Reliability and Generalization of Gait Biometrics Using 3D Inertial Sensor Data and 3D Optical System Trajectories

Particularities in the individuals’ style of walking have been explored for at least three decades as a biometric trait, fueling the automatic gait recognition field. Whereas, gait recognition works usually focus on improving end-to-end performance measures, and this work aims at understanding which individuals’ traces are more relevant to improve subjects’ separability. For such, a manifold projection technique and a multi-sensor gait dataset were adopted to investigate the impact of each data source characteristics on this separability. The assessments have shown it is hard to distinguish individuals based only on their walking patterns in a subject identification scenario. In this scenario, the subjects’ separability is more related to their physical characteristics than their movements related to gait cycles and biomechanical events. However, this study’s results also points to the feasibility of learning identity characteristics from individuals’ walking patterns learned from similarities and differences between subjects in a verification setup. The explorations concluded that periodic components occurring in frequencies between 6Hz and 10Hz are more significant for learning these patterns than events and other biomechanical movements related to the gait cycle, as usually explored in the literature.

An Adaptive Decomposition Ensemble Learning Approach for Metro Passenger Flow Forecasting

Accurate and timely metro passenger flow forecasting is critical for the successful deployment of intelligent transportation systems. However, it is quite challenging to propose an efficient and robust forecasting approach due to the inherent randomness and variations of metro passenger flows. In this study, we present a novel adaptive decomposition ensemble learning approach to accurately forecast the volume of metro passenger flows that combines the complementary advantages of variational mode decomposition (VMD), seasonal autoregressive integrated moving averaging (SARIMA), a multilayer perceptron (MLP) network and a long short-term memory (LSTM) network. Our proposed decomposition ensemble learning approach consists of three important stages. The first stage applies VMD to decompose the metro passenger flow data into periodic components, deterministic components and volatility components. Then, we employ the SAIMA model to forecast the periodic component, the LSTM network to learn and forecast the deterministic component and the MLP network to forecast the volatility component. In the last stage, these diverse forecasted components are reconstructed by another MLP network. The empirical results show that our proposed decomposition ensemble learning approach not only has the best forecasting performance compared with the relevant benchmark models but also appears to be the most promising and robust based on the historical passenger flow data in the Shenzhen subway system and several standard evaluation measures.

Separating neural oscillations from aperiodic 1/f activity: challenges and recommendations

Electrophysiological power spectra typically consist of two components: An aperiodic part usually following an 1/f power law P∝1/fβ and periodic components appearing as spectral peaks. While the investigation of the periodic parts, commonly referred to as neural oscillations, has received considerable attention, the study of the aperiodic part has only recently gained more interest. The periodic part is usually quantified by center frequencies, powers, and bandwidths, while the aperiodic part is parameterized by the y-intercept and the 1/f exponent β. For investigation of either part, however, it is essential to separate the two components. In this article, we scrutinize two frequently used methods, FOOOF (Fitting Oscillations & One-Over-F) and IRASA (Irregular Resampling Auto-Spectral Analysis), that are commonly used to separate the periodic from the aperiodic component. We evaluate these methods using diverse spectra obtained with electroencephalography (EEG), magnetoencephalography (MEG), and local field potential (LFP) recordings relating to three independent research datasets. Each method and each dataset poses distinct challenges for the extraction of both spectral parts. The specific spectral features hindering the periodic and aperiodic separation are highlighted by simulations of power spectra emphasizing these features. Through comparison with the simulation parameters defined a priori, the parameterization error of each method is quantified. Based on the real and simulated power spectra, we evaluate the advantages of both methods, discuss common challenges, note which spectral features impede the separation, assess the computational costs, and propose recommendations on how to use them.

