Sound Prediction Research Articles

Research on sound quality of roller chain transmission system based on multi-source transfer learning

To establish the sound quality evaluation model of roller chain transmission system, we collect the running noise under different working conditions. After the noise samples are preprocessed, a group of experienced testers are organized to evaluate them subjectively. Mel frequency cepstral coefficient (MFCC) of each noise sample is calculated, and the MFCC feature map is used as an objective evaluation. Combining with the subjective and objective evaluation results of the roller chain system noise, we can get the original dataset of its sound quality research. However, the number of high-quality noise samples is relatively small. Based on the sound quality research of various chain transmission systems, a novel method called multi-source transfer learning convolutional neural network (MSTL-CNN) is proposed. By transferring knowledge from multiple source tasks to target task, the difficulty of small sample sound quality prediction is solved. Compared with the problem that single source task transfer learning has too much error on some samples, MSTL-CNN can give full play to the advantages of all transfer learning models. The results also show that the MSTL-CNN proposed in this paper is significantly better than the traditional sound quality evaluation methods.

Aeroacoustic Source Potential Based on Poisson’s Equation

Poisson’s equation is an important equation to postprocess the aerodynamic fields into linearized momentum modes and was recently found to be important for the computation of an isotropic pressure-like source for scalar aeroacoustic wave models, like the aeroacoustic wave equation based on Pierce’s operator (AWE-PO). Mathematically viable boundary conditions of the Poisson equation, which computes the AWE-PO source, are investigated. For the different source fields, the wave propagation is computed using the AWE-PO, and the details of the sound prediction results are compared to a reference direct numerical simulation of a mixing layer. The different boundary conditions of the Poisson equation were found to have a minor influence on the overall sound prediction characteristics of the AWE-PO equation. The AWE-PO is reformulated into a simplified version of the Phillips’s equation, which mitigates the intermediate step of computing an isotropic source potential. By doing so, a previously obtained interference radiation valley in the radiated acoustic intensity of the AWE-PO results is attributed to a missing shear-noise source term.

A Survey on Prediction of Heart Functionality Sound by using Deep Learning Algorithm

The primary objective of this study is to enhance the precision of predicting cardiovascular illnesses by introducing a sophisticated system that leverages deep learning techniques. Traditional diagnostic methods have relied on the analysis of cardiac sounds, achieving an accuracy of approximately 87.5% through the use of machine learning algorithms such as Random Forest and Decision Trees. However, these conventional approaches have their limitations. In contrast, the proposed hybrid approach aims to surpass these limitations by focusing on capturing intricate patterns and temporal relationships within heart sound data using deep learning. This innovative approach seeks to demonstrate a substantial improvement in prediction accuracy when compared to existing methods. If proven effective, this deep learning-based diagnostic tool has the potential to provide a more nuanced understanding of heart function, enabling early identification of anomalies and ultimately leading to improved patient outcomes in cardiovascular health.

Small Sample Sound Quality Prediction Method of Hy-Vo Chain Transmission System Based on Fuzzy Generation

To improve the noise comfort of the whole machine, it is necessary to establish the sound quality prediction model of the Hy-Vo chain transmission system. Compared with the silent chain transmission system, the Hy-Vo chain transmission system normally operates at a lower speed and cannot have too much load at the limit speed. It is difficult to obtain a sufficient quantity of high-quality noise samples because there are few different working conditions. For small sample sound quality prediction, we use a sample enhancement method called fuzzy generation based on fuzzy mathematics. Firstly, audio samples of the Hy-Vo chain transmission system are collected through noise tests. Secondly, the processed samples are evaluated objectively and subjectively. After a correlation test of the subjective evaluation results, correct subjective evaluation scores of each noise sample are obtained. With the help of fuzzy generation, we can obtain a sufficient number of new samples. By mixing the original samples with the generated samples, a new dataset is created. Through using a general regression neural network (GRNN), support vector regression (SVR) model, and ridge regression (RR) method, the sound quality of the Hy-Vo chain transmission system can be predicted. Different from prediction results under the original dataset, using the fuzzy generation method can not only significantly reduce the prediction error of the model but also improve stability.

