non-Gaussian Characteristics Research Articles

Enhancing water quality prediction with advanced machine learning techniques: An extreme gradient boosting model based on long short-term memory and autoencoder

Achieving carbon neutrality in the pulp and paper industry necessitates effectively recycling pulp and papermaking wastewater, where continuous monitoring of effluent quality indices is crucial. This study suggests a novel machine learning-based model named LSTMAE-XGBOOST that integrates the feature extraction capabilities of autoencoder, the sequential feature learning capabilities of LSTM, and the high prediction accuracy of XGBOOST. This model is capable of extracting the complex relationships, non-Gaussian characteristics, and time series features from the papermaking wastewater data, and it demonstrates superior predictive performance. Compared to traditional machine learning models, the proposed model exhibits higher prediction accuracy. Specifically, when contrasted with partial least squares regression, LSTMAE-XGBOOST achieves a 40% increase in R2 and a 35% reduction in RMSE. Further comparative assessments against other machine learning-based hybrid models with similar structures confirm the superiority of integrating LSTM and XGBOOST within the hybrid model approach. This study contributes a compelling methodology for modeling effluent quality indices, offering significant implications for environmental management in the pulp and paper industry.

Effects of oncoming wind speed and turbulence intensity on wind characteristics of a simplified hill and ridge

Supertall buildings and long-span bridges are significantly affected by wind-induced vibrations, and the wind fields in mountainous areas are highly complex and influenced by the oncoming wind speed and turbulence intensity. To accurately determine the variation patterns of wind characteristics in mountainous areas under different oncoming wind speeds and turbulence intensities, large eddy simulation (LES) was employed to analyze wind fields over simplified hill and ridge models. By setting different basic wind speeds (5, 10, 15, and 20 m/s) and turbulence intensities (1%, 3%, 5%, and 10%) at the inlet, the variation patterns of wind characteristics over the simplified hill and ridge under atmospheric boundary layer inflow were studied, revealing wind field flow mechanisms. The results indicate that the wind characteristics on the leeward side of the simplified hill and ridge are significantly influenced by the oncoming wind speed and turbulence intensity. Increasing the oncoming wind speed and turbulence intensity leads to decreased wind profile deceleration, reduced changes in wind direction and attack angles, and increased wind speed amplification factor. In addition, the turbulence fluctuations, power spectra, and coherence function between two points on the leeward side increase with the oncoming wind speed and turbulence intensity. The turbulent integral scale decreases with an increase in the oncoming wind speed and turbulence intensity. As the wind speed and turbulence increase, the size of the recirculation bubble gradually decreases, with its center moving closer to the wall surface. In the vorticity field, the number of smaller-scale three-dimensional turbulent vortices near the hill and ridge increases. These differences in flow characteristics are the fundamental causes of the changes in wind characteristics. High wind speeds and turbulence intensities typically result in higher kurtosis and skewness, with significant non-Gaussian characteristics in the fluctuating wind. Traditional wind load design specifications for building architecture based on a Gaussian distribution may not be applicable to mountainous terrain. In practical engineering, the influences of the oncoming wind speed and turbulence intensity on wind characteristics should be fully considered.

Credibility-based multi-sensor fusion for non-Gaussian conversion error mitigation

In a complex environment, a multi-sensor fusion algorithm can compensate for the limitations of a single sensor’s performance. In a distributed fusion algorithm, sensors need to transmit local estimates to a central coordinate system, and the existence of coordinate transformation uncertainty can undermine the performance of data transmission. Therefore, this paper proposes a multi-sensor distributed fusion method based on trustworthiness. Firstly, considering the presence of non-Gaussian conversion errors, a credibility-based multi-sensor fusion framework is constructed. Secondly, to address the difficulty in estimating conversion errors when measurement errors follow a non-Gaussian distribution, an optimization model is constructed based on actual measurement information to estimate the distribution of non-Gaussian conversion errors. Then, in response to the non-linear and non-Gaussian characteristics of the target optimization function, a particle swarm optimization algorithm based on trustworthiness adaptive weights is proposed to estimate the coordinate transformation errors. Finally, given the inconsistency in local estimates due to missing sensor measurements or significant errors in a non-Gaussian complex environment, a maximum correntropy consensus algorithm is proposed to avoid the trustworthiness calculation being affected by the current measurement errors, thereby improving the accuracy of the global estimation.

