Wind Turbine Sound Pattern Analysis

Acoustic monitoring system for equipment to detect component failures and abnormal conditions by using microphones to record sounds generated by the equipment during operation. The system compares the operational audio signature to a baseline signature to detect deviations indicative of issues. This allows monitoring components that are hard to access or where wear is hard to detect visually. It can provide early warning of failures and malfunctions, enabling proactive maintenance.

Source

Aiming at the problems of insufficient feature expression capability and high model computation complexity in traditional wind turbine group abnormal sound detection methods, this paper proposes a lightweight method based on improved MobileNetV3 network. First, SincNet bandpass filter Mel spectrum are integrated to construct multi-dimensional acoustic features, taking into account original signal time-domain features frequency-domain features. Second, dynamic frequency-aware convolution (DFC) module is introduced network architecture adaptively adjust parameters kernel through attention mechanism strengthen frequency capture sounds; downsampling process optimized by combining with SoftPool reduce loss high-frequency information. On dataset Danish University Science Technology, AUC reaches 94.71%, number only 2.38M, which 62.3% lower than mainstream ResNet-18, providing high-precision edge-end solution for status monitoring.

Source

Abstract. We present a digital shadow Kalman filtering approach based on the direct linearization of multibody aeroservoelastic model wind turbine. In contrast to approaches ad hoc models, reuse existing trusted models reduces development time and duplication effort, leverages resources invested in tuning validation, eventually increases confidence results. This has already been pursuded by others, but it is here improved with respect several main aspects formulation. To extend applicability non-symmetric, waked, yaw-misaligned conditions, filter-internal addition tower fore-aft rotor rotational dynamics now also includes side-side flapwise edgewise degrees freedom blades. make aware inflow conditions at disk, estimators are used detect real during operation rotor-equivalent values speed, vertical shear, horizontal shear (on account waked conditions), yaw misalignment (in support wake-steering control). These parameters schedule model, adapting its behavior current experienced Furthermore, white-box augmented data-driven corrections improve predictive accuracy. Two explored for c... Read More

Source

Studies that involve mitigating aerodynamic noise in rotating components such as rotors of wind turbines or propellers Unmanned Aerial Vehicles have gained immense interest the research community over last few years. The present study explores mitigation potential passive compliant coatings through Computational Aeroacoustics Analysis (CAA) and experimentation tunnel testing. CAA was performed on a flat plate for chord-based Reynolds number Re c = 460,000 using SST k- Improved Delayed Detached Eddy Simulation Ffowcs Williams Hawkings acoustic analogy. Trailing edge (TE) accurately predicted from 750 to 7000 Hz. Noise results were compared with cases where different material properties are applied onto plate. It observed coating-1 (Dow Corning Silastic S-2) may increase TE by 10 15 dB/Hz throughout frequency range interest, an Overall Sound Pressure Level (OASPL) 2.89 dB. Whereas coating-2 Sylgard 184) shifted energy content lower reduced 2 4 600 1575 Additionally, it resulted 1.85 dB reduction OASPL, thus demonstrating choice coating materials viscoelastic plays crucial role... Read More

Source

Early detection of wind turbine faults using acoustic analysis. The method involves measuring and analyzing the sounds emitted by the turbine to detect components coming from abnormal areas. The steps are: measure turbine sounds, analyze spectrograms of the sounds over time, look for signals from abnormal areas synchronized with turbine rotation, and if threshold exceeded, determine turbine is abnormal.

Source

Optimizing wind farm noise reduction using AI to improve accuracy and efficiency compared to manual methods. The method involves using a noise reduction optimization model along with the wind turbine power generation model to find the optimal noise reduction solution that meets noise standards while maximizing power output. This is done by iteratively adjusting turbine operating parameters like blade pitch and tip speed ratio. The AI model learns the noise-power tradeoff relationship through training on wind farm data.

Source

Method to predict and control amplitude modulation (AM) noise generated by wind turbines, with the goal of mitigating AM noise levels perceived by people living near wind farms. The method involves: 1. Obtaining wind turbine data like conditions, sensor readings, and power output. 2. Predicting AM noise at a distance from the turbine using a machine learning model trained on historical turbine data. 3. Generating control data for the turbine based on the predicted AM noise. 4. Compare the predicted and measured AM noise to assess turbine contribution. 5. Retrain the model with measured noise if prediction accuracy is poor. 6. Store reduction data from control interventions to refine rankings.

