Short-time Fourier Transform Research Articles

Assessing seismicity in Bangladesh: an application of Guttenberg-Richter relationship and spectral analysis

Bangladesh has a high risk of earthquakes because the Dauki, Jamuna, and Chittagong-Myanmar faults are still active. However, the assessment of seismicity remains a big challenge due to the complex geologic setting of Bangladesh. This study employed the Guttenberg-Richter relationship and the spectral models to assess and analyze the earthquake conditions in Bangladesh. Besides, an instrumental earthquake catalogue, obtained from the Bangladesh Meteorological Department (BMD), covering 1985–2017, is established. The results revealed that the Guttenberg-Richter constants of a and b were 2.981 and 0.392, which propagated a strain release from 1992 to 2017. The spectral model analyses, e.g. wavelet transform (WT), short-time Fourier transformation (STFT), and multitaper model (MTM), demonstrated the magnitude and strain release anomalies of the same magnitude ranging from 4.8 to 5.7, indicating the probable precursor of an upcoming earthquake. Notably, magnitudes have been running around 4.5–5.8, which may act as a signal to major earthquakes that have not been evident before. The proposed models allowed for the completion of the Bangladesh earthquake catalogue and provided a platform for future seismicity assessment and earthquake probability analysis. These results should be considered in determining how likely earthquakes are to happen in an area or region.

Detection of valvular heart diseases combining orthogonal non-negative matrix factorization and convolutional neural networks in PCG signals

Background and objective:Valvular heart disease (VHD) is associated with elevated mortality rates. Although transthoracic echocardiography (TTE) is the gold standard detection tool, phonocardiography (PCG) could be an alternative as it is a cost-effective and noninvasive method for cardiac auscultation. Many researchers have dedicated their efforts to improving the decision-making process and developing robust and precise approaches to assist physicians in providing reliable diagnoses of VHD. Methods:This research proposes a novel approach for the detection of anomalous valvular heart sounds from PCG signals. The proposed approach combines orthogonal non-negative matrix factorization (ONMF) and convolutional neural network (CNN) architectures in a three-stage cascade. The aim of the proposal is to improve the learning process by identifying the optimal ONMF temporal or spectral patterns for accurate detection. In the first stage, the time–frequency representation of the input PCG signal is computed. Next, band-pass filtering is performed to locate the spectral range that is most relevant for the presence of such cardiac abnormalities. In the second stage, the temporal and spectral cardiac structures are extracted using the ONMF approach. These structures are utilized in the third stage and fed into the CNN architecture to detect abnormal heart sounds. Results:Several state-of-the-art CNN architectures, such as LeNet5, AlexNet, ResNet50, VGG16 and GoogLeNet, have been evaluated to determine the effectiveness of using ONMF temporal features for VHD detection. The results reveal that the integration of ONMF temporal features with a CNN classifier significantly improve VHD detection. Specifically, the proposed approach achieves an accuracy improvement of approximately 45% when ONMF spectral features are used and 35% when time–frequency features from the short-time Fourier transform (STFT) spectrogram are used. Additionally, feeding ONMF temporal features into low-complexity CNN architectures yields competitive results comparable to those obtained with complex architectures. Conclusions:The temporal structure factorized by ONMF plays a critical role in distinguishing between normal heart sounds and abnormal heart sounds since the repeatability of normal heart cycles is disrupted by the presence of cardiac abnormalities. Consequently, the results highlight the importance of appropriate input data representation in the learning process of CNN models in the biomedical field of valvular heart sound detection.