Study on The Response Characteristics of Monopile Offshore Wind Turbines under Non-Gaussian loads in Typhoons

The fan structure is often seriously damaged under typhoon, so the study of the fan structure under typhoon is of great significance for its disaster prevention and mitigation. The actual measurement shows that the typhoon has obvious non-Gaussian characteristics, and the influence mechanism of this characteristic on the fan response characteristics is not clear. Therefore study the response characteristics of fans under non-Gaussian load. First, based on the autoregressive AR model, combined with the Johnson conversion model (JTM); second, the wind load is determined based on the specification; then the wind load. Finally, the response characteristics of 5MW fan are studied by characterizing the numerical cases of non-Gaussian fan with different strength of typhoon. The results show that the non-high characteristics of non-Gaussian wind field and the non-Gaussian characteristics become stronger with the strength of the non-Gaussian characteristics of wind field.

Numerical simulation of flow around a transversely oscillating square cylinder at Re=22000

The study utilized OpenFOAM software, combined with the k-ω shear stress transport (SST) turbulence model and Arbitrary Lagrangian-Eulerian (ALE) dynamic mesh method, to investigate the influence of the oscillation amplitude on the flow dynamics around a square cylinder at Re = 2.2 × 104. It was found that various oscillation amplitudes affect the non-Gaussian characteristics of fluctuating pressure differently along the cylinder's surface: diminishing at the leading edge while intensifying at the trailing-edge corner. Concurrently, secondary recirculation bubbles near the cylinder's leading and trailing edges gradually diminish and eventually disappear with increasing oscillation amplitudes. Analysis of the amplitude spectrum of transverse velocity indicates the exclusive presence of odd-order harmonics along the central axis, while both odd and even-order harmonics are observed on either side of the central axis. Additionally, heightened oscillation amplitudes result in increased energy transfer from the cylinder to the fluid, with the energy density function predominantly exhibiting negative values and periodic variations occurring at twice the frequency of cylinder displacement or lift coefficient. As the oscillation amplitude increases, the vortex in the wake region transitions from a single-row to a double-row configuration. The research results provide important theoretical guidance for deepening the understanding of the flow characteristics around the oscillating square cylinder.

Investigation on typhoon-induced aero-elastic response of membrane structures by wind tunnel test and numerical simulation

Extreme winds, such as typhoons, can lead to serious vibration and damage for flexible membrane roofs. An understanding of the aeroelastic behavior experienced by membrane structures during typhoons is therefore significant to allow well designed in practice. This paper investigates the aeroelastic response of umbrella shaped membrane structures under typhoon experimentally and numerically. The flexible scaled model is tested in typhoon field simulated in wind tunnel to investigate the aeroelastic characteristics varying with wind velocities and wind directions, including displacement response, non-Gaussian characteristics, frequency, modal shape and damping ratios et al. The full coupled fluid-structure interaction numerical model proposed is benchmarked and expanded in parameter discussions. The results indicate that non-Gaussian characteristics appear significant with positive skewness in pressure region and negative skewness in suction region. The probabilistic distribution proves leptokurtic type with kurtosis beyond three. The displacement response in statistics increases almost linearly with wind velocity while the non-Gaussian characteristics remain robust. The high-order mode shapes can be excited in typhoon, and their frequencies and damping ratios vary with wind velocities. The effects of both wind velocity and membrane pretension are proved to be more remarkable than rise-span ratio. This study can address the deficiency of current studies and provisions on the dynamic response of membrane structures in typhoons.

Data-model driven probability model of branch power flow for distribution networks and its application to analysis of overload and line loss

The fluctuation of highly penetrated distributed generations (DGs) in distribution networks (DNs) increases branch overload risks, which makes the analysis uncertainty of branch power flow more complex. The probability analysis of branch power can grasp the power flow operation characteristics under uncertain conditions, which supports the power flow optimization management and ensures the safe and economic operation of DNs. Therefore, this paper proposes a data-model driven probability analysis method for branch power flow in DNs. An approximate branch power model is first derived to reveal the analytical relationship between branch power and node power. Then, based on the branch power approximation model and the forecasting error, the datasets of branch apparent power are constructed to capture the complex nonlinear characteristics of the power flow. The non-Gaussian distribution characteristics of branch apparent power and line loss are described by the optimal probability fitting method. Finally, the probability level of branch overload is evaluated and the probability distribution of line loss is analyzed. The case studies demonstrate that the branch power approximation model is highly accurate, and the probability distribution of the branch apparent power presents different characteristics. The proposed method can quickly and accurately calculate the branch overload probability level.