Source

Predicting and mitigating noise generated by wind turbines using machine learning. The method involves predicting amplitude modulation (AM) noise, a low frequency pulsing sound, from wind turbine data using a machine learning model. This allows controlling the turbine operation to reduce AM noise at distances from the turbine. The AM noise prediction is based on factors like wind speed, direction, turbulence, etc. If measured AM exceeds the prediction, indicating nearby turbines contribute, control triggers like power reduction are set. If measured AM is lower, control is avoided. After intervention, AM reduction is monitored to refine predictions.

Source

Wind power equipment data acquisition system for monitoring and optimizing wind turbine performance. The system uses sensors on wind turbines to collect real-time data on parameters like wind speed, rotor speed, temperature, and vibration. This data is transmitted to a cloud for analysis. By comparing real-time and predicted performance, issues can be identified, efficiency improved, and downtime reduced. Sensors also monitor noise intensity to detect turbine anomalies. The system provides timely performance data for maintenance scheduling and prevents safety hazards.

Source

Adjusting wind turbine operation to meet noise limits in nearby sensitive areas without requiring additional on-site noise measurement devices. The method calculates the noise in the sensitive area based on wind turbine operating parameters and nearby rainfall. It then adjusts the turbine to keep the total noise below a threshold. This allows optimizing power generation while avoiding excessive noise in sensitive locations.

Source

Adaptive trailing edge comb on wind turbine blades to reduce noise emissions. The trailing edge comb has serrations that can be adjusted in geometry and orientation based on parameters like air density, wind speed, and load. This allows optimizing the comb shape for specific operating conditions to reduce noise compared to fixed comb designs.

Source

Method and arrangement for controlling a wind turbine to reduce noise while meeting local noise limits. The method involves estimating the turbine noise based on wind conditions and operational parameters. If the estimated noise exceeds a reference, the turbine is controlled to reduce noise. This can involve power reduction, blade pitch adjustment, or operating at a different power-speed curve. The noise estimation and control is implemented using an onboard arrangement with modules for noise estimation, noise reference management, and noise control.

Source

Diagnosing abnormal sounds in wind turbine yaw systems using acoustic analysis to monitor and detect faults. The method involves capturing acoustic signals from the yaw system components, extracting features from the signals, and using machine learning to classify the features as normal or abnormal. This allows early detection and diagnosis of yaw system faults through acoustic analysis instead of relying on manual inspection or vibration sensors.

Source

Accurately calculating wind farm noise levels taking into account terrain effects to improve prediction accuracy. The method involves modifying the wind turbine noise calculation model using terrain correction parameters for the specific site. This accounts for how terrain affects noise generation. The modified model is then used to calculate noise levels at predicted points in the site. Initial noise values are obtained. Then noise attenuation is calculated for factors beyond terrain. The final site-specific noise values are determined by subtracting attenuation from initial values.

Source

A wind farm noise and vibration reduction system that uses active noise control and passive noise control techniques to effectively reduce wind turbine blade noise and vibration. The system involves installing secondary sound sources on the wind turbine blades, collecting blade noise signals, and using adaptive filtering algorithms to generate control signals that drive the secondary sources to emit offsetting sound waves. Passive mufflers are also used to control mid- and high-frequency noise. This allows real-time adaptation to blade noise changes, improving noise reduction and energy savings without increasing blade resistance or impacting performance.

Source

Optimizing wind farm noise mitigation without reducing power output. The method involves using a noise propagation model that considers interactions between wind turbines and predicts noise levels at surrounding locations. To keep noise below limits, turbines are selectively operated based on optimization with constraints like max noise and total power. This reduces noise at critical points without excessive power loss.

Source

Method for controlling noise generated by a wind farm to reduce noise levels without significantly impacting power production. The method involves using a noise propagation model that takes into account wind farm layout, wind conditions, and turbine operating states to predict noise levels at nearby locations. When noise limits are exceeded, the method selects specific turbines to reduce power or stop based on the model to minimize overall farm power loss. This targeted approach allows meeting noise constraints without overly decreasing farm output.

Source

Computer-aided simulation method to accurately predict noise generated by wind turbine blades without the need for physical prototypes. The method involves using a digital representation of the blade geometry and importing it into a simulation software. The simulation calculates segmental airflow parameters, boundary layer transitions, and acoustic noise based on blade element momentum theory and viscous airfoil polar calculations. This allows predicting blade noise accurately without building physical prototypes.

