Nonstationary Behavior Research Articles

Population Prevalence and Trends of Oral Clefts in Colombia: Analysis by Departments.

Determine the population prevalence and trends of cleft lip and/or palate (CL/P) by department for Colombia in the period 2009 to 2015. Prevalence study based on Individual Registry of Health Services in general population from 2009 to 2015. All people diagnosed with CL/P were included for all ages, type of diagnosis, and any type of health services in the mentioned period. The prevalence was calculated by period and point for each year, for each department and according to the type of cleft. Stationarity on time series was evaluated using (Dickey-Fuller) and (Phillips-Perron). A trend test was applied to estimate whether the increase in prevalence was significant in the period. The trend test used was a Poisson regression. A total 15 225 people with CL/P were identified, where 53.3% were men. The national period prevalence of CL/P is 3.37 per 10 000 with upward trend (prevalence ratio = 1.34, P = .05) and nonstationary behavior. The national period prevalence of cleft lip is 0.93 per 10 000, cleft palate 1.17 per 10 000, and cleft lip and palate (CLP) 1.26 per 10 000, where CLP is subclassify into unilateral CLP (0.83 per 10 000) and bilateral CLP (0.43 per 10 000). At the departmental level, the highest CL/P prevalence is Guaviare (11.2), followed by Guainía (8.4) and the lowest Quindío (0.49). In Colombia, the national period prevalence of CL/P is 3.37 per 10 000 with upward trend at national level indicates an increase in prevalence from 2009 to 2015. The 77% of the total CL/P population are infant or adolescent. Geographically, the central region has the highest availability of technologies.

An ensemble of a boosted hybrid of deep learning models and technical analysis for forecasting stock prices

For several years the modeling as well as forecasting of the prices of stocks have been extremely challenging for the business community and researchers as a result of the existence of noise in samples and also the non-stationary behaviour of information samples. Notwithstanding these drawbacks with improved deep learning, it is now possible to design schemes that will efficiently perform the feature learning task. For this work, we proposed a brand-new end to end algorithm labeled EHTS toward solving the stock price forecasting problem. The AB-CNN and CB-LSTM modules extract features from the stock price dataset and soon after amalgamating the results. Thus, the output of the concatenation stage was feed into the concluding stage which is a stand-alone MLP module. The inclusion of the LSTM and Attention Mechanism in our architecture is to extract long-range and exceptionally long-term stock price information. We experiment the proposed algorithm on two popular stocks both from the NYSE stock market namely “Johnson & Johnson” code-named, “JNJ” and the Bank of America (BAC). In terms of the rMSE, MAE and MAPE error metrics, our proposed scheme gives the lowest error value in all for all datasets. Also, five percentage training window sizes are experimented and EHTS outperforms all the baseline schemes for the different window sizes in all the two datasets with the 70% window size having the highest performance. In terms of number of epochs, EHTS uses the lowest number of epochs for training than the other schemes in all the datasets. Finally, we as well study our stock’s information to point out short-range trading opportunities by performing simulations on our stock price data. The metrics considered in the simulation are as follows: Moving Average (MA), Moving Average Convergence Divergence (MACD) curve, MACD histogram, Signal line, Relative Strength Index (RSI), Returns (R), Annual Returns (AR), Sharpe Ratio (SR), Annual Volatility (V), Maximum DrawDown (MDD) and Daily WinningRate (DWR). For all the aforementioned metrics, EHTS performs better than the baselines. Experimental results revealed that our proposed scheme outperformed the stand-alone deep learning schemes, statistical algorithms, and machine learning models from the beginning to the end.

Joint modelling wind speed and power via Bayesian Dynamical models

The relationship of dependence between wind speed and wind power variables has a degree of complexity that has motivated several scientific studies over the years. Much of this research seeks to understand the stochastic nature of both phenomena, either for the purpose of marginal analysis or for joint analyses, aiming to improve prediction of wind power. The present study proposes three dynamic Bayesian models for wind energy that account for the complexity of both variables via hierarchical structures such as temporal dependence, nonstationary behaviour, and truncation of power due to turbine specifications. This hierarchy considers wind energy modelling conditioned on wind speed and a marginal model for wind speed. This approach allows joint analysis to be carried out via univariate time series modelling. Analysis of a rich dataset from Bahia state (Brazil) indicates that the proposed models are accurate for both short-term and long-term wind power forecasts.