INDUCED DRAFT FAN DOMINANT FREQUENCY DETECTION USING SHORT-TIME FOURIER TRANSFORM METHOD

Weak suction and large vibrations indicate an Induced Draft Fan (IDF) problem. The Fast Fourier Transform (FFT) method cannot be applied to non-stationary vibration signals. Therefore, this study aims to analyze non-stationary vibration signals for IDF vibration signals at start-up so that the source of damage to the IDF can be found. The research process begins with a brief measurement of both bearing locations with horizontal and axial axes. Processing of the vibration signal from the measurement using the FFT method and the Short Time Fourier Transform (STFT). Based on the STFT spectrogram graph for measurements on the horizontal and axial axes, the dominant frequency values are the same. The frequency with the largest amplitude value is at one RPM IDF or 25 Hz. High vibration at 1 RPM is a big indication that the IDF is experiencing unbalance.

Optical Time-Mapped Spectrograms (II): Fractional Talbot Designs

Real-time implementations of joint time-frequency analysis over instantaneous bandwidths above the GHz range remain challenging. In a companion paper, we have proposed an analog photonic processing scheme that enables computing a short-time Fourier transform (STFT), or spectrogram (SP), of an incoming arbitrary broadband signal, over tens-of-GHz analysis bandwidths, in a continuous, gapless and real-time manner. The proposed method involves a temporal sampling of the signal under test (SUT) with a periodic train of interfering, linearly chirped optical pulses followed by group-velocity dispersion to map the spectra of consecutive and overlapping truncated sections of the SUT along the time domain. This scheme offers a notable design versatility to customize the performance specifications of the computed SP, but it generally involves a sub-optimal non-uniform sampling of the SUT and it requires the use of a bulky and expensive pulsed optical source. In this communication, we show that this previous general scheme can be easily configured to ensure an optimal, uniform sampling of the SUT by simply setting the involved dispersive lines to satisfy a fractional self-imaging condition, while keeping all the advantages (e.g., design versatility) of the original scheme. Moreover, the resulting design is further adapted to entirely avoid the need for a pulsed source, using instead a more efficient and simpler phase-only temporal sampling of the SUT, e.g., implemented through electro-optic phase modulation. We derive the design conditions and performance trade-offs of the proposed time-mapped STFT schemes based on dispersion-induced fractional Talbot self-imaging. Through numerical simulations and experimental demonstration, we confirm the potential of this simple and efficient approach for real-time SP analysis of arbitrary signals over instantaneous bandwidths above a few tens of GHz, with MHz frequency resolutions and ultrahigh processing speeds, approaching billions of FTs per second.

A New Deep Learning Framework for Imbalance Detection of a Rotating Shaft.

Rotor unbalance is the most common cause of vibration in industrial machines. The unbalance can result in efficiency losses and decreased lifetime of bearings and other components, leading to system failure and significant safety risk. Many complex analytical techniques and specific classifiers algorithms have been developed to study rotor imbalance. The classifier algorithms, though simple to use, lack the flexibility to be used efficiently for both low and high numbers of classes. Therefore, a robust multiclass prediction algorithm is needed to efficiently classify the rotor imbalance problem during runtime and avoid the problem's escalation to failure. In this work, a new deep learning (DL) algorithm was developed for detecting the unbalance of a rotating shaft for both binary and multiclass identification. The model was developed by utilizing the depth and efficacy of ResNet and the feature extraction property of Convolutional Neural Network (CNN). The new algorithm outperforms both ResNet and CNN. Accelerometer data collected by a vibration sensor were used to train the algorithm. This time series data were preprocessed to extract important vibration signatures such as Fast Fourier Transform (FFT) and Short-Time Fourier Transform (STFT). STFT, being a feature-rich characteristic, performs better on our model. Two types of analyses were carried out: (i) balanced vs. unbalanced case detection (two output classes) and (ii) the level of unbalance detection (five output classes). The developed model gave a testing accuracy of 99.23% for the two-class classification and 95.15% for the multilevel unbalance classification. The results suggest that the proposed deep learning framework is robust for both binary and multiclass classification problems. This study provides a robust framework for detecting shaft unbalance of rotating machinery and can serve as a real-time fault detection mechanism in industrial applications.