Mathematical modeling of the creation process and aging of polymer composite materials

Based on the analysis of the literature on the possibility of using neural networks to create new materials with high functional properties, the article considers a solution to the problem of determining the operational stability of polymeric composite materials by creating physical and chemically sound mathematical prediction models. Epoxy resins of the UP-637 and EA brands with an isophorone diamine hardener were chosen as the matrix of the model composite material, and oligobutadiene rubber of the SKN-10 KTR brand was chosen as the modifier. It justifies directions of work necessary for development of new materials creation methodology with optimal characteristics, building a model for changing the properties of materials at variation of composition and implementation of full-scale mathematical modeling of physical and chemical processes of polymer composite materials aging at changing level and time of climatic factors influence. Verification of the obtained dependence of service characteristics on the composition of the material and the level of influencing climatic factors was carried out on the basis of data from full-scale tests in a temperate climate. The proposed methodology for modelling the properties of polymer composite materials will reduce the development time of new materials and allow creation of polymer composites based on epoxy resin containing fillers of various natures (carbon, mineral and polymer) with high performance parameters.

Revisiting Tensor Basis Neural Network for Reynolds stress modeling: Application to plane channel and square duct flows

Recent advances in computing power devoted to Computational Fluid Dynamics (CFD) have opened up a new avenue for addressing the turbulence closure problem through Machine Learning (ML). Several Tensor Basis Neural Network (TBNN) frameworks aimed at enhancing turbulence Reynolds-Averaged Navier–Stokes (RANS) modeling have recently been proposed in the literature as data-driven constitutive models for systems with known invariance properties. However, persistent ambiguities remain regarding the physical adequacy of applying the General Eddy Viscosity Model (GEVM) within these frameworks. This work aims at investigating this aspect in an a priori stage for better predictions of the Reynolds stress anisotropy tensor, while preserving the Galilean and rotational invariances. In particular, we propose a general framework providing optimal tensor basis models for two types of canonical flows with increasing complexity: Plane Channel Flow (PCF) and Square Duct Flow (SDF). Subsequently, deep neural networks based on these optimal models are trained using state-of-the-art strategies to achieve a balanced and physically sound prediction of the full anisotropy tensor. A priori results obtained by the proposed framework are in very good agreement with the reference Direct Numerical Simulations (DNS) data. Notably, our shallow network with three layers excels in providing accurate predictions of the anisotropy tensor for PCF at unobserved friction Reynolds numbers, both in interpolation and extrapolation scenarios. The learning of the SDF case is more challenging because of its physical nature but also due to a lack of training data at various regimes. We propose a numerical strategy to alleviate this problem based on Transfer Learning (TL). In order to more efficiently generalize to an unseen intermediate Reτ regime, we take advantage of our prior knowledge acquired from a training with a larger and wider dataset. Our results indicate the potential of the developed network model, and demonstrate the feasibility and efficiency of the TL process in terms of training data size and training time. Based on these results, we believe there is a promising future by integrating these neural networks into an adapted in-house RANS solver.

Granular activated carbon sound absorption predictions made using measured material parameters

Granular activated carbon (GAC) is hierarchical porosity material that yields better in low frequency sound absorption performance compared with many traditional porous materials due to the sorption effect created by nanometer-scale pores within the particles. In 2017, a triple porosity model accounting for the sorption effect in the micropores was proposed by Venegas et al. In the proposed model, some of input parameters (i.e., mesopore size and micropore effective diffusion coefficient) were fitted by matching the model to measured the surface impedance of a GAC stack; however, some fitted values are questionable from an inorganic material perspective. In the present work, the GAC material parameters, such as micro- and meso-pore sizes, were estimated from a standard isotherm measurement. The measurement results showed a standard Langmuir type isotherm behavior at different temperatures: i.e., 273.15 and 293.15 K. Thus, if the isotherms can be described with the Langmuir model, then the Langmuir constant and heat of adsorption can be estimated based on the isotherms. Finally, these parameters were used as the input to the GAC model and the absorption coefficient was calculated. The calculated result was also compared with the absorption coefficient measured following the E1050 standard.

Predicting ocean pressure field with a physics-informed neural network.

Ocean sound pressure field prediction, based on partially measured pressure magnitudes at different range-depths, is presented. Our proposed machine learning strategy employs a trained neural network with range-depth as input and outputs complex acoustic pressure at the location. We utilize a physics-informed neural network (PINN), fitting sampled data while considering the additional information provided by the partial differential equation (PDE) governing the ocean sound pressure field. In vast ocean environments with kilometer-scale ranges, pressure fields exhibit rapidly fluctuating phases, even at frequencies below 100 Hz, posing a challenge for neural networks to converge to accurate solutions. To address this, we utilize the envelope function from the parabolic-equation technique, fundamental in ocean sound propagation modeling. The envelope function shows slower variations across ranges, enabling PINNs to predict sound pressure in an ocean waveguide more effectively. Additional PDE information allows PINNs to capture PDE solutions even with a limited amount of training data, distinguishing them from purely data-driven machine learning approaches that require extensive datasets. Our approach is validated through simulations and using data from the SWellEx-96 experiment.