Vibration fault identification of a turbojet engine based on cepstrum analysis

A vibration fault identification method based on vibration state characteristics of a turbojet engine and cepstrum analysis technology was proposed in this paper, and the application of cepstrum in vibration analysis of an aero-engine was also discussed. The vibration data of the turbojet engine in three different test cases of 0.8 rated state, max power state, and afterburning state were analyzed using the cepstrum analysis method. The periodic components and the characteristics of multi-component side-frequency complex signals in the dense overtone vibration signals were separated and extracted, which reflected the sensitivity of the positions of the compressor casing and the turbine casing to the harmonic vibration components of high- and low-pressure rotors and the characteristic difference of different vibration parts. Thus, effective identification of vibration faults was achieved. The results shows that the cepstrum analysis technique applied to the vibration analysis of the turbojet engine can better identify the sideband components of the frequency domain modulated signal and enhance the recognition capability of the fault frequency component, which is helpful to identify the engine vibration fault quickly and accurately.

Intra-annual and long-periodic components in the changes of precipitation over the Antarctic Peninsula and their possible causes

In order to identify and study the main mechanisms of the formation of atmospheric precipitation, in the article the monthly and annual amounts of precipitation were analyzed from the observations results at Vernadsky, Bellingshausen and Grytviken stations. For the last station, a small linear trend of precipitation increase was detected, while at Vernadsky and Bellingshausen station it is practically absent. At the next stage of the study, the characteristics of intra-annual component of the precipitation variability for these stations were obtained. In the annual course, the component of precipitation variability is represented by 3 peaks – March, July and October (at Bellingshausen station March and July only), with a well-pronounced 4-year periodicity. However, data from Vernadsky station indicates a decrease of the seasonal component in time, at Grytviken station the seasonal component is stable, while at Bellingshausen station is increasing of the seasonal component in time. The analysis of long-period components of the precipitation variability of was carried out on the remains of the data obtained after the analysis of the intra-annual component. For the long-period component of precipitation variability at Vernadsky station, five statistically significant harmonics were obtained, which are reflected in periods of 6.8, 2.4, 4.0, 5.1, and 5.3 years. For Grytviken and Bellingshausen stations, 4 statistically significant harmonics were obtained, the periods of which are 4.2, 0.8, 1.7, 8.9 years and 1.5, 2.0, 2.8, 0.2 years, respectively. Today, the main phases of solar activity are well known, which are about 11 years old. The long-period components of precipitation variability obtained in the work for the stations under consideration (to 10.3, 12 and 34.1 years) are identical (close) to the mentioned phase of solar activity. This allowed the authors to draw preliminary conclusions about the influence of solar activity on the conditions for the formation of precipitation in the region under study. However, direct correlation analysis did not confirm this, as in the case of the El Niño influence.

Experimental Observations in Tire Cavity Resonance and Interactions with Periodic Noise Components

Tire cavity resonance is one of the major sources of tire-related in-cabin noise and vibration. It has gained more attention in recent years with the growth of the electric vehicle market. This is due to the absence of masking noise from the internal combustion engine and powertrain. Thus, the mitigation of this issue has become a critical task for tire and vehicle manufacturers. The excited cavity resonant frequency in an unloaded condition is typically between 170 - 220 Hz. However, multiple studies have shown that loading the tire will result in two dominant resonances transmitted into the cavity. Their corresponding mode shapes are typically described in terms of the direction of their characteristic acoustic pressure variation i.e., fore-aft cavity mode and vertical cavity mode. As the tire's rotational speed increases, in-cabin measurements show that the tire cavity resonant frequencies separate from each other. Further, interactions with the periodic component of tire noise at certain speeds are also observed. These periodic components can be attributed to tire non-uniformities and tread pattern related excitation. This interaction is perceived as tonal noise inside the vehicle cabin at discrete speeds. This work presents experimental results summarizing these findings.