Material Classification and Aging Time Prediction of Structural Metals Using Laser-Induced Breakdown Spectroscopy Combined with Probabilistic Neural Network.

In this paper, laser-induced breakdown spectroscopy (LIBS) combined with a probabilistic neural network (PNN) was applied to classify engineering structural metal samples (valve stem, welding material, and base metal). Additionally, utilizing data from the plasma emission spectrum generated by laser ablation of samples with different aging times, an aging time prediction model based on a firefly optimized probabilistic neural network (FA-PNN) was established, which can effectively evaluate the service performance of structural materials. The problem of insufficient features obtained by principal component analysis (PCA) for predicting the aging time of materials is addressed by the proposal of a time-frequency feature extraction method based on short-time Fourier transform (STFT). The classification accuracy (ACC) of time-frequency features and principal component features was compared under PNN. The results indicate that, in comparison to the PCA feature extraction approach, the time-frequency feature extraction method based on STFT demonstrates higher accuracy in predicting the time of aging materials. Then, the relationship between classification accuracy (ACC) and settings of PNN was discussed. The ACC of the PNN model for both the material classification test set and the aging time test set achieved 100% with Firefly (FA) optimization algorithms. This result was also compared with the ACC of ANN, KNN, PLS-DA, and SIMCA for the aging time test set (95%, 87.5%, 85%, and 62.5%, respectively). The experimental results demonstrated that the classification model using LIBS combined with FA-PNN could realize better classification accuracy.

Blind extraction of acoustic signals from rolling bearing composite faults based on ISCA-IMCKD

In production practice, the signal gathered by a sensor often includes strong ambient noise, and its composition is complex. Focusing on the problem that traditional methods are difficult to separate and extract fault frequencies from strong background noise, a novel compound fault blind extraction method based on improved sparse component analysis (ISCA) and improved maximum correlation kurtosis deconvolution (IMCKD)-named ISCA‐IMCKD- is suggested. Initially, the signal that the sensor has collected is shifted into time-frequency area signal by short-time Fourier transform (STFT). In addition, the single source domain characteristic data is screened by improved single source point detection to determine the number of sources. Second, the ISCA method is optimized by using cosine distance improved fuzzy C-means clustering, which is utilized to further process the characteristic data to calculate the mixing matrix. Moreover, the estimated source signal is initially extracted according to the membership degree of clustering results. Finally, the estimated source signal is shifted into the time area by inverse STFT transform, and the IMCKD is employed to enhance the characteristics of the projected source signals. Meanwhile, the initially estimated source signal is completely separated, and the defect frequencies of the composite faults are finally extracted by envelope analysis. Simulation experiments and measured data are employed to certify the viability of the proposed means. The defect detection of rolling bearings is finished while the time cost is significantly saved.

Evaluation and Exploration of Machine Learning and Convolutional Neural Network Classifiers in Detection of Lung Cancer from Microarray Gene-A Paradigm Shift.

Microarray gene expression-based detection and classification of medical conditions have been prominent in research studies over the past few decades. However, extracting relevant data from the high-volume microarray gene expression with inherent nonlinearity and inseparable noise components raises significant challenges during data classification and disease detection. The dataset used for the research is the Lung Harvard 2 Dataset (LH2) which consists of 150 Adenocarcinoma subjects and 31 Mesothelioma subjects. The paper proposes a two-level strategy involving feature extraction and selection methods before the classification step. The feature extraction step utilizes Short Term Fourier Transform (STFT), and the feature selection step employs Particle Swarm Optimization (PSO) and Harmonic Search (HS) metaheuristic methods. The classifiers employed are Nonlinear Regression, Gaussian Mixture Model, Softmax Discriminant, Naive Bayes, SVM (Linear), SVM (Polynomial), and SVM (RBF). The two-level extracted relevant features are compared with raw data classification results, including Convolutional Neural Network (CNN) methodology. Among the methods, STFT with PSO feature selection and SVM (RBF) classifier produced the highest accuracy of 94.47%.

