Search

This study focuses on evaluating several machine learning algorithms for the condition assessment of utility timber poles. An efficient feature extraction technique combining Hilbert Huang transform (HHT) and wavelet packet transform (WPT) is adopted from the authors' previous work and implemented to determine damage-sensitive features from vibration data related to five serviceable and eight unserviceable in-situ timber poles. Then, these features are pre-processed using correlation heat map analysis for feature selection. Principal component analysis (PCA) is adopted as the final step of pre-processing for reducing noise interference and enhancing classification accuracy. Afterwards, a feature matrix is formed, which is fed into the selected classifiers for pattern recognition. In addition, the information gain method is also implemented and compared against PCA to examine the effect of feature selection. Finally, selected classifiers are employed using those dominant features, and their performance is evaluated based on six parameters – accuracy, precision, recall, F1-score, confusion matrix and Receiver Operating Characteristic (ROC) curve. It was found that choosing the best feature set related to the health state helps to improve the performance of the pattern recognition algorithms to a great extent. [ABSTRACT FROM AUTHOR]

Bridge structural health monitoring (BSHM) is extensively employed to assess whether bridge damage exceeds maintenance limits and helps maintenance decision-making. Frequency identification is an essential part of BSHM, which utilizes the first-order modal frequency as an evaluation indicator of bridge health status. This paper proposes a method for bridge damage identification based on extracting the first-order modal frequency of the bridge from dynamic responses of vehicle bogies. Firstly, the feasibility of extracting bridge modal frequency from bogie accelerations is theoretically deduced via a vehicle–bridge interaction model. The result indicates that the bogie accelerations encompass three frequency components, namely vehicle driving frequency, vehicle pitching frequency and bridge first-order modal frequency. Then, a multi-domain conjoint analysis method incorporating the time domain, frequency domain and time–frequency domain is proposed to extract the first-order frequency of damaged bridge from bogie accelerations. By defining a damage sensitivity index of the sliding window, the time-domain sensitive boundaries of bogie acceleration to bridge damage are pinpointed. Leveraging prior knowledge of the bridge modal frequency range, the frequency-domain sensitive boundaries bogie acceleration to bridge frequency are determined. On this basis, the sensitive region of bogie acceleration to bridge frequency can be precisely localized in the Hilbert Huang transform spectrum, ultimately identifying bridge frequency through searching local maximum instantaneous energy points within this region. Finally, comprehensive experiments, considering different vehicle speeds and track spectra scenarios, are carried out to investigate the effectiveness and robustness of the proposed method. The results affirm the performance of the method to accurately identify the first-order modal frequency of the damaged bridge, further underscoring the promising prospect of on-board BSHM. [ABSTRACT FROM AUTHOR]

To address the shortage of damage identification methods for modal bridge expansion joints (MBEJs). Two damage indexes, DI(IPH) and DI(HS) were constructed using the instantaneous phase (IP) and Hilbert spectrum of Hilbert Huang transform (HHT). Three specimens were designed and fabricated to identify the damage to the concrete in the center beam, support bar and anchorage zone of MBEJs. The impact vibration of MBEJs was excited by the hammer impulses and acceleration sensors were used to obtain the vibration response. The damage of the center beam, support bar and concrete in anchorage zone was all classified into five levels (level 1–level 5). DI(IPH) and DI(HS) increase gradually with increasing damage levels. After the center beam was completely broken, the DI(IPH) and DI(HS) improved by 40.31 and 40.61, respectively. The DI(IPH) and DI(HS) increased by 25.25 and 55.91 after the complete breakage of the support bar. The DI(IPH) and DI(HS) of the concrete in the anchorage zone at Level 5 showed more minor numerical changes than the center beam and support bar, with an increase of 7.49 and 35.45, respectively. The experimental results showed that HHT could effectively identify the damage in various parts of MBEJs with high accuracy. There is a clear indication of the degree of damage, which is suitable for the identification of damage in MBEJs. [ABSTRACT FROM AUTHOR]