MLTF: Model less time-series forecasting

Multi-step ahead long term forecasting remains a pertinent challenge in time series literature due to non-stationary behaviour of real-world data. Predominantly most traditional time series models are parametric in nature and they use the predicted values to generate forecast for future time steps. This leads to error accumulation in each step of the forecasting horizon which causes increasingly poorer forecast in long-term. Other than the problem of error accumulation, most parametric algorithms also require significant pre-processing, hyper-parameter tuning, training and post-processing which can often put high computational burden on the system. Therefore, this paper proposes, Model Less Time-series Forecasting (MLTF), a non-parametric approach for forecasting which does not require any pre-processing or traditional training (i.e. Backpropagation). MLTF is a non-parametric method which uses statistical representations such as trend, linearity, entropy etc. to cluster series from a pre-defined repository and the series from same cluster are tagged as similar series. The trajectory of the target series is extracted from these similar series after applying an adaptive re-sampling technique. There is minimal training involved in MLTF, therefore this framework is computationally very efficient. The model-less nature also enables it to not suffer from error accumulation in long-horizon forecast. MLTF is validated empirically with a rich set of experiments involving M1, M3 competition dataset, Electricity, Volatility and COVID-19 data (over 4500 independent uni-variate series of different frequencies i.e. Hourly, Daily, Monthly, Quarterly and Yearly). The experiments demonstrate that, MLTF is significantly faster while being similar (or better) in terms of forecasting accuracy than the state-of-the-art DL methods and other non-parametric time series model.

Recurrence Quantitative Analysis of Wavelet-Based Surrogate Data for Nonlinearity Testing in Heart Rate Variability.

Exploring the presence of nonlinearity through surrogate data testing provides insights into the nature of physical and biological systems like those obtained from heart rate variability (HRV). Short-term HRV time series are of great clinical interest to study autonomic impairments manifested in chronic diseases such as the end stage renal disease (ESRD) and the response of patients to treatment with hemodialysis (HD). In contrast to Iterative Amplitude Adjusted Fourier Transform (IAAFT), the Pinned Wavelet Iterative Amplitude Adjusted Fourier Transform (PWIAAFT) surrogates preserve nonstationary behavior in time series, a common characteristic of HRV. We aimed to test synthetic data and HRV time series for the existence of nonlinearity. Recurrence Quantitative Analysis (RQA) indices were used as discriminative statistics in IAAFT and PWIAAFT surrogates of linear stationary and nonstationary processes. HRV time series of healthy subjects and 29 ESRD patients before and after HD were tested in this setting during an active standing test. Contrary to PWIAAFT, linear nonstationary time series may be erroneously regarded as nonlinear according to the IAAFT surrogates. Here, a lower proportion of HRV time series was classified as nonlinear with PWIAAFT, compared to IAAFT, confirming that the nonstationarity condition influences the testing of nonlinear behavior in HRV. A contribution of nonlinearity was found in the HRV data of healthy individuals. A lower proportion of nonlinear time series was also found in ESRD patients, but statistical significance was not found. Although this proportion tends to be lower in ESRD patients, as much as 60% of time series proved to be nonlinear in healthy subjects. Given the important contribution of nonlinearity in HRV data, a nonlinear point of view is required to achieve a broader understanding of cardiovascular physiology.