Tool Wear Monitoring Based on the Gray Wolf Optimized Variational Mode Decomposition Algorithm and Hilbert–Huang Transformation in Machining Stainless Steel

The online monitoring and prediction of tool wear are important to maintain the stability of machining processes. In most cases, the tool wear condition can be evaluated by signals such as force, sound, vibration, and temperature, which are often processed via Fourier-transform based methods, typically, the short-time Fourier transform (STFT). However, the fixed-width window function in STFT has many limitations. In this paper, a novel tool wear monitoring method based on variational mode decomposition (VMD) and Hilbert–Huang transformation (HHT) were developed to monitor the wear of carbide tools in machining stainless steel. In this method, the intrinsic mode function (IMF) was used as the fitness function, and the (K alpha) parameter sets for VMD were optimized by the gray wolf optimization (GWO). The results show that the characteristic frequency in the GWO-VMD-HHT method is more significant with no aliasing compared with the EMD-HHT method, and an obvious characteristic frequency shift phenomenon is present. By utilizing the energy value of IMF3 as the feature to classify the wear state of the cutting tool, the increase of energy reached 85.48% when 260–315 milling passes were in severe wear state. GWO, which can accurately find the best parameters for VMD, not only solves the problem that the Entropy Function is not suitable for force signals, but also provides reference for the selection of parameters of VMD.

Classification of Epileptic and Psychogenic Nonepileptic Seizures via Time-Frequency Features of EEG Data.

The majority of psychogenic nonepileptic seizures (PNESs) are brought on by psychogenic causes, but because their symptoms resemble those of epilepsy, they are frequently misdiagnosed. Although EEG signals are normal in PNES cases, electroencephalography (EEG) recordings alone are not sufficient to identify the illness. Hence, accurate diagnosis and effective treatment depend on long-term video EEG data and a complete patient history. Video EEG setup, however, is more expensive than using standard EEG equipment. To distinguish PNES signals from conventional epileptic seizure (ES) signals, it is crucial to develop methods solely based on EEG recordings. The proposed study presents a technique utilizing short-term EEG data for the classification of inter-PNES, PNES, and ES segments using time-frequency methods such as the Continuous Wavelet transform (CWT), Short-Time Fourier transform (STFT), CWT-based synchrosqueezed transform (WSST), and STFT-based SST (FSST), which provide high-resolution time-frequency representations (TFRs). TFRs of EEG segments are utilized to generate 13 joint TF (J-TF)-based features, four gray-level co-occurrence matrix (GLCM)-based features, and 16 higher-order joint TF moment (HOJ-Mom)-based features. These features are then employed in the classification procedure. Both three-class (inter-PNES versus PNES versus ES: ACC: 80.9%, SEN: 81.8%, and PRE: 84.7%) and two-class (Inter-PNES versus PNES: ACC: 88.2%, SEN: 87.2%, and PRE: 86.1%; PNES versus ES: ACC: 98.5%, SEN: 99.3%, and PRE: 98.9%) classification algorithms performed well, according to the experimental results. The STFT and FSST strategies surpass the CWT and WSST strategies in terms of classification accuracy, sensitivity, and precision. Moreover, the J-TF-based feature sets often perform better than the other two.

A Long-Term Recurrent Convolutional Network for SNR Estimation of Cone-Shaped Target

Signal-to-Noise Ratio (SNR) is an important prior information for many micro-motion echo processing approaches and techniques. Estimating SNR in advance can effectively enhance the performance of such technologies. This letter proposes a Long Term Recurrent Convolutional Network (LRCN) based SNR estimation method for cone-shaped target. First, mathematical expression of the echo is derived by studying the cone-shaped target micro-motion model, and hence the characteristics of the echo and its Short-Time Fourier Transform (STFT) are analyzed. Second, the motivation for the proposed are analyzed, and the echo is converted as an time-frequency graph (TFG) with STFT, so that an estimation technique based on LRCN (a hybrid of convolutional neural network (CNN) and recurrent neural networks (RNN)) is designed to estimate the SNR. Experiments indicate that the proposed method outperforms other already existing methods with better accuracy and robustness.