Tendon fatigue analysis of tension leg platform using a new time-domain quasi-dynamic method

Due to the non-Gaussian characteristics of the tendon tension response of tension leg platform (TLP), the common statistical model is not adequate to estimate the fatigue damage accurately in the frequency domain, thus the time domain analysis is required. A new quasi-dynamic method is proposed to calculate the tendon tension response considering various factors. An equivalent analytical tendon model is employed and the post-treatment has been carried out for the reconstruction of the high-frequency stress response based on the vertical motion QTFs of each tendon by HydroSTAR. Forty sets of environmental cases are extracted from a scatter diagram. The rainflow counting method is applied to deal with the stress time series and the fatigue damage can be obtained combined with the relevant S-N curve. The number of cycles of tension for each tendon is discussed, considering the contribution of the first-order wave load, the second-order diff-frequency wave load, and the second-order sum-frequency wave load. The comparison of annual fatigue damage based on the quasi-dynamic method and coupled method by Orcaflex is carried out. The study proves that the quasi-dynamic method is an efficient way to estimate the fatigue life of each tendon in the time domain considering numerous cases of discrete sea states.

Double-Layer Distributed and Integrated Fault Detection Strategy for Non-Gaussian Dynamic Industrial Systems.

Currently, with the increasing scale of industrial systems, multisensor monitoring data exhibit large-scale dynamic Gaussian and non-Gaussian concurrent complex characteristics. However, the traditional principal component analysis method is based on Gaussian distribution and uncorrelated assumptions, which are greatly limited in practice. Therefore, developing a new fault detection method for large-scale Gaussian and non-Gaussian concurrent dynamic systems is one of the urgent challenges to be addressed. To this end, a double-layer distributed and integrated data-driven strategy based on Laplacian score weighting and integrated Bayesian inference is proposed. Specifically, in the first layer of the distributed strategy, we design a Jarque-Bera test module to divide all multisensor monitoring variables into Gaussian and non-Gaussian blocks, successfully solving the problem of different data distributions. In the second layer of the distributed strategy, we design a dynamic augmentation module to solve dynamic problems, a K-means clustering module to mine local similarity information of variables, and a Laplace scoring module to quantitatively evaluate the structural retention ability of variables. Therefore, this double-layer distributed strategy can simultaneously combine the different distribution characteristics, dynamism, local similarity, and importance of variables, comprehensively mining the local information of the multisensor data. In addition, we develop an integrated Bayesian inference strategy based on detection performance weighting, which can emphasize the differential contribution of local models. Finally, the fault detection results for the Tennessee Eastman production system and a diesel engine working system validate the superiority of the proposed method.

A novel integrated method for improving the forecasting accuracy of crude oil: ESMD-CFastICA-BiLSTM-Attention

The volatility of crude oil price can significantly impact the stability of the crude oil market and even the global economy. Effectively predicting global crude oil price and volatility provides a scientific basis for decision-making for market regulators and investors worldwide, promoting the sound development of the global economy. In this study, we integrate signal processing with deep learning methods to present an optimal forecasting strategy for global crude oil price and volatility. We select the daily price of the WTI crude oil market from April 4, 1983, to December 12, 2023, for calculating volatility. Subsequently, employing extreme-point symmetric empirical mode decomposition (ESMD), K-means clustering, and fast independent component analysis method, we decompose and reconstruct the forecasting data, obtaining independent components with non-Gaussian characteristics. These components serve as inputs to estimate the accuracy of various models, including BiLSTM, Attention, LSTM, SVR, RF, and their combinations, in predicting crude oil price and volatility from both a point prediction and interval prediction perspective. Empirical results demonstrate that data decomposition, reconstruction, and the BiLSTM-Attention model outperform other models in predicting crude oil price and volatility.

Estimation of the set of states obtained in particle number measurement schemes

The paper investigated a set of non-Gaussian states generated by measuring the number of particles in one of the modes of a two-mode entangled Gaussian state. It was demonstrated that all generated states depend on two types of parameters: some parameters are responsible for Gaussian characteristics, while other parameters are responsible for non-Gaussian characteristics. Among all generated states, we identified those optimally generated in terms of the generation probability and the magnitude of non-Gaussianity.