Source

Method for accurately representing noise of a rotating wind turbine blade using computer simulations and blending techniques. The method involves importing a 3D blade geometry, extracting constructive parameters like airfoil profile and chord, and calculating low-order airflow using BEMT. This provides sectional angle of attack and velocity for scale-resolving CFD simulations. Noise spectra are computed at virtual microphones for each section. These are blended into a smooth variation over a rotor revolution. Inverse Fourier transforms generate synthetic signals with phase variations. Doppler correction and summation across revolutions complete the noise representation. Ground and atmospheric absorption are also applied.

Source

Controlling wind turbine noise levels without sacrificing power output by optimizing add-on device usage. The method involves setting a target noise level and controlling the turbine's add-on aerodynamic devices to stay below that level while maximizing power generation. It limits the number of devices allowed to extend at once based on the noise requirements. This prevents reducing turbine speed or output power to lower noise.

Source

Active noise control for wind turbines to reduce noise levels near residential areas without shutting down the turbines. The method involves monitoring the noise levels in the surrounding sensitive areas and selectively adjusting the wind turbine operation to mitigate the noise. This is done by comparing the sensitive area noise levels to a background level and choosing a noise reduction control scheme based on that difference. The chosen scheme, like varying generator speed, is then implemented to reduce the wind turbine noise. The goal is to effectively reduce wind turbine noise without resorting to shutdowns or other measures that impact power generation.

Source

A method to reduce noise and minimize power loss in wind turbines by optimizing fan operation based on a noise control model. The method involves acquiring sound pressure data, establishing a sound pressure distribution model, processing it to obtain a modified model, and creating optimization indices for noise and power. These indices are then optimized using an algorithm to find the best fan operation parameters that minimize noise and loss.

Source

Reducing noise emissions of a wind turbine by applying a perturbation signal to the optimal operating set point to increase temporal variation and prevent resonances from building up. The method involves receiving wind data, determining the optimal operating set point based on that data, and applying a perturbation signal to the set point to modify it. This modified set point with greater variation is then used to control the wind turbine and reduce noise emissions compared to using the original set point.

Source

Method, device, and system for predicting wind farm noise levels and optimizing wind turbine operations to mitigate noise pollution. The method involves calculating noise levels from individual turbines at real-time wind speeds, accounting for propagation losses, and summing to predict overall farm noise. By selectively adjusting turbine power instead of blanket shutdowns, it maximizes farm output while keeping total noise below limits.

Source

Analysis of noise emission from wind turbines to improve the reliability of the wind turbine. The analysis includes identifying tonal noise in a spectra of noise data, identifying the operating parameters of the wind turbine that are responsible for the generation of the tonal noise.

Source

Adaptive noise control method for wind farms that reduces noise emissions while maximizing power production compared to fixed noise limits. It involves dynamically adjusting wind turbine rotational speeds based on noise levels at surrounding locations. The method finds the turbine with the highest noise impact and reduces its speed. If the critical location's noise is below the limit, it increases speeds of turbines with lower noise impact. This balances noise reduction and power generation. The method uses a central park controller to calculate optimized rotor speeds recursively based on turbine noise models, conditions, and distances.

Source

Method to identify operating parameters and components of a wind turbine that promote the generation of tonal noise. The method involves analyzing noise data and operating parameter data from a wind turbine to determine if tonal noise is present and which operating parameters are correlated with it. By dividing the operating parameter space into intervals and analyzing only the relevant intervals, it's possible to accurately identify the operating parameters that cause tonal noise. This information can then be used to adjust turbine operation to avoid those parameter values and reduce tonal noise.

Source

Method to analyze and predict tonal noise emissions from wind turbines and control turbine operation to reduce or avoid tonal noise. The method involves identifying the vibration sensor whose data correlates with tonal noise in a specific region of interest. This sensor can then be used to predict and monitor tonal noise generation. By capturing vibration data from multiple sensors around the turbine, the method determines which sensor's vibrations best indicate tonal noise. This allows more accurate prediction and prevention compared to just using overall turbine vibration levels.

Source

Method for optimizing wind farm noise emissions while maximizing energy production. It involves dynamically adjusting wind turbine rotational speeds based on noise levels measured at critical locations. The method involves determining total noise levels at multiple locations, identifying the most critical location, reducing the speed of the turbine with highest noise-energy impact there, and increasing the speed of the turbine with lowest impact. This allows meeting noise limits at critical locations without significantly reducing overall energy production.