Flaw Detection in Wire and Arc Additive Manufacturing Using In-Situ Wide Frequency Bandwidth Acoustic Pressure

Wire and arc additive manufacturing (WAAM) is an additive manufacturing (AM) process that can produce large metallic components with low material waste and high production rates. However, WAAM’s high deposition rates require high heat input which can result in potential defects such as pores, cracks, lack of fusion or distortion. For practical implementation of the WAAM process in an industrial environment it is necessary to ensure defects-free production. However, the NDT inspection using traditional NDT techniques (ultrasound, eddy currents, x-ray, for example) is a very demanding task, especially during part production. Therefore, reliable online NDT inspection and monitoring techniques are needed for the industrial spread of WAAM. The objective of this work is to detect flaw formation on WAAM produced parts using in-situ acquired acoustic data with a frequency bandwidth from 10 to 1MHz. WAAM parts were processed with deliberately introduced contaminations while its acoustic signal was obtained to correlate different signal characteristics with defects. To identify flaw formation, two distinct types of microphones were employed to acquire data from the same deposition process. The processing of the signal consisted of applying time and frequency domain techniques, namely, Power Spectral Density and Short Time Fourier Transform. The acoustic signatures obtained allowed for the differentiation between flawed and flaw free signals and for the spatial location of the contaminations. The acoustic signal acquired also showed that the data acquired by conventional microphones is not enough to fully characterize the acoustic spectrum emitted by the WAAM process. This work demonstrates the potential of acoustic data and signal processing in the online inspection of WAAM produced parts.

Improving the Accuracy and Resolution of Filter- and Frequency-to-Time Mapping-Based Time and Frequency Acquisition Methods by Broadening the Filter Bandwidth

In this article, a method for improving the accuracy and resolution of the filter-and frequency-to-time mapping (FTTM)-based time and frequency acquisition system in the fast sweep scenario is proposed by broadening the filter bandwidth. It is found that the width of the generated pulse via FTTM is mainly determined by the impulse response of the filter when the sweep rate is very fast. Therefore, appropriately increasing the filter bandwidth can significantly reduce the pulsewidth, so as to improve the measurement accuracy and resolution. To verify the validity, short-time Fourier transform (STFT) and microwave frequency measurement (MFM) systems employing the stimulated Brillouin scattering (SBS)-based FTTM are further investigated by inserting a frequency-sweep pump wave for SBS gain bandwidth broadening. The improvement of accuracy and frequency resolution is well confirmed by comparing the measurement results with and without SBS gain spectrum broadening. The MFM accuracy is improved by more than ten times compared with other reported works under a similar fast sweep rate, and the frequency resolution of the STFT is also much improved compared with our former results.

A Fast and Robust Local-Based Protection Algorithm Based on the High-Frequency Transient for HVDC Grid

With the expansion of HVDC grids, short-circuit faults and their protection issues are among the main challenges of these networks, which should be given more attention. This study presents an investigation of the DC side fault current in HVDC grids during the transient period. Also, a fast and robust protection algorithm is introduced that is essential for secure and appropriate HVDC grids operation. The Short-Time Fourier Transform (STFT) is applied to extract the frequency components of the fault current, and the time features of these components are utilized as the basis of the proposed algorithm. A four-terminal cable-based HVDC grid with bipolar configuration is modeled via EMTP, and a wide range of simulation scenarios are applied for various fault locations, fault impedances, and operation points. Furthermore, the algorithm thresholds are determined with the explicit procedure. Simulation results indicate that the proposed method is fast, selective, sensitive, and robust against AC-side fault, operating condition variations, fault distance, and fault direction. Also, it has an appropriate performance during high impedance faults.

