniform array wake-up system for acoustic signature monitoring of wind turbine blades, characterized in that, include: Voiceprint acquisition array, signal processing module, and wake-up trigger module; The acoustic signature acquisition array includes multiple acoustic sensors arranged along a preset direction. The spacing between adjacent acoustic sensors varies non-linearly along the preset direction to form a low-frequency sensing area with relatively large adjacent spacing and a high-frequency sensing area with relatively small adjacent spacing in the acoustic signature acquisition array along the preset direction. The low-frequency sensing area is used to acquire low-frequency acoustic signals generated by blade rotation, and the high-frequency sensing area is used to acquire high-frequency acoustic signals generated by blade damage. The signal processing module is connected to the acoustic pattern acquisition array and is used to extract low-frequency components from the signals acquired by the acoustic sensors in the low-frequency sensing area and process them, based on the spatial position of each acoustic sensor in the acoustic pattern acquisition array, and to extract high-frequency components from the signals acquired by the acoustic sensors in the high-frequency sensing area and process them. The wake-up trigger module is located inside the hub of the wind turbine or at the front end of the nacelle. It is used to continuously monitor the characteristic signals generated by the rotation of the blades and generate a wake-up signal when the characteristic signals meet the preset trigger conditions. The voiceprint acquisition array and / or the signal processing module are configured to switch from a sleep state to a working state upon receiving the wake-up signal.

2. The non-uniform array wake-up system for wind turbine blade acoustic signature monitoring according to claim 1, characterized in that, The position coordinates of the i-th sound sensor in the voiceprint acquisition array Defined by the following log-interval distribution function: ； Where L is the total length of the acoustic signature acquisition array, and M is the total number of acoustic sensors. To control the nonlinear coefficient of the spacing change rate, and .

3. The non-uniform array wake-up system for wind turbine blade acoustic signature monitoring according to claim 1, characterized in that, The wake-up trigger module includes: The signal acquisition unit is used to continuously acquire the characteristic signals generated by the rotation of the blades. The signal conditioning unit is used to filter and amplify the acquired feature signals. The feature extraction unit is used to extract the amplitude features of the conditioned feature signal; The feature recognition unit is used to determine whether the preset triggering condition is met based on the amplitude feature, and outputs the wake-up signal when the condition is met.

4. The non-uniform array wake-up system for wind turbine blade acoustic signature monitoring according to claim 3, characterized in that, The preset triggering conditions include: the amplitude of the feature signal exceeds a preset amplitude threshold, and the duration of the amplitude exceeding the preset amplitude threshold is within a preset time window.

5. The non-uniform array wake-up system for wind turbine blade acoustic signature monitoring according to claim 4, characterized in that, The preset triggering conditions also include: within a preset observation period, detecting a feature signal that continuously exceeds a preset amplitude threshold multiple times and whose duration exceeds the preset amplitude threshold is within a preset time window, and the error between the interval of adjacent feature signals and the theoretical blade passage period is less than a preset value.

6. The non-uniform array wake-up system for wind turbine blade acoustic signature monitoring according to claim 1, characterized in that, The signal processing module includes a multi-rate filter bank, which is used to divide the time-domain signals collected by each acoustic sensor into multiple sub-bands, and to use delay summation beamforming for low-frequency sub-bands with frequencies lower than a first preset frequency, and to use minimum variance distortion-free response beamforming for high-frequency sub-bands with frequencies higher than a second preset frequency.

7. The non-uniform array wake-up system for wind turbine blade acoustic signature monitoring according to claim 1, characterized in that, It also includes a power management module, which provides independent power to the wake-up trigger module and controls the power domain switching of the voiceprint acquisition array and the signal processing module.

8. The non-uniform array wake-up system for wind turbine blade acoustic signature monitoring according to claim 1, characterized in that, After being woken up, the acoustic fingerprint acquisition array or the signal processing module enters the working state. When the time without a valid blade acoustic fingerprint signal exceeds the preset sleep time, it automatically returns to the sleep state. When the wake-up pulse interval generated by the wake-up trigger module is less than the preset time multiple times in a row, the acoustic fingerprint acquisition array or the signal processing module is in a cyclic sleep-wake-up mode.

9. The non-uniform array wake-up system for wind turbine blade acoustic signature monitoring according to claim 1, characterized in that, The acoustic signature array is mounted on a bracket at the connection between the wind turbine nacelle and the tower. The bracket is either a ring bracket or a semi-ring bracket, and the orientation of each acoustic sensor is perpendicular to the blade rotation plane.

10. A non-uniform array wake-up method for acoustic signature monitoring of wind turbine blades, characterized in that, include: The wake-up trigger module, located inside the hub or at the front of the nacelle of the wind turbine, continuously monitors the characteristic signals generated by the rotation of the blades and generates a wake-up signal when the characteristic signals meet the preset trigger conditions. The wake-up signal is sent to the voiceprint acquisition array to control the voiceprint acquisition array to switch from sleep state to working state; The acoustic signature signal of the blade is collected by multiple acoustic sensors arranged along a preset direction in the acoustic signature acquisition array. The spacing between adjacent acoustic sensors varies non-linearly along the preset direction, so as to form a low-frequency sensing area with relatively large adjacent spacing and a high-frequency sensing area with relatively small adjacent spacing in the acoustic signature acquisition array along the preset direction. Based on the spatial position of each acoustic sensor in the acoustic pattern acquisition array, low-frequency components are extracted from the signals acquired by the acoustic sensors in the low-frequency sensing area and processed, and high-frequency components are extracted from the signals acquired by the acoustic sensors in the high-frequency sensing area and processed.

Images