Acoustics Australia最新文献

The number of research papers dealing with vibration-based condition monitoring has been exponentially growing in recent decades. As a consequence, one may identify some trends that emerge from this vast literature. The present paper delineates a methodology that can be recognized in several research works, which is rooted in a succession of three stages. The first stage embodies a linear transform of the data, typically in the form of a filterbank, the second stage reduces the dimension of the data through a nonlinear functional, typically in the form of health indicators, and the last stage supplies a statistical decision. Although several variants of this methodology exist, its fundamental principles seem to have converged to a general consensus, at least implicitly. This paper provides a critical overview of this methodology. It discusses its working assumptions under some typical scenarios and formulates several caveats. It also provides a few prospects that may nourish future research.

This paper discusses trends in condition monitoring of modern offshore wind turbines. First an overview is given of design changes that have been made over the years to large offshore wind turbines and how this resulted in novel opportunities from a monitoring perspective. Similarly, the evolution in data source availability is discussed. From these opportunities, some ongoing research activities in the field are discussed and how they fit with the open challenges. This list is far from exhaustive. It gives an overview of some capita selecta. Particularly, the fields of advanced signal processing and requirement for innovations towards prognostic frameworks are highlighted.

This paper provides an overview of the recent advances made in the field of fault diagnosis of industrial machines operating under variable speed conditions. First, the shortcomings of the traditional techniques in extracting reliable fault information are laid down, followed by a discussion on the different approaches adopted to overcome these issues. Next, these approaches are discussed by categorizing them as resampling based and resampling free methods. The principle and implementation procedures of these methods are discussed by summarizing the key literature in this area. Finally, the paper is concluded by highlighting the future challenges to address in this area.

With the huge success of applying deep learning (DL) methodologies to image recognition and natural language processing in recent years, researchers are now keen to use them in the machine condition monitoring (MCM) context. There are numerous papers in applying various DL techniques, such as auto-encoder, restricted Boltzmann machine, convolutional neural network and recurrent neural network, to MCM problems ranging from component level condition monitoring (machine tool wear prediction, bearing fault diagnosis and classification and hydraulic pump fault diagnosis) to system level health management (aircraft and spacecraft diagnosis). In this paper, we give a brief overview in the area of DL for MCM with a focus on reviewing the most recent papers published since 2019. In Part 1, we present some critical views regarding whether any breakthrough has been achieved from an MCM domain expert perspective, with the main conclusion that DL has great potential for MCM applications and a major breakthrough could come soon since the shortfalls lie more in data than in the DL methodologies. Our overall impression is that (a) DL models are not really showing their great potentials with only a small training data; (b) faulty-condition data is hard to come by for training DL, but normal condition data is abundant, so anomaly detection makes more sense; (c) applying DL only to the Case Western Reserve University (CWRU) bearing fault dataset is not sufficient for real-world industrial applications as it was from a very simple test rig, and applying DL to data from complex systems like helicopter gearbox data may deliver much more convincing results. In Part 2, we enhance the main conclusion of the critical review with supplement views and a case study on analyzing Bell-206B helicopter main gearbox planet bearing failure data using some traditional MCM techniques in contrast to applying the long short-term memory (LSTM) DL method. We can conclude from the case study that the DL-based methods are not necessarily always superior to the traditional MCM techniques for dataset from moderately complex machinery.

A simplified finite element method (FEM) simulation method has been established and validated for analyzing the sound absorption mechanism of structures with periodic axisymmetric cavities. Combined with genetic algorithm, the simplified FEM method is used to optimize the sound absorption coefficient of the structure containing periodic cylindrical cavities and variable cross section cavities. The result of variable section cavities is much better than the case of cylindrical cavities. The effect of cavity shape on sound absorption mechanism is discussed through energy dissipation, structure deformation and modal analysis of the absorption structures. It is found that the cavity structure resonances include bending vibration of the surface layer and radial motion of particles near the cavities. The radial motion also changes along the axial direction. Adding geometric design parameters of the cavity cross section are conducive to moving the radial mode to low frequency. The radial vibration has a great influence on absorption performance, which is more conducive to promoting the conversion of longitudinal waves into transverse waves with more energy dissipation. Finally, a better sound absorption performance is obtained by introducing the material parameter of Young's modulus into the optimization model, indicating that comprehensive consideration of geometry and material parameters for optimization is expected to obtain the desired sound absorption structure in engineering practice.