Development of an acoustic design support tool for HVAC ventilation units

HVAC equipment is one of the main sources of noise inside or outside a building. Furthermore, many people are exposed daily to HVAC noise, which can lead to health-related troubles. Therefore, an HVAC mechanical unit must be selected to provide an acceptable sound level transmitted to the occupied spaces of a building and not disturb the community. This paper presents a prediction tool named Sound Prediction Tool (SPT) developed by MJM Acoustical Consultants Inc. in 2021 that allows the engineer (i) to design an HVAC unit with specific mechanical components (filter, coil, etc.) according to his input parameters and (ii) to predict and optimize the sound levels produced by the selected design through its casing and openings (inlet and outlet air). This tool has been validated firstly with acoustical tests ‘’in situ’’ on several separate Air Handling Units (AHU) manufactured by our customer and secondly with Finite Elements Method (FEM) models using Comsol-Multiphysics. Several examples are presented to demonstrate the usefulness of the proposed tool.

A Method for Sound Speed Profile Prediction Based on CNN-BiLSTM-Attention Network

In response to the current challenges in efficiently acquiring sound speed profiles and ensuring their representativeness, considering the need to fully leverage historical sound speed profiles while accounting for their spatiotemporal variability, we introduce a model for sound speed profile prediction based on a CNN-BiLSTM-Attention network, which integrates a convolutional neural network (CNN), a bidirectional long short-term memory network (BiLSTM), and an attention mechanism (AM). The synergy of these components enables the model to extract the spatiotemporal features of sound speed profiles more comprehensively. Utilizing the global ocean Argo grid dataset, the model predicted the sound speed profiles of an experimental zone in the Western Pacific Ocean. In predicting sound speed profiles of a single point, the model achieved a root mean square error (RMSE), relative error (RE), and accuracy (ACC) of 0.72 m/s, 0.029%, and 0.99971, respectively, surpassing comparative models. For regional sound speed profile prediction, the mean RMSE, RE, and ACC of different water layers were 0.919 m/s, −0.016%, and 0.9995, respectively. The experimental outcomes not only confirm the high accuracy of the model, but also highlight its superiority in sound speed profile prediction, particularly as an effective compensatory approach when profile measurements are untimely or contain representational errors.

Prediction of Water Temperature Based on Graph Neural Network in a Small-Scale Observation via Coastal Acoustic Tomography

Coastal acoustic tomography (CAT) is a remote sensing technique that utilizes acoustic methodologies to measure the dynamic characteristics of the ocean in expansive marine domains. This approach leverages the speed of sound propagation to derive vital ocean parameters such as temperature and salinity by inversely estimating the acoustic ray speed during its traversal through the aquatic medium. Concurrently, analyzing the speed of different acoustic waves in their round-trip propagation enables the inverse estimation of dynamic hydrographic features, including flow velocity and directional attributes. An accurate forecasting of inversion answers in CAT rapidly contributes to a comprehensive analysis of the evolving ocean environment and its inherent characteristics. Graph neural network (GNN) is a new network architecture with strong spatial modeling and extraordinary generalization. We proposed a novel method: employing GraphSAGE to predict inversion answers in OAT, using experimental datasets collected at the Huangcai Reservoir for prediction. The results show an average error 0.01% for sound speed prediction and 0.29% for temperature predictions along each station pairwise. This adequately fulfills the real-time and exigent requirements for practical deployment.

Drug solubility in biorelevant media in the context of an inhalation-based biopharmaceutics classification system (iBCS)