Enhancing non-profiled side-channel attacks by time-frequency analysis

Side-channel analysis (SCA) has become an increasing important method to assess the physical security of cryptographic systems. In the process of SCA, the number of attack data directly determines the performance of SCA. With sufficient attack data, the adversary can achieve a successful SCA. However, in reality, the cryptographic device may be protected with some countermeasures to limit the number of encryptions using the same key. In this case, the adversary cannot use casual numbers of data to perform SCA. The performance of SCA will be severely dropped if the attack traces are insufficient. In this paper, we introduce wavelet scatter transform (WST) and short-time fourier transform (STFT) to non-profiled side-channel analysis domains, to improve the performance of side-channel attacks in the context of insufficient data. We design a practical framework to provide suitable parameters for WST/STFT-based SCA. Using the proposed method, the WST/STFT-based SCA method can significantly enhance the performance and robustness of non-profiled SCA. The practical attacks against four public datasets show that the proposed method is able to achieve more robust performance. Compared with the original correlation power analysis (CPA), the number of attack data can be reduced by 50–95%.

Recognition of Radar-Based Deaf Sign Language Using Convolution Neural Network

The difficulties in the communication between the deaf and normal people through sign language can be overcome by implementing deep learning in the gestures signal recognition. The use of the Convolution Neural Network (CNN) in distinguishing radar-based gesture signals of deaf sign language has not been investigated.This paper describes the recognition of gestures of deaf sign language using radar and CNN. Six gestures of deaf sign language were acquired from normal subjects using a radar system and processed. Short-time Fourier Transform wasperformed to extract the gestures features and the classification was performed using CNN. The performance of CNN was examined using two types of inputs; segmented and non-segmented spectrograms. The accuracy of recognising the gestures is higher (92.31%)using the non-segmented spectrograms compared to the segmented spectrogram. The radar-based deaf sign language could be recognised accurately using CNN without segmentation.

Speech Enhancement Based on Discrete Wavelet Packet Transform and Itakura-Saito Nonnegative Matrix Factorisation

Nonnegative matrix factorization (NMF) is one of the most popular machine learning tools for speech enhancement (SE). However, there are two problems reducing the performance of the traditional NMFbased SE algorithms. One is related to the overlap-and-add operation used in the short time Fourier transform (STFT) based signal reconstruction, and the other is the Euclidean distance used commonly as an objective function; these methods can cause distortion in the SE process. In order to get over these shortcomings, we propose a novel SE joint framework which combines the discrete wavelet packet transform (DWPT) and the Itakura-Saito nonnegative matrix factorisation (ISNMF). In this approach, the speech signal was first split into a series of subband signals using the DWPT. Then, the ISNMF was used to enhance the speech for each subband signal. Finally, the inverse DWPT (IDWT) was utilised to reconstruct these enhanced speech subband signals. The experimental results show that the proposed joint framework effectively enhances the performance of speech enhancement and performs better in the unseen noise case compared to the traditional NMF methods.

NRC-Net: Automated noise robust cardio net for detecting valvular cardiac diseases using optimum transformation method with heart sound signals