On the capability of preprocessing techniques for suspended sediment load prediction using artificial intelligence methods

ABSTRACT With growing interest in global environmental change, it is crucial to have a reliable model to accurately predict the suspended sediment load (SSL). This study investigates the efficiency and capability of two time-series data decomposition algorithms, namely empirical mode decomposition (EMD) and ensemble EMD (EEMD), to address the complexity and non-stationary behaviour of the monthly SSL time series data. These algorithms were coupled with three artificial intelligence models – gene expression programming (GEP), extreme learning machine (ELM), and extreme gradient boosting (XGBoost) – to predict SSL at two stations in Tajan Basin, northern Iran. The results indicate that EEMD, with coefficient of determination values higher than 0.9, performed better than EMD when coupled with XGBoost as the ensemble model. To verify the the predicted SSL data, Monte Carlo simulation was used. The results confirmed the lower uncertainty of the EEMD-XGBoost model in predicting SSL at a monthly scale.

Estimation of Non-Stationary Behavior in Annual and Seasonal Surface Freshwater Volume Discharged into the Gorgan Bay, Iran

Estimation of Non-Stationary Behavior in Annual and Seasonal Surface Freshwater Volume Discharged into the Gorgan Bay, Iran

Analytics for IoT‐Enabled Human–Robot Hybrid Sortation: An Online Optimization Approach

Motivated by a realworld practice of a China Post sortation center, this study considers the deployment of Internet of Things (IoT) technology to improve the efficiency of a human–robot hybrid sortation system. In this system, IoT technology enables responsive adjustment of the manual capacity to alleviate the congestion effect of increasing parcel flow on robotic sorting efficiency. We begin with a predictive analysis of a real‐life dataset to quantify the congestion effect on robotic sorting efficiency, and examine the nonstationary behavior of the parcel flow process. Subsequently, we design an online algorithm using IoT advance information (which refers to advance observation of some parcel flows before their actual arrivals) for efficiently adjusting the manual capacity. We develop theoretical guarantee for the effectiveness of the online algorithm by bounding its regret, and also demonstrate that the long‐term expected regret rate is zero under mild conditions. Via extensive simulation experiments, we find that our online algorithm outperforms the China Post sortation center's current policy and conventional Markov dynamic program in terms of computational efficiency and solution quality. The simulation results also demonstrate that the value of IoT technology to the sortation center can be significant, and shed insights on IoT investment by revealing the diminishing return of expanding the horizon of IoT advance information.

Maximum Visibility: A Novel Approach for Time Series Forecasting Based on Complex Network Theory

This article presents Maximum Visibility Approach (MVA), a new time series forecasting method based on the Complex Network theory. MVA initially maps time series data into a complex network using the visibility graph method. Then, based on the similarity measures between the nodes in the network, MVA calculates the one-step-ahead forecasts. MVA does not use all past terms in the forecasting process, but only the most significant observations, which are indicated as a result of the autocorrelation function. This method was applied to five different groups of data, most of them showing trend characteristics, seasonal variations and/or non-stationary behavior. We calculated error measures to evaluate the performance of MVA. The results of statistical tests and error measures revealed that MVA has a good performance compared to the accuracy obtained by the benchmarks considered in this work. In all cases, MVA surpassed other forecasting methods in Literature, which confirms that this work will contribute to the field of time series forecasting not only in the theoretical aspect, but also in practice.

Detection of Cyber Attacks on a Water Treatment Process

This paper investigates the detection of cyber attacks in a kind of critical infrastructure, the water treatment processes. For this purpose, the real data from the secure water treatment testbed (SWaT) in the iTrust Centre for Research in Cyber Security at the Singapore University of Technology and Design have been used in the investigation. More than 30 different attacks happen on the communication network between the programmable logic controller (PLC) and the sensors/actuators. After understanding the basic working mechanism of the SWaT process, an attack detection system is developed systematically, which consists of a limit value check and safety rule unit, a model-based monitoring unit and a data-driven monitoring unit. To cope with the nonstationary behavior of some water chemical properties such as pH and oxidation reduction potential, the recursive principal component analysis (RPCA) is applied in the data-driven part. It is shown how to divide the SWaT process suitably to integrate the model-based and data-driven monitoring for the SWaT process. The comparison shows that the proposed attack detection system outperforms other existing attack detection systems for the SWaT process.