With the huge success of applying deep learning (DL) methodologies to image recognition and natural language processing in recent years, researchers are now keen to use them in the machine condition monitoring (MCM) context. There are numerous papers in applying various DL techniques, such as auto-encoder, restricted Boltzmann machine, convolutional neural network and recurrent neural network, etc., to MCM problems ranging from component-level condition monitoring (machine tool wear prediction, bearing fault diagnosis and classification and hydraulic pump fault diagnosis) to system-level health management (aircraft and spacecraft diagnosis). In this paper, we give a brief overview in the area of DL for MCM with a focus on reviewing the most recent papers published since 2019. In Part 1, we present some critical views regarding whether any breakthrough has been achieved from an MCM domain expert perspective, with the main conclusion that DL has great potential for MCM applications, and a major breakthrough could come soon since the shortfalls lie more in data than in the DL methodologies. Our overall impression is that (a) DL models are not really showing their great potentials with only a small training data; (b) faulty-condition data is hard to come by for training DL, but normal condition data is abundant, so anomaly detection makes more sense; (c) applying DL only to the Case Western Reserve University (CWRU) bearing fault dataset is not sufficient for real world industrial applications as it was from a very simple test rig, and applying DL to data from complex systems like helicopter gearbox data may deliver much more convincing results. In Part 2, we enhance the main conclusion of the critical review with supplement views and a case study on analysing Bell-206B helicopter main gearbox planet bearing failure data using some traditional MCM techniques in contrast to applying the long short-term memory (LSTM) DL method. We can conclude from the case study that the DL-based methods are not necessarily always superior to the traditional MCM techniques for dataset from moderately complex machinery.

Helmholtz resonator is considered and widely accepted as a basic acoustic model in engineering applications and research. In this paper, the normal incidence sound absorption characteristics of series and parallel configurations of Helmholtz resonators is studied analytically, numerically and experimentally. The proposed analytical model for series configuration of HRs comprises of Johnson–Champoux–Allard model and transfer matrix method while parallel configuration of HRs is described using parallel transfer matrix method. The results from proposed analytical models fit well with the finite element method (FEM) results obtained from COMSOL multiphysics. Incorporation of parallel configuration and proper tuning of geometric parameters helps to overcome the trade-off between broad band sound absorption and minimum space utilization. Also, the experimental observations of one of the parallel configuration substantiates the FEM results. Moreover, the FEM models are more accountable for the variation in neck position and also provide better visualization of acoustic absorption with frequency.

To expand the spatial and temporal scales of passive acoustic monitoring of animals, automatically detecting target sounds among noises with similar acoustic properties is essential but challenging. In particular, the classification of tonal vocalisations and tonal noise remains a universal problem in bioacoustics research. The vocalisations of dugong, which is an endangered marine mammal that inhabits coastal seas, need to be monitored to enhance our understanding of its habitat use. However, detecting dugong tonal vocalisations is difficult due to the presence of tonal noise in the same frequency band. In this study, a classification method was developed for these signals to handle large acoustic data by reducing the labour required for manual inspection. Mel-frequency cepstral coefficients (MFCC) were extracted to characterise background sounds along with a few parameters of the signal contour, and a support vector machine was trained for binary classification. The classifier achieved an 84.4% recall and a 93.5% precision on the testing dataset even in a noisy shallow marine environment. This methodology enables the effective classification of dugong calls and similar tonal noises by combining contour and MFCC features and can extend the spatial and temporal scale of acoustic monitoring of the endangered dugong. This technique is potentially applicable to the monitoring of other endangered marine mammals that produce tonal vocalisations.

The cascade’s response function can be used to effectively deal with the unsteady response of the interaction between the harmonic turbulence and the cascades. Based on this function, this paper presents the formulas for the stator broadband and tonal noises, whose inflow models are different from each other. The broadband noise comes from the impact of the random turbulence wave in the rotor wake on the stator, while the tonal noise comes from the interaction between the periodic rotor wake and the stator. According to the formulas for predicating the two kinds of noise, their inflow models are different from each other. Comparing with the test models of subsonic fan, the prediction models of the broadband and tonal noise prove to be correct. Meanwhile, the influence of the blade number on stator tonal and broadband noise is carried out through the prediction models, and the results are summarized that (1) the greater the number of blades, the higher the broadband sound power level of the stator in the high frequency. (2) An increase in the number of blades can “cut off” the tonal noise of the stator at BPF. So, reasonable arrangement of the number of stator and rotor blades is significant to the passive suppression of stator noise.

Long-term data of underwater passive acoustic monitoring (PAM) collected from two sites of pygmy blue whale presence within Australia, the continental shelf off Portland (38.5° S, 141.2° E) and the Perth Canyon (32° S, 115° E) were analysed to compare the acoustic behaviour of eastern Indian Ocean pygmy blue (EIOPB) whales. Pygmy blue whale song detection was consistently higher at the Perth Canyon site than at the Portland sample site. Statistical analysis found there to be a significant difference in the production of song and phrase variants between sites (p < 0.01) with a shorter two-unit (P2) song variant being more common in the Perth Canyon area, while the traditional three-unit (P3) song variant was more frequent off Portland. This was supported by manual and feature space analysis techniques. Increasing song complexity was observed in the form of phrases with broken song units, a phenomenon that was first observed at the Portland site on isolated occasions but has occurred and proliferated in the Perth Canyon area from 2016 onwards. Analysis of environmental conditions indicated that increased background noise due to multiple EIOPB whales vocalising, as well as water depth, may influence song length. This was reflected by songs made up of shorter phrases dominating in higher background noise conditions and deeper water, while longer more complex phrase types dominate in quieter, shallower conditions. Further research is recommended to isolate any potential influence of environmental factors on song production.