Objective:Cardiovascular Diseases (CVDs) can be effectively treated when detected early, reducing mortality rates significantly. Traditionally, Phonocardiogram (PCG) signals have been utilized for detecting cardiovascular disease due to their cost-effectiveness and simplicity. Nevertheless, various environmental and physiological noises frequently affect the PCG signals, compromising their essential distinctive characteristics. The prevalence of this issue in overcrowded and resource-constrained hospitals can compromise the accuracy of medical diagnoses. Therefore, this study aims to discover the optimal transformation method for detecting CVDs using noisy heart sound signals and propose a noise-robust network to improve the CVDs classification performance. Methods:For the identification of the optimal transformation method for noisy heart sound data, Mel Frequency Cepstral Coefficients (MFCCs), Short-Time Fourier Transform (STFT), Constant-Q Nonstationary Gabor Transform (CQT), and Continuous Wavelet Transform (CWT) has been used with Visual Geometry Group with 16-layer deep model architecture (VGG16). Furthermore, we propose a novel Convolutional Recurrent Neural Network (CRNN) architecture called Noise Robust Cardio Net (NRC-Net), which is a lightweight model to classify Mitral Regurgitation, Aortic Stenosis, Mitral Stenosis, Mitral Valve Prolapse, and normal heart sounds using PCG signals contaminated with respiratory and random noises. An attention block is included to extract important temporal and spatial features from the noisy corrupted heart sound. Results:The results of this study indicate that CWT is the optimal transformation method for noisy heart sound signals. When evaluated on the GitHub heart sound dataset, CWT demonstrates an accuracy of 95.69% for VGG16, which is 1.95% better than the second-best CQT transformation technique. Moreover, our proposed NRC-Net with CWT obtained an accuracy of 97.4%, which is 1.71% higher than the VGG16. Conclusion:Based on the outcomes illustrated in the paper, the proposed model is robust to noisy data and can be used in polyclinics and hospitals to detect valvular cardiac diseases accurately.

Detection of Anomalies in Additively Manufactured Metal Parts Using CNN and LSTM Networks

The process of metal additive manufacturing (AM) involves creating strong, complex components by using fine metal powders. Extensive use of AM methods is expected in near future for the production of small and medium-sized batches of end-use products and tools. The ability to detect loads and defects would enable AM components to be used in critical applications and improve their value. In this study, the Surface Response to Excitation (SuRE) method was used to investigate wave propagation characteristics and load detection on AM metallic specimens. With completely solid infills and the same geometry, three stainless steel test bars are produced: one conventionally and two additively. To investigate the effect of infills, four bars with the same geometries are 3D printed with triangular and gyroid infills with either 0.5 mm or 1 mm skin thickness. Two piezoelectric disks are attached to each end of the test specimens to excite the parts with guided waves from one end and monitor the dynamic response to excitation at the other end. The response to excitation was recorded when bars were in a relaxed condition and when compressive loads were applied at five levels in the middle of them. For converting time-domain signals into 2D time-frequency images, the Short-Time Fourier Transform (STFT) and Continuous Wavelet Transform (CWT) were implemented. To distinguish the data based on fabrication characteristics and level of loading, two deep learning models (Long Short-term Memory algorithm (LSTM) and Convolutional Neural Networks (2D CNN)) were utilized. Time-frequency images were used to train 2D CNN, while raw signal data was used to train LSTM. It was found that both LSTM and 2D CNN could estimate solid parts' loading level with an accuracy of more than 90%. In parts with infills, CNN outperformed LSTM for the classification of over five classes (internal geometry and loading level simultaneously).

Abnormal Respiratory Sound Identification Using Audio-Spectrogram Vision Transformer.

Respiratory disease, the third leading cause of deaths globally, is considered a high-priority ailment requiring significant research on identification and treatment. Stethoscope-recorded lung sounds and artificial intelligence-powered devices have been used to identify lung disorders and aid specialists in making accurate diagnoses. In this study, audio-spectrogram vision transformer (AS-ViT), a new approach for identifying abnormal respiration sounds, was developed. The sounds of the lungs are converted into visual representations called spectrograms using a technique called short-time Fourier transform (STFT). These images are then analyzed using a model called vision transformer to identify different types of respiratory sounds. The classification was carried out using the ICBHI 2017 database, which includes various types of lung sounds with different frequencies, noise levels, and backgrounds. The proposed AS-ViT method was evaluated using three metrics and achieved 79.1% and 59.8% for 60:40 split ratio and 86.4% and 69.3% for 80:20 split ratio in terms of unweighted average recall and overall scores respectively for respiratory sound detection, surpassing previous state-of-the-art results.