Extension of Harmonic Balance Approach for Large-Eddy Simulation of Unsteady Flows in Cascade

A full scale-resolving simulation of cascades flutter is time consuming because of computational inefficiency owing to its low nondimensional frequencies. To improve the efficiency and reliability of the numerical analyses for such flows, we propose an efficient scale-resolving simulation method dedicated to time-periodic flows by extending the harmonic balance approach to a large-eddy simulation. This method combines convergence calculations of the steady-state problem based on the harmonic balance method for periodic components, and the nonlinear time-marching method for small scale turbulent fluctuations. Using the proposed method, deterministic periodic components and stochastic turbulent fluctuations are calculated simultaneously, and the effect of turbulent fluctuations on deterministic periodic components is directly calculated without using turbulence models. In this paper, we present the algorithm of the simulation technique and the progress of validation calculations for channel flow excited in the streamwise direction.

A time series analysis of permanent GNSS stations in the northwest network of Iran

The aim of this study is to estimate reliable velocities along with their realistic uncertainties based on a robust time series analysis including analysis of deterministic and stochastic (noise) models. In the deterministic model analysis part, we use a complete station motion model comprised of jump effects, linear and nonlinear trend, periodic components, and post-seismic deformation model. This part also consists of jump detection, outlier detection, and statistical significance of jumps. We perform the deterministic model analysis in an iterative process to elevate its efficiency. In the noise analysis part, first, we remove the spatial correlation of observations using the weighted stacking method based on the common mode error (CME) parameter. Next, a combination of white and flicker noises is used to determine the stochastic model. This time series analysis is applied for 11-year time series of 25 permanent GNSS stations from 2006 to 2016 in the northwest network of Iran. We reveal that there is a nonlinear trend in some stations, although most stations have a linear trend. In addition, we found that a combination of logarithmic and exponential functions is the most appropriate post-seismic deformation model in our study region. The result of the noise analysis shows that the spatial filtering reduces the norm of post-fit residual vector by 19.34%, 17.51%, and 12.44% on average for the east, north, and up components, respectively. Furthermore, the uncertainties obtained from the combination of white and flicker noises at the east, north, and up components are 5.0, 4.8, and 4.4 times greater than those of the white noise model, respectively. The results indicate that the stations move horizontally with an average velocity of 36.0 ± 0.3 mm/yr in the azimuth of 52.66° NE which is compatible with velocities obtained from MIDAS. We obtained the vertical velocity of most stations in the range of -5 to 5 mm/yr. However, in three stations of GGSH, ORYH, and BNAB, which are in the proximity of Lake Urmia, the vertical velocities are estimated to be -80.9 mm/yr, -50.6 mm/yr, and -11.4 mm/yr, respectively. Moreover, we found that these three stations possess large periodic signal amplitudes in all three coordinate components as well as a nonlinear trend in the up component.

A Novel Decomposition-Ensemble Learning Model Based on Ensemble Empirical Mode Decomposition and Recurrent Neural Network for Landslide Displacement Prediction

As vital comments on landslide early warning systems, accurate and reliable displacement prediction is essential and of significant importance for landslide mitigation. However, obtaining the desired prediction accuracy remains highly difficult and challenging due to the complex nonlinear characteristics of landslide monitoring data. Based on the principle of “decomposition and ensemble”, a three-step decomposition-ensemble learning model integrating ensemble empirical mode decomposition (EEMD) and a recurrent neural network (RNN) was proposed for landslide displacement prediction. EEMD and kurtosis criteria were first applied for data decomposition and construction of trend and periodic components. Second, a polynomial regression model and RNN with maximal information coefficient (MIC)-based input variable selection were implemented for individual prediction of trend and periodic components independently. Finally, the predictions of trend and periodic components were aggregated into a final ensemble prediction. The experimental results from the Muyubao landslide demonstrate that the proposed EEMD-RNN decomposition-ensemble learning model is capable of increasing prediction accuracy and outperforms the traditional decomposition-ensemble learning models (including EEMD-support vector machine, and EEMD-extreme learning machine). Moreover, compared with standard RNN, the gated recurrent unit (GRU)-and long short-term memory (LSTM)-based models perform better in predicting accuracy. The EEMD-RNN decomposition-ensemble learning model is promising for landslide displacement prediction.