An inhalation-based Biopharmaceutics Classification System for pulmonary drugs (iBCS) holds the perspective to allow for scientifically sound prediction of differences in the in vivo performance of orally inhaled drug products (OIDPs).A set of nine drug substances were selected, that are administered via both the oral and pulmonary routes. Their solubility was determined in media representative for the oral (Fasted State Simulated Intestinal Fluid (FaSSIF)) and pulmonary (Alveofact medium and Simulated Lung Fluid (SLF)) routes of administration to confirm the need for a novel approach for inhaled drugs. The complexity of these media was then stepwise reduced with the purpose of understanding the contribution of their components to the solubilizing capacity of the media. A second reason for varying the complexity was to identify a medium that would allow robust but accurate dissolution testing. Hence, Hank’s balanced salt solution (HBSS) as a medium used in many in vitro biological tests, non-buffered saline solution, and water were included.For some drug substances (salbutamol sulfate, tobramycin, isoniazid, and tiotropium bromide), no significant differences were observed between the solubility in the media used. For other drugs, however, we observed either just small (rifampicin, budesonide, salmeterol) or unexpectedly large differences (beclomethasone dipropionate).Based on the minimum theoretical solubility required for their common pulmonary dose in 10 ml of lung lining fluid, drug solubility was classified as either high or low. Two high solubility and two low solubility compounds were then selected for refined solubility testing in pulmonary relevant media by varying their content of phospholipids, surfactant proteins and other proteins.The solubility of drug substances in simulated lung lining fluids was found to be dependent on the physicochemical properties of the drug substance and the composition of the media. While a pulmonary dissolution medium that would fit all drugs could not be established, our approach may provide guidance for finding the most suitable dissolution medium for a given drug substance and better designing in vitro tests for predicting the in vivo performance of inhalable drug products.

Interaction network structure explains species' temporal persistence in empirical plant-pollinator communities.

Despite clear evidence that some pollinator populations are declining, our ability to predict pollinator communities prone to collapse or species at risk of local extinction is remarkably poor. Here, we develop a model grounded in the structuralist approach that allows us to draw sound predictions regarding the temporal persistence of species in mutualistic networks. Using high-resolution data from a six-year study following 12 independent plant-pollinator communities, we confirm that pollinator species with more persistent populations in the field are theoretically predicted to tolerate a larger range of environmental changes. Persistent communities are not necessarily more diverse, but are generally located in larger habitat patches, and present a distinctive combination of generalist and specialist species resulting in a more nested structure, as predicted by previous theoretical work. Hence, pollinator interactions directly inform about their ability to persist, opening the door to use theoretically informed models to predict species' fate within the ongoing global change.

Effect of sound source predicted by large eddy simulation on aerodynamic sound prediction of a box fan

Effect of sound source predicted by large eddy simulation on aerodynamic sound prediction of a box fan

Mathematical modelling of Alzheimer's disease biomarkers: Targeting Amyloid beta, Tau protein, Apolipoprotein E and Apoptotic pathways.

Introduction: The kinetics of brain cell death in Alzheimer's disease (AD) is being studied using mathematical models. These mathematical models utilize techniques like differential equations, stochastic processes, and network theory to explore crucial signalling pathways and interactions between different cell types. One crucial area of research is the intentional cell death known as apoptosis, which is crucial for the nervous system. The main purpose behind the mathematical modelling of this is for identification of which biomarkers and pathways are most influential in the progression of AD. In addition, we can also predict the natural history of the disease, by which we can make early diagnosis. Mathematical modelling of AD: Current mathematical models include the Apolipoprotein E (APOE) Gene Model, the Tau Protein Kinetics Model, and the Amyloid Beta Peptide Kinetic Model. The Bcl-2 and Bax apoptosis theories postulate that the balance of pro- and anti-apoptotic proteins in cells determines whether a cell experiences apoptosis, where the Bcl-2 model, depicts the interaction of pro- and anti-apoptotic proteins, it is also being used in research on cell death in a range of cell types, including neurons and glial cells. How peptides are produced and eliminated in the brain is explained by the Amyloid beta Peptide (Aβ) Kinetics Model. The tau protein kinetics model focuses on production, aggregation, and clearance of tau protein processes, which are hypothesized to be involved in AD. The APOE gene model investigates the connection between the risk of Alzheimer's disease and the APOE gene. These models have been used to predict how Alzheimer's disease would develop and to evaluate how different inhibitors will affect the illness's course. Conclusion: These mathematical models reflect physiological meaningful characteristics and demonstrates robust fits to training data. Incorporating biomarkers like Aβ, Tau, APOE and markers of neuronal loss and cognitive impairment can generate sound predictions of biomarker trajectories over time in Alzheimer's disease.

Development of sound field simulation and optimization over large acoustic space by use of novel SEA method

Panasonic is committed to providing customers with optimal PA system. The ability to predict PA system performance from digital mock-up is key to help designing optimal system under various conditions. Sound predictions over large spaces are challenging with FEM/BEM methods due to high computational cost and response sensitivity. Additionally, adding detail in source or boundary is difficult with Ray-tracing methods. Finally, traditional SEA methods don't provide sufficient resolution of response. We have investigated a novel SEA method with ability to predict spatial variations in acoustic response. Sound propagation in large public spaces over the full audio range was predicted, accounting for detailed loudspeaker source characteristics, room geometry and boundaries absorption. Predictions were validated against test in a Panasonic auditorium. Method was then used to predict PA system performance in public space under realistic noise conditions. Inverse method was used to synthesize background noise source from limited test data. STI predictions were then performed and optimization carried out to identify PA system loudness requirements in order to achieve best STI performance over wide area while minimizing signal loudness at any position for various background noise conditions.