The Mansurov effect: Statistical significance and the role of autocorrelation

The Mansurov effect is related to the interplanetary magnetic field (IMF) and its ability to modulate the global electric circuit, which is further hypothesized to impact the polar troposphere through cloud generation processes. We investigate the connection between IMFBy-component and polar surface pressure by using daily ERA5 reanalysis for geopotential height since 1980. Previous studies produce a 27-day cyclic response during solar cycle 23 which appears to be significant according to conventional statistical tests. However, we show here that when statistical tests appropriate for strongly autocorrelated variables are applied, there is a fairly high probability of obtaining the cyclic response and associated correlation merely by chance. Our results also show that data from three other solar cycles produce similar cyclic responses as during solar cycle 23, but with seemingly random offset in respect to the timing of the signal. By generating random normally distributed noise with different levels of temporal autocorrelation and using the real IMFBy-time series as forcing, we show that the methods applied to support the Mansurov hypothesis up to now are highly susceptible to random chance as cyclic patterns always arise as artifacts of the methods. The potential non-stationary behavior of the Mansurov effect makes it difficult to achieve solid statistical significance on decadal time scales. We suggest more research on, e.g., seasonal dependence of the Mansurov effect to understand better potential IMF effects in the atmosphere.

A problem of non – linear deformation of five–layer conical shells with allowance for discrete ribs

A problem of non – linear deformation of multiplayer conical shells with allowance for discrete ribs under non – stationary loading is considered. The system of non – linear differential equations is based on the Timoshenko type theory of rods and shells. The Reissner’s variational principle is used for deductions of the motion equations. An efficient numerical method with using Richardson type finite difference approximation for solution of problems on nonstationary behaviour of multiplayer shells of revolution with allowance distcrete ribs which permit to realize solution of the investigated wave problems with the use of personal computers. As a numerical example, the problem of dynamic deformation of a five-layer conical shell with rigidly clamped ends under the action of an internal distributed load was considered.

Time‐varying frequency model for electric arc furnaces

Accurate modelling of the Electric arc furnace (EAF) is necessary to study its impact on the power systems. Here, a new model for the EAF is presented in which both non-linearity and time-varying characteristics are taken into account. The model consists of seven parallel current sources which represent the fundamental and harmonic currents. Magnitude and phase of the fundamental harmonic and magnitude of the second to seventh harmonics are considered as time-varying which are updated every 0.01 s. To do so, the actual data recorded from eight EAFs in the Mobarakeh steel company (MSC), Isfahan, Iran are used. Autoregressive moving average (ARMA) models are employed to derive the parameters of the time-varying model. Also, different orders and coefficients of the ARMA model are considered at each run of the model, which enables the proposed model to predict the non-stationary and stochastic behaviours of the EAF over long-time periods. It is shown that the proposed combined time-varying-harmonic model can even model the generated inter-harmonics. To evaluate the accuracy of the proposed model, the derived power spectral density (PSD) of the EAF time-varying parameters, the instantaneous flicker, and short-term flicker by the proposed model are compared with the actual recorded data.

Fuzzy Mathematics-Based Outer-Loop Control Method for Converter-Connected Distributed Generation and Storage Devices in Micro-Grids

The modern changes in electric systems present new issues for control strategies. When power converters and distributed energy resources are included in the micro-grid, its model is more complex than the simplified representations used, sometimes losing essential data. This paper proposes a unified fuzzy mathematics-based control method applied to the outer loop of a voltage source converter (VSC) in both grid-connected and islanded modes to avoid using simplified models in complex micro-grids and handle the uncertain and non-stationary behaviour of nonlinear systems. The proposed control method is straightforwardly designed without simplifying the controlled system. This paper explains the design of a fuzzy mathematics-based control method applied to the outer-loop of a VSC, a crucial device for integrating renewable sources and storage devices in a micro-grid. Simulation results validated the novel control strategy, demonstrating its capabilities for real field applications.