The prediction on non-Gaussian characteristics of wind pressure for the long-span roof in the mountainous area using proper orthogonal decomposition–deep learning framework

To make an accurate prediction of the non-Gaussian characteristics of wind pressure for the long-span roof, this study combines the proper orthogonal decomposition (POD) technique, convolutional neural network (CNN), and long short-term memory (LSTM) network to propose a novel POD-CNN-LSTM framework. Then, the proposed framework was well validated based on the wind tunnel testing of a long-span roof structure, and some error criteria, such as mean square root error and correlation coefficient, were adopted to evaluate the prediction accuracy of the non-Gaussian characteristics. Furthermore, two other methods, POD-CNN and POD-LSTM, were also used to conduct a comparative study. The obtained results illustrate that compared to POD-CNN and POD-LSTM, the proposed framework can achieve better performance on the pulsating wind pressure coefficient. For predictions of non-Gaussian characteristics, the output results of the proposed POD-CNN-LSTM show fewer errors, which means the predictions are close to the measured results, including skewness, kurtosis, and wind pressure probability density distributions. To summarize, the proposed POD-CNN-LSTM framework shows superiority over others, which means the proposed framework has good potential for the practical application of non-Gaussian prediction of the engineering structure.

Estimation of extreme wind pressure coefficients of buildings using POT with automated threshold selection

Extreme wind pressure coefficients of the high-rise buildings are critical for determining wind loads on components and claddings. To address the issue of insufficient samples in the widely used block maximum (BM) method, the peak over threshold (POT) method has been introduced in wind engineering for more effective and accurate estimation of extreme wind pressure coefficients. However, it is quite a challenge to appropriately select the threshold. This study proposes an automated threshold selection procedure using L-moment theory. Monte Carlo simulation is conducted to substantiate the theoretical feasibility of this procedure. Utilizing data from wind-tunnel pressure measurements of a high-rise building model, the study automatically selects thresholds and estimates extreme wind pressure coefficients. Comparative analysis with the BM method demonstrates that thresholds determined through automated threshold selection procedure exhibit stability with minimal deviation from standard extreme values. Furthermore, the developed procedure can effectively adapt to non-Gaussian characteristics of the sample time histories.

Numerical study of leakage rate in reciprocating seal with non-Gaussian topography characteristics using percolation mechanism

The topography significantly affects the leakage prevention of the sealing surface. However, the topography effect on sealing performance is not yet fully understood. The leakage mechanism cannot be revealed by average flow model using a statistical distribution to characterize the surface. This paper proposes a new leakage analysis method for reciprocating seals based on the percolation theory. The contact and lubrication analysis of non-Gaussian surfaces with different characteristics is conducted to establish three-dimensional percolation channels. The leakage rate can be calculated from the flow velocity across the cross-section at the critical constriction in the channel. The results show that compared with experimental data, the simulation results of the percolation model are more accurate than those of the average flow model.

Wind-induced interference effect of complex building clusters on long-span roof structures

The presence of complex buildings introduces interference effects and complicates the wind field around the long-span roof structure. The complexity of the problem makes it challenging to predict wind load of long-span roof structures, resulting in a scarcity of design data available to engineers and designers. Thus, the influence of interference effect on mean and peak pressure coefficients of a long-span tri-centered cylindrical roof structure was investigated through wind tunnel tests conducted in an atmospheric boundary layer wind tunnel. The study focused on the interference effects caused by two cooling towers, chimney, and some low-rise buildings, considering all wind directions. The results show that the interference effects are the result of a combination of amplification effects from the adjacent cooling towers and chimney and shielding effects from the surrounding low-rise buildings. Due to the shielding effect of low-rise buildings, the wind suction on the roof is mitigated compared to that of a single building, while at the end of the roof, wind suction is amplified by cooling tower effects. The interference effect of the building clusters, however, amplifies the fluctuating wind pressure coefficient and the peak pressure coefficient on the roof. To accurately estimate the wind load of building components and envelopes, furthermore, this study investigates the influence mechanism of interference effect on non-Gaussian characteristics of wind pressure combining with fluctuating wind pressure spectrum, spatial correlation of fluctuating wind pressure, and probability distribution characteristics of standardized wind pressure coefficient. The spatial correlation of wind pressure in the roof interference area exhibits a strong association, and the probability density distribution characteristics of the standardized wind pressure coefficient significantly deviate from the Gaussian curve, with a peak factor exceeding 3.5. Thus, it can be concluded that the wind pressure within this region demonstrates significant non-Gaussian characteristics. Hence, the zone values of peak pressure coefficients are determined using the peak factor approach to inform the wind resistance design of the envelope structure.