Source

Control method for wind turbines that optimizes power production while minimizing noise levels. The method involves calculating predicted operating trajectories of turbine parameters like blade pitch and rotor speed, and using those predictions to calculate noise metrics. The turbine is then controlled based on the predicted noise levels. This allows proactive noise reduction without sacrificing power output. The method can be applied to individual turbines or entire wind farms.

Source

Method to reduce noise from wind turbines without significantly reducing power output. It involves controlling the blade pivot angle of hinged blades to optimize noise and power tradeoff. The method involves biasing the blade toward the minimum pivot angle when wind speed is low, increasing rotor diameter. At high wind speeds, instead of reducing blade tip speed, the blade bias is reversed to limit blade tip speed. This prevents noise reduction by rotor speed decrease. The bias force is based on allowable noise level or corrosion risk. By independently controlling blade angle and tip speed, noise can be reduced without sacrificing power.

Source

Determining the sound value of wind turbine noise to accurately assess the audibility of wind turbine noise compared to background noise. The method involves collecting operating state and shutdown state noise data, screening the shutdown state data to remove noise sources, analyzing the screened data to determine tonal audibility, and correcting the operating state tonal audibility based on the shutdown state results to get the true audibility. This considers background noise influence and provides a more accurate sound value measurement.

Source

Predicting wind farm noise levels and optimizing power generation while meeting noise limits. The method involves calculating the noise level at a detection point by summing the individual turbine noise at collected wind speeds using lookup tables. If the total noise exceeds a threshold, it finds power levels to reduce noise without loss of overall farm output using an optimization algorithm. The turbines with the greatest impact on noise are targeted for power reductions.

Source

Online noise monitoring system and method for the transmission chain of a wind turbine nacelle to diagnose and maintain fan noise issues. The system involves placing a unique noise sensor inside the nacelle above the gearbox to collect noise signals during operation. The signals are processed and uploaded to a data system for analysis. This allows real-time monitoring of nacelle transmission chain noise levels and changes under different operating conditions. It provides accurate and efficient detection of fan noise issues without complex measurements or sensors in the surrounding environment.

Source

Monitoring tonal noise emissions from wind turbines using sensor data and noise notifications received from nearby communities. The method involves correlating sensor data from wind turbine operating parameters with noise notifications received from nearby communities to determine if the turbine is generating tonal noise. If so, the turbine's operating parameters can be adjusted to reduce or mask the tonal noise.

Source

Reducing tonal noise emissions from wind turbines by optimizing the transmission path of vibrations from the source to the radiating components. The method involves modeling the wind turbine components as a network of elements connected by acoustic paths. Vibrations from the source are applied at the interface and the transmission through each element determined. Modifications are made to elements along the vibration path to reduce distance to the source and radiating components. This can involve changing stiffness, mass, and damping.

Source

A wind power noise evaluation method based on cluster analysis and neural networks to realistically evaluate the noise levels of wind turbines in operation without needing long-term test databases. The method uses wind turbine main control system (SCADA) data along with wind turbine noise test results to achieve real-time evaluation of the noise radiation of operating wind turbines. It involves five stages: 1) wind turbine noise testing, 2) SCADA data collection and verification, 3) clustering analysis of SCADA data to group similar operating conditions, 4) training a neural network model using noise test data and clustered SCADA data, and 5) obtaining noise evaluation results using the trained model and clustered SCADA data. This allows real-time, efficient, and economical wind farm noise evaluation using existing SCADA data instead of extensive test databases.

Source

Method to predict tonal noise produced by wind turbines, especially from the drivetrain, by correlating vibration data with actual noise levels. The method involves acquiring vibration data from sensors around the drivetrain during testing, then acquiring noise data from the turbine. By identifying a sensor whose vibration correlates with the tonal noise in the noise data, relationships between vibration and tonal noise can be determined. This allows predicting tonal noise based on vibration levels during normal operation.

Source

A method for predicting noise distribution of wind farms with complex terrain that provides accurate and efficient noise prediction for wind farms in hilly or mountainous areas. The method involves using a terrain database and wind turbine specifications to precompute noise levels for different wind speeds and turbine orientations. This precomputed data is then interpolated to estimate noise levels for specific wind farm locations. This avoids the need for time-consuming numerical simulations on complex terrain.