Bitcoin daily close price prediction using optimized grid search method

Abstract Cryptocurrencies are digital assets that can be stored and transferred electronically. Bitcoin (BTC) is one of the most popular cryptocurrencies that has attracted many attentions. The BTC price is considered as a high volatility time series with non-stationary and non-linear behavior. Therefore, the BTC price forecasting is a new, challenging, and open problem. In this research, we aim the predicting price using machine learning and statistical techniques. We deploy several robust approaches such as the Box-Jenkins, Autoregression (AR), Moving Average (MA), ARIMA, Autocorrelation Function (ACF), Partial Autocorrelation Function (PACF), and Grid Search algorithms to predict BTC price. To evaluate the performance of the proposed model, Forecast Error (FE), Mean Forecast Error (MFE), Mean Absolute Error (MAE), Mean Squared Error (MSE), as well as Root Mean Squared Error (RMSE), are considered in our study.

Event-based analysis of wetland hydrologic response in the Prairie Pothole Region

Previous studies in the Prairie Pothole Region mainly assessed wetland hydrologic function at seasonal, annual and decadal scales. While many studies have looked at water balance dynamics and local flow generation processes in wetlands, no study looked at hydrological dynamics in response to individual rainfall events in prairie pothole wetlands (here after referred as wetlands) across a gradient of alteration. The present study aimed to investigate: (1) the most important metrics needed to characterize the spatial variability of wetland hydrologic response to rainfall-runoff events; (2)the temporal variability of individual wetland hydrologic response; (3) the spatial and temporal variability of wetland-stream interaction; and (4) the temporal persistence of various spatial controls on individual wetland hydrologic response characteristics. High-frequency water level data was collected over two years for ten intact, three consolidated, and seven drainage ditches associated with fully drained wetland(s), as well as a creek located in southwestern Manitoba, Canada. The hydrologic response of the studied wetlands to individual rainfall-runoff events was characterized using a range of metrics. Several data analysis methods were used, including principal component analysis, graphical assessments of wetland-stream hysteresis dynamics, and correlations analyses between wetland response metrics and spatial characteristics. Results suggested that wetland alteration status (i.e., drained versus intact, open-water wetlands) plays an important role in explaining differences in the event-scale hydrologic behaviour of wetlands. Climatic and antecedent storage conditions (i.e., surface and subsurface storage in wetland basin) also had a strong influence on the hydrologic responses of wetlands during individual rainfall-runoff events and appeared to override the influence of spatial controls such as wetland area, volume or catchment area. Antecedent storage also seemed to be the driving factor of wetland-stream interactions. A lack of persistent correlations between wetland spatial characteristics and response metrics was observed and suggested nonstationary wetland hydrological behaviours and controls, a conclusion that has significant implications for wetland classification and modelling.

Quali-quantitative characterization of biogas with the temporal behavior of organic load on wastewater treatment plant with upflow anaerobic sludge blanket reactors through measurement in full‑scale systems

This study aims to present the time behavior of wastewater flow parameters, organic matter, biogas flow, biogas composition, and its relations, measured through online sensors, in a municipal wastewater treatment plant (WWTP) operating full-scale upflow anaerobic sludge blanket (UASB) reactors, installed in the south of Brazil. WWTP has online measurement devices to evaluate some physicochemical variables of the sewage and the biogas. The COD analyzer (UV– Vis probe), ultrasonic flow meter, biogas flow meter, and biogas composition analyzer were the equipment used. The monitoring occurred for two time periods each of 72 h and one time period for 48 h in the year 2018. Data were checked with descriptive statistics, data independence was checked through the autocorrelation Box– Ljung test, normality behavior was checked with several tests (Shapiro– Wilk, Kolmogorov–Smirnov, Lilliefors, Anderson–Darling, D’Agostino K2, and Chen–Shapiro), and Spearman’s correlation coefficient was used to evaluate the correlations among the parameters. The mean sewage flow was 345 ± 120 L.s-1; removed organic load was, in average, 48%; biogas quality values were 82.32% ± 3.62% v/v (CH4), 2.66% ± 1.19% v/v (CO2), and 3453 ± 1268 ppm (H2S); and the production per capita obtained was 4.51 ± 1.65 NL.hab-1.d-1. It was estimated an electric power generation of 3118.6 kWh.d-1, which is equivalent to an installed power of 130 KW. The behavior of removed organic load and biogas flow (Nm3.h-1), produced in the treatment plant, showed variable, periodic, and nonstationary time behavior.