Exploring non-Gaussian sea ice characteristics via observing system simulation experiments

Exploring non-Gaussian sea ice characteristics via observing system simulation experiments

A novel method for the non-Gaussian wind pressure simulation based on autoregressive model and maximum entropy method

The simulation of wind pressures with non-Gaussian characteristics is a prerequisite to apply the Monte Carlo method in the structural wind-resistant reliability analysis. Due to its simplicity and accuracy, the autoregressive (AR) model has been widely used for the non-Gaussian process simulation. However, AR-based simulation methods perform poorly for the wind pressures with strong non-Gaussian characteristics. Moreover, they cannot reproduce the bimodal characteristics of some hardening wind pressures. To improve the simulation accuracy, a novel non-Gaussian simulation approach is proposed in this study based on AR model and maximum entropy method. In this approach, a set of formulas to determine the first ten moments of the input process as a function of the corresponding moments of the output process in the AR model have been firstly derived. Then, the probability distribution of the input process is reconstructed by the enhanced maximum entropy method. Finally, the translation process theory is introduced to simulate the non-Gaussian wind pressure process combined with AR model. Numerical results show the proposed method has higher simulation accuracy for the wind pressures with strong non-Gaussianity compared with the conventional AR-based method. Additionally, it can reproduce the bimodal characteristic for the wind pressures with bimodal distribution.

Non-Gaussian feature of the output field from a double qubit-cavity ultrastrong coupling system

We investigate the non-Gaussian feature of radiation in a circuit quantum electrodynamics (QED) system where two qubits are strongly coupled to a single-mode cavity. In the regime of ultrastrong coupling (USC), the rotating-wave approximation is not valid, and the Rabi Hamiltonian contains counter-rotating wave terms, leading to level crossing and avoided crossings in the energy spectrum. We further analyze the intensity-amplitude correlation of the output field in these two novel scenarios. In the USC regime, the creation and annihilation operators in the correlation function are replaced, allowing for the identification of non-Gaussian features in the output field. Our findings reveal that despite the absence of squeezing effects in the output light, significant non-Gaussian characteristics are present. Additionally, we demonstrate that as the driving or coupling strength increases, the non-Gaussian features of the output field become more pronounced. This suggests that USC systems hold broad potential applications in the realms of nonlinear optics and the generation of non-Gaussian states.

Comparison of viscoelastic flows in two- and three-dimensional serpentine channels.

Polymer solutions in the dilute regime play a significant role in industrial applications. Due to the intricate rheological properties of these highly viscoelastic fluids, especially in complex flow geometries, a thorough numerical analysis of their flow dynamics is imperative. In this research, we present a numerical investigation of purely elastic instability occurring in two- and three-dimensional serpentine channels under conditions where fluid inertia is negligible and across a broad spectrum of polymer relaxation times. Our findings reveal a strong qualitative agreement between the existing experimental results obtained from dilute solutions of flexible polymers in microfluidic devices and the numerical simulations conducted in two and three dimensions using the Oldroyd-B model. Spatial flow observations and statistical analysis of temporal flow features indicate that this purely elastic turbulent flow exhibits nonhomogeneous, non-Gaussian, and anisotropic characteristics across all scales. Additionally, our comparison of two- and three-dimensional simulations demonstrates that the elastic instability is primarily driven by the curvature of the streamlines induced by the flow geometry, rather than the weak secondary flow in the azimuthal direction. Therefore, our two-dimensional numerical simulations successfully replicate, at least qualitatively, the features observed in three-dimensional experiments. Furthermore, spectral analysis suggests that, in comparison to elastic turbulence in the dilute regime, the range of scales for the excited fluctuations is narrower.

An approach for fatigue damage estimation under broadband non-Gaussian random loadings based on the Johnson transformation model

Estimating fatigue damage of marine structures using spectral methods under broadband random loadings is a complex and challenging task. In the presence of non-Gaussian loading, the spectral method typically employs the Hermite polynomial model (HPM) to address the non-Gaussianity. However, the applicability of HPM is limited in certain non-Gaussian cases due to the requirement of monotonicity. This paper adopts the Johnson transformation model (JTM) instead of HPM and integrates it with the non-Gaussian TB method (Ding and Chen, 2015) to overcome the constraints in broadband non-Gaussian fatigue estimation. To verify the performance of the new approach, this paper analyses the transformation accuracy of the two models. Additionally, taking the time-domain fatigue damage obtained by the rainflow counting method and the linear damage accumulation rule as a benchmark, this paper compares the fatigue damage estimates obtained from the spectral methods. The comparative assessment encompasses various spectra, S-N curves, and non-Gaussian characteristics. The results indicate the performance of JTM within the non-Gaussian TB method is comparable to HPM in most cases and even surpasses it in hardening non-Gaussian cases. Given the broader applicability of JTM across ranges of skewness and kurtosis, it emerges a superior option for the non-Gaussian TB method in estimating broadband non-Gaussian fatigue damage. Furthermore, the limitations of the proposed approach are discussed.