Pumped storage is an important component of electrified wire netting. The safe and stable operation of pump turbines is extremely important. Among them, pressure pulsation is one of the main causes of pump turbine vibration. The characteristics of pressure pulsation are relatively complex, and it is difficult to directly observe their temporal changes using commonly used FFT methods. The division of frequency characteristics is often vague. Meanwhile, it is difficult to explain some phenomena such as frequency doubling. This article focuses on a certain model of pump turbine and uses SST model to numerically simulate the runaway condition of the pump turbine. And the Hilbert Huang transform method is used to analyze the pressure pulsation in the vaneless region and draft tube. The results show that the main characteristic frequencies of the vaneless region are blade passing frequency 112.5 Hz and rotational frequency 12.5 Hz. The main characteristic frequencies of the draft tube are vortex rope frequency near 3 Hz which energy ratio is up to 50%, rotational frequency, and blade passing frequency. The pressure pulsation characteristics in the vaneless region have changed from a complex composition of double blade passing frequency and rotational frequency to a distribution dominated by blade passing frequency. In the passage of the guide vane, the pressure pulsation is almost only characterized by blade passing frequency. The frequency characteristics of the vaneless region between the runner and the guide vane become complex again. Meanwhile, the results show that the characteristic frequencies of the vaneless region and the draft tube propagate upstream and downstream. [ABSTRACT FROM AUTHOR]

In this work, a surface-mounted circular cylinder with a fixed aspect ratio (ratio of height of the cylinder to its diameter) of 5 is subjected to a non-zero mean oscillating flow with a range of frequencies and amplitudes. Three-dimensional direct numerical simulations are then conducted on this finite-height cylinder. The mass and momentum equations are resolved using the finite volume-based Open Source Field Operation and Manipulation (OpenFOAM). A fixed Reynolds number Re = U o D / ν of 250 is used in this study, which is defined based on mean velocity at the inlet ( U o ) and cylinder diameter (D). Here ν is the kinematic viscosity of the working fluid. Non-dimensional velocity oscillation amplitude ( A ∗ = a / U o ) is varied from 0.1 to 0.3, while the non-dimensional oscillation frequency ( f ∗ = f / f o ) takes the values of 0.33, 0.5, 1, 2, and 3. Here a and f are the dimensional oscillation amplitude and frequency, respectively and f o is the vortex shedding frequency corresponding to a uniform flow at Re = 250. The three-dimensional vortex structures, presented with the help of iso- Q surfaces, show that the oscillating flow changes the size and shape of the hairpin-shaped vortices. Wake is found to be synchronized with the oscillation frequency at f * = 2 for each value of the A * and results in the maximum lift force on the cylinder. Hilbert Huang transformation analysis of the transverse velocity signals at a specific point in the wake reveals that the wake is more complex and aperiodic in nature for f * values of 0.33, 0.5, and 1, whereas it is periodic for f * = 2 and 3. In order to further disclose the nonlinearity associated with the oscillating flow, the degree of stationarity is discussed corresponding to each value of A * and f *. Dynamic mode decomposition is exploited to obtain information about the coherent vortical structures and their spatial and temporal behavior in the wake with a change in the value of f *. Effects of A * and f * on the dynamic characteristics are also investigated. [ABSTRACT FROM AUTHOR]

This research aims to provide the groundwork for smartly categorizing hand movements for use with prosthetic hands. The hand motions are classified using surface electromyography (sEMG) data. In reaction to a predetermined sequence of fibre activation, every single one of our muscles contracts. They could be useful in developing control protocols for bio-control systems, such human-computer interaction and upper limb prostheses. When focusing on hand gestures, data gloves and vision-based approaches are often used. The data glove technique requires tedious and unnatural user engagement, whereas the vision-based solution requires significantly more expensive sensors. This research offered a Deep Neural Network (DNN) automated hand gesticulation recognition system based on electromyography to circumvent these restrictions. This work primarily aims to augment the concert of the hand gesture recognition system via the use of an artificial classifier. To advance the recognition system's classification accuracy, this study explains how to build models of neural networks and how to use signal processing methods. By locating the Hilbert Huang Transform (HHT), one may get the essential properties of the signal. When training a DNN classifier, these characteristics are sent into it. The investigational results reveal that the suggested technique accomplishes a better categorization rate (98.5 % vs. the alternatives).