Record tests to detect non-stationarity in the tails with an application to climate change

The analysis of trends and other non-stationary behaviours at the extremes of a series is an important problem in global warming. This work proposes and compares several statistical tools to analyse that behaviour, using the properties of the occurrence of records in i.i.d. series. The main difficulty of this problem is the scarcity of information in the tails, so it is important to obtain all the possible evidence from the available data. First, different statistics based on upper records are proposed, and the most powerful is selected. Then, using that statistic, several approaches to join the information of four types of records, upper and lower records of forward and backward series, are suggested. It is found that these joint tests are clearly more powerful. The suggested tests are specifically useful in analysing the effect of global warming in the extremes, for example, of daily temperature. They have a high power to detect weak trends and can be widely applied since they are non-parametric. The proposed statistics join the information of M independent series, which is useful given the necessary split of the series to arrange the data. This arrangement solves the usual problems of climate series (seasonality and serial correlation) and provides more series to find evidence. These tools are used to analyse the effect of global warming on the extremes of daily temperature in Madrid.

Recent Advances in Dynamic Modeling and Process Control of PVA Degradation by Biological and Advanced Oxidation Processes: A Review on Trends and Advances

Polyvinyl alcohol (PVA) is an emerging pollutant commonly found in industrial wastewater, owing to its extensive usage as an additive in the manufacturing industry. PVA’s popularity has made wastewater treatment technologies for PVA degradation a popular research topic in industrial wastewater treatment. Although many PVA degradation technologies are studied in bench-scale processes, recent advancements in process optimization and control of wastewater treatment technologies such as advanced oxidation processes (AOPs) show the feasibility of these processes by monitoring and controlling processes to meet desired regulatory standards. These wastewater treatment technologies exhibit complex reaction mechanisms leading to nonlinear and nonstationary behavior related to variability in operational conditions. Thus, black-box dynamic modeling is a promising tool for designing control schemes since dynamic modeling is more complicated in terms of first principles and reaction mechanisms. This study seeks to provide a survey of process control methods via a comprehensive review focusing on PVA degradation methods, including biological and advanced oxidation processes, along with their reaction mechanisms, control-oriented dynamic modeling (i.e., state-space, transfer function, and artificial neural network modeling), and control strategies (i.e., proportional-integral-derivative control and predictive control) associated with wastewater treatment technologies utilized for PVA degradation.

A Nonstationary Wind Speed Frequency Model over South Korea: In the Context of Bayesian Mixture Distribution Model

Kim, H.-J.; Kim, Y.-T.; Kim, T.-W., and Kwon, H.-H., 2021. A nonstationary wind speed frequency model over South Korea: In the context of Bayesian mixture distribution model. In: Lee, J.L.; Suh, K.-S.; Lee, B.; Shin, S., and Lee, J. (eds.), Crisis and Integrated Management for Coastal and Marine Safety. Journal of Coastal Research, Special Issue No. 114, pp. 196–200. Coconut Creek (Florida), ISSN 0749-0208. It has been well recognized that extreme wind speed processes often feature nonstationary behavior, which may not be effectively modeled within a stationary frequency modeling framework. Furthermore, univariate Gumbel distribution has been commonly used for wind speed frequency analysis in Korea. However, the distributional changes in extreme wind speed have been globally observed, including Korea. More specifically, the univariate Gumbel distribution based wind speed frequency analysis often failed to describe multimodal behaviors which are mainly influenced by distinct climate conditions. In this perspective, this study explores a mixture distribution based nonstationary frequency (MDNF) model in a changing climate within a Bayesian framework. It was found that the MDNF model can effectively account for the time-varying moments (i.e., mean and variance) in a two-component mixture distribution. The MDNF model showed more robust results for describing the upper tail of the distribution, which plays a crucial role in estimating the design wind speed.