Source

A method for designing low-noise wind turbine blades that balances power output and noise levels. The method involves optimizing chord length and twist distribution to increase power while reducing blade noise. The optimization uses a blade noise model, turbulence model, and load constraints. It calculates noise and power for each blade section, then iteratively adjusts chord and twist to maximize power-to-noise ratio.

Source

Method for predicting and optimizing noise levels in flat terrain wind farms to meet human and animal noise requirements. It involves calculating wind farm noise attenuation coefficients using a database without changing turbine models. This allows fast prediction of noise contours across the wind farm layout. It takes into account the influence of turbine wakes on noise distribution. This enables optimizing wind farm layout at different wind speeds and directions to mitigate noise impacts on people and wildlife.

Source

Cooperative active control of wind turbine noise and power generation to reduce noise levels in residential areas without shutting down turbines. The method involves monitoring noise and wind data near homes using sensors. When noise exceeds standards, turbines near the homes adjust speed and blade pitch to lower noise without shutting down. This allows optimal power generation while meeting noise limits. It uses real-time monitoring and intelligent control instead of blanket noise reduction methods.

Source

Method for controlling wind turbines to avoid tonal noise production by using vibration sensors placed at different locations on the turbine. The method involves identifying a vibration sensor whose data correlates with tonal noise in a region of interest. Wind turbine operating parameters are then adjusted based on the vibration level detected by that sensor to avoid or reduce tonal noise production. This leverages the fact that vibrations at certain frequencies can indicate potential tonal noise issues. By finding the vibration sensor that correlates with tonal noise, turbine operation can be adjusted to mitigate tonal noise before it becomes audible.

Source

Method to predict noise and optimize layout of flat terrain wind farms that addresses the issue of excessive wind farm noise impacting humans and animals. The method involves calculating a database of attenuation coefficients for wind farm noise propagation that doesn't change with wind turbine models. This allows fast prediction of wind farm noise distribution without needing detailed turbine simulations. The noise contour is used to optimize wind farm layout considering noise impacts for different wind speeds and directions. It enables meeting noise requirements for people and animals across the wind farm.

Source

Reducing noise of wind turbines during high wind speeds by controlling the nacelle yaw angle. When wind speed exceeds a threshold, the nacelle is yawed away from the nominal wind direction to increase the blade angle of attack. This reduces noise from the pressure side of the blade, mitigating aerodynamic noise caused by excessive pitching in high winds.

Source

A numerical method to accurately predict wind turbine noise using a hybrid aerodynamic-acoustic approach. The method involves solving the acoustic field around the wind turbine using the boundary element method (BEM) combined with a multi-zone BEM to handle singularities. It also uses a mesh segmentation technique called Principal Axes to optimize the BEM mesh generation. This allows calculating the noise at any point in the sound field using the virtual surface pressure of the wind turbine blades. The BEM with fast multipole acceleration is used to efficiently compute the sound field near the turbine wake.

Source

A method for predicting noise levels at wind farms that takes into account factors like wind turbine operating states, terrain, weather, and barriers to provide more accurate and realistic predictions compared to traditional models. The method involves modeling and forecasting wind farm noise using a noise model of individual wind turbines, considering factors like wind speed changes after the farm, terrain, climate, and turbine loading. This allows more directive and accurate real-time noise prediction for wind farms compared to just using turbine-level models.

Source

Monitoring acoustic noise generated by wind turbines in real-time to optimize performance and reduce noise. The system involves mounting microphones on the wind turbine nacelle to continuously monitor noise levels. Data is correlated with remote noise measurements using a transfer function. This correlated data is used to adjust the wind turbine operating parameters in real-time to minimize noise without shutting down unnecessarily. This allows longer turbine operation and higher power output within noise limits.

Source

Reducing noise from wind turbine blades by using a detachable trailing edge noise reduction device that can be controlled based on blade angle of attack. The device has teeth that protrude from the trailing edge and can rotate and adjust height. A control system calculates the optimal tooth rotation and spacing for each angle of attack to minimize trailing edge noise. This allows customization of the device to mitigate noise at different operating conditions.

Source

Optimizing wind turbine noise reduction by steering the wake flow to lower noise levels at nearby sensitive sites. The method involves estimating wake profiles based on weather conditions, estimating far-field sound propagation, and generating a yaw signal to steer the nacelle relative to the tower. If estimated noise exceeds a threshold, the yaw signal is adjusted to reduce noise. By aligning the wake flow with sound propagation, the turbine can operate at higher power and lower noise at sensitive sites.

Source