Pumped storage is an important component of electrified wire netting. The safe and stable operation of pump turbines is extremely important. Among them, pressure pulsation is one of the main causes of pump turbine vibration. The characteristics of pressure pulsation are relatively complex, and it is difficult to directly observe their temporal changes using commonly used FFT methods. The division of frequency characteristics is often vague. Meanwhile, it is difficult to explain some phenomena such as frequency doubling. This article focuses on a certain model of pump turbine and uses SST model to numerically simulate the runaway condition of the pump turbine. And the Hilbert Huang transform method is used to analyze the pressure pulsation in the vaneless region and draft tube. The results show that the main characteristic frequencies of the vaneless region are blade passing frequency 112.5 Hz and rotational frequency 12.5 Hz. The main characteristic frequencies of the draft tube are vortex rope frequency near 3 Hz which energy ratio is up to 50%, rotational frequency, and blade passing frequency. The pressure pulsation characteristics in the vaneless region have changed from a complex composition of double blade passing frequency and rotational frequency to a distribution dominated by blade passing frequency. In the passage of the guide vane, the pressure pulsation is almost only characterized by blade passing frequency. The frequency characteristics of the vaneless region between the runner and the guide vane become complex again. Meanwhile, the results show that the characteristic frequencies of the vaneless region and the draft tube propagate upstream and downstream.

Effective water resource management requires accurate estimation of the volume of water withdrawn through anthropogenic groundwater pumping. However, insufficient pumping information makes accurate estimation difficult. To address this issue, this study proposes a novel method for estimating pumping volumes that does not rely on pumping information. The proposed method uses high-resolution groundwater level (GWL) data from densely monitored wells and integrates Hilbert–Huang transform and empirical orthogonal function analysis to extract pump-associated signals, which can then be used to estimate total pumping. Our method was validated by analyzing GWL data from monitoring stations on the Choshui River alluvial fan in Taiwan. The results indicate the estimated pumping amount of 1. 71 × 1 0 9 – 2. 05 × 1 0 9 m 3 /year consistent with previous estimates obtained using survey and physics-based methods ranging from 9 × 1 0 8 m 3 /year to 2. 2 × 1 0 9 m 3 /year. In the areas with abundant groundwater head observations, the proposed method can efficiently estimate pumping rates with a high space–time resolution, and may serve as a useful reference for groundwater management. • This study proposes a method utilizing groundwater data to estimate spatiotemporal pumping rate. • Multi-year high space–time resolution of pumping rate distribution estimation was efficiently derived from hourly groundwater data. • The pumping rate distribution aligns with geological and agricultural water usage patterns. [ABSTRACT FROM AUTHOR]

This research aims to provide the groundwork for smartly categorizing hand movements for use with prosthetic hands. The hand motions are classified using surface electromyography (sEMG) data. In reaction to a predetermined sequence of fibre activation, every single one of our muscles contracts. They could be useful in developing control protocols for bio-control systems, such human-computer interaction and upper limb prostheses. When focusing on hand gestures, data gloves and vision-based approaches are often used. The data glove technique requires tedious and unnatural user engagement, whereas the vision-based solution requires significantly more expensive sensors. This research offered a Deep Neural Network (DNN) automated hand gesticulation recognition system based on electromyography to circumvent these restrictions. This work primarily aims to augment the concert of the hand gesture recognition system via the use of an artificial classifier. To advance the recognition system's classification accuracy, this study explains how to build models of neural networks and how to use signal processing methods. By locating the Hilbert Huang Transform (HHT), one may get the essential properties of the signal. When training a DNN classifier, these characteristics are sent into it. The investigational results reveal that the suggested technique accomplishes a better categorization rate (98.5 % vs. the alternatives)., Competing Interests: The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper., (© 2024 The Authors.)