Engineering process for eVTOLs interior sound predictions

Urban Air Mobility (UAM) is quickly developing with the objective to transport passengers and cargos at lower altitudes in urban areas using automated aircrafts. Noise is one of the important challenges facing engineers developing these aircrafts as they will be operating in cities and are thus subjected to noise regulations. In addition, customers will be expecting similar interior noise performance as in luxury automotive vehicles. This paper presents an engineering process developed to predict the acoustic performance inside Electric Vertical Take-off and Landing (eVTOL) aircrafts. The process starts with Computational Fluid Dynamics (CFD) simulations to calculate the noise sources generated by the propellers. A Boundary Element Model (BEM) is then used to propagate these sources around the aircraft and calculate the pressure distribution on the fuselage exterior, which is then applied, in the last step, as excitation in a Statistical Energy Analysis (SEA) model to predict the interior noise. The paper uses a demonstrator model to illustrate the process and results. This process is now available to support engineers when optimizing their eVTOL design for costs, weight and interior noise performance without the need for physical prototypes.

Adding attention to the neural ordinary differential equation for spatio-temporal prediction

Explainable spatio-temporal prediction gains attraction in the development of geospatial artificial intelligence. The neural ordinal differential equation (NODE) emerges as a new solution for explainable spatio-temporal prediction. However, challenges still need to be solved in most existing NODE-based prediction models, such as difficulty modeling spatial data and mining long-term temporal dependencies in data. In this study, we propose a spatio-temporal attentional NODE (STA-ODE) to address the two challenges above. First, we define a spatio-temporal ordinary differential equation to predict a value at each time iteratively by a novel spatio-temporal derivative network. Second, we develop an attention mechanism to fuse multiple prediction values for capturing long-term temporal dependencies in data. To train the STA-ODE model, we design a loss function that aligns the prediction results in spatial dimension with prediction results in temporal dimension to calibrate the parameters of the model. The proposed model was validated with three real-world spatio-temporal datasets (traffic flow dataset, PM2.5 monitoring dataset, and temperature monitoring dataset). Experimental results showed that STA-ODE outperformed seven existing baselines regarding prediction accuracy. In addition, we used visualization to demonstrate the sound interpretability and prediction accuracy of the STA-ODE model.

Impact sound insulation prediction and uncertainty quantification using a detailed source model in the modal Transfer Matrix Method

The main goal of this work is to provide an efficient and accurate prediction of the impact sound insulation of layered structures for the entire building acoustics frequency range (50-5000 Hz). To this end, the modal Transfer Matrix Method (mTMM) has been developed to predict the impact sound radiation from layered floors resulting from tapping machine impacts, while accounting for simply supported boundary conditions. A detailed source model is presented which accounts for both the impact locations of the individual hammers and the discrete 2 Hz force spectrum. It is also discussed how this detailed source model can be implemented into the mTMM. Monte Carlo simulations provide insight into the uncertainty of the radiated sound power and the impact sound insulation improvement for bare and layered floors, resulting from the impact force location. Numerical validation is provided by comparison to finite element models of various floors. Narrow-band sound pressure level measurements on a layered floor have been performed to validate the detailed source model and its use in the mTMM. A comparison of measured and predicted results indicates the importance of a detailed source approach, especially at low frequencies

Uncertainty quantification of the diffuse sound field assumption in sound insulation predictions

The transmission loss of a partition depends on the sound fields in the adjoining rooms, which are often taken to be diffuse. A diffuse sound field is a random field, representing a conceptual ensemble of rooms with the same volume and reverberation time, but otherwise any possible arrangement of boundaries and small objects that have a wave scattering effect. Uncertainty due to random wave scattering is therefore inherently present in diffuse transmission loss values. In this work, closed-form expressions are presented for quantifying this uncertainty, such that it becomes possible to estimate by how much the transmission loss of a particular room-wall-room system may deviate from the nominal, ensemble average value which corresponds to omnidirectional incidence and halfspace radiation. The first expression is exact yet its evaluation requires knowledge of the dynamic stiffness of the partitioning structure and of the radiation impedances of the surrounding fluids. The second expression is approximate and depends on energetic quantities only, so it is also applicable in an experimental setting. The expressions are validated in a Monte-Carlo simulation study involving a single wall.