Nonstationary Parts Research Articles

A hierarchical modeling strategy for condition monitoring and fault diagnosis of wind turbine using SCADA data

The development and utilization of wind energy can help promote the carbon neutrality. Condition monitoring and fault diagnosis based on supervisory control and data acquisition (SCADA) can effectively improve wind turbine reliability and reduce operation and maintenance (O&M) costs. However, the complex data characteristics (e.g., time-varying and distribution-free) make it difficult to use a unified model to comprehensively evaluate the turbine state. To finely detect and analyze the turbine faulty mode, a hierarchical condition monitoring and fault diagnosis (CMFD) method is proposed, which consists of variable-level and turbine-level. First, the SCADA data is divided into two blocks: the nonstationary part with time-varying distribution, and the stationary part with time-invariant but non-Gaussian distribution. At the variable-level, this paper proposed a local monitoring unit based on sparse cointegration analysis (SCA) and independent component analysis (ICA) to detect faulty samples of different variable blocks. And the fault-related variables are isolated based on the proposed distributed reconstruction. At the turbine-level, the evaluation results from variable-level are fused: on the one hand, Bayesian inference is used to integrate monitoring statistics of stationary and nonstationary parts to comprehensively evaluate the operational state of the turbine; on the other hand, the root fault subsystem is deduced from the fault-related variables isolated from the two parts based on Gaussian process regression. The proposed method is applied in the operation and maintenance of a real wind farm, and the performance advantages are verified by the comparison with popular benchmarks.

Improved BIGRU Model and Its Application in Stock Price Forecasting

In order to obtain better prediction results, this paper combines improved complete ensemble EMD (ICEEMDAN) and the whale algorithm of multi-objective optimization (MOWOA) to improve the bidirectional gated recurrent unit (BIGRU), which makes full use of original complex stock price time series data and improves the hyperparameters of the BIGRU network. To address the problem that BIGRU cannot make full use of the stationary data, the original sequence data are processed using the ICEEMDAN decomposition algorithm to derive the non-stationary and stationary parts of the data and modeled with the BIGRU and the autoregressive integrated moving average model (ARIMA), respectively. The modeling process introduces a whale algorithm for multi-objective optimization, which improves the probability of finding the best combination of parameter vectors. The R2, MAPE, MSE, MAE, and RMSE values of the BIGRU algorithm, ICEEMDAN-BIGRU algorithm, MOWOA-BIGRU algorithm, and the improved algorithm were compared. An average improvement of 14.4% over the original algorithm’s goodness-of-fit value will greatly improve the accuracy of stock price predictions.

Prediction of Long-Term Elbow Flexion Force Intervals Based on the Informer Model and Electromyography

Accurate and long-term prediction of elbow flexion force can be used to recognize the intended movement and help wearable power-assisted robots to improve control performance. Our study aimed to find a proper relationship between electromyography and flexion force. However, the existing methods must incorporate biomechanical models to produce accurate and timely predictions of flexion force. Elbow flexion force is largely determined by the contractile properties of muscles, and the relationship between flexion force and the motor function of muscles has to be thoroughly analyzed. Therefore, based on the investigation on the contributions of different muscles to the flexion force, original electromyography signals were decomposed into non-linear and non-stationary parts. We selected the mean absolute value (MAV) of the non-linear part and the variance of the non-stationary part as inputs for an Informer prediction model that does not require detailed a priori knowledge of biomechanical models and is optimized for processing time sequences. Finally, a long-term flexion force probability interval is proposed. The proposed framework performs well in predicting long-term flexion force and outperforms other state-of-the-art models when compared to experimental results.

Iterative envelope mean fractal dimension filter for the separation of crackles from normal breath sounds

This paper presents a new method of separating pulmonary crackles from normal breath sounds: the iterative envelope mean fractal dimension (IEM-FD) filter. Crackles are an important physiological parameter for evaluating lung condition of an individual and their automatic separation from normal breath sounds can provide an objective way of diagnosing or monitoring different cardiopulmonary diseases. The filter combines the new iterative envelope mean (IEM) method with the established fractal dimension (FD) technique. The IEM method estimates the non-stationary and stationary parts of the lung sound signal and then the FD technique is applied to the estimated non-stationary output of the IEM method for further refining the separation process. The IEM-FD filter is tested using a publicly available dataset and, compared with an established crackle separation technique. The IEM-FD achieves high accuracy for crackle detection in the presence of noise with SNR >= -1 dB for fine crackles and SNR > +1 dB for coarse crackles, and has low computational cost, with minimal under- or over- estimation and good preservation of crackle morphology. The method is shown to have an overall performance suitable for automated analysis to determine accurately the number and characteristics of pulmonary crackles in a recorded lung sound.

A nonlinear full condition process monitoring method for hot rolling process with dynamic characteristic

As a typical complex industrial process, hot rolling process (HRP) is different from chemical process. Strip steels are produced coil by coil, that means there is a long idle period between coils. The rolling speed is very high and the producing time of each coil is usually a few minutes. Previous researches mostly focus on fault detection in loaded condition and very few attempts have been made to exploit the monitoring of idle condition. In order to monitor the whole process, not only the loaded condition, but also the idle one, a novel nonlinear full condition process monitoring model is developed in this work. First, a dissimilarity index (DI) is defined for condition identification and a support data vector description (SVDD) model is established to monitor the idle condition. Second, t-distributed stochastic neighbor embedding (t-SNE) is used to extract nonlinear principal components (NPC) for slow feature analysis (SFA) and cointegration analysis (CA). Nonlinear cointegration analysis (NCA) can reveal the long-run dynamic relations of nonstationary parts, while nonlinear slow feature analysis (NSFA) can extract the latent temporal dynamic and static variations of stationary ones. Finally, the monitoring performance of the proposed model is verified through a real HRP.

COVID-19 Time Series Forecast Using Transmission Rate and Meteorological Parameters as Features

The number of confirmed cases of COVID-19 has been ever increasing worldwide since its outbreak in Wuhan, China. As such, many researchers have sought to predict the dynamics of the virus spread in different parts of the globe. In this paper, a novel systematic platform for prediction of the future number of confirmed cases of COVID-19 is proposed, based on several factors such as transmission rate, temperature, and humidity. The proposed strategy derives systematically a set of appropriate features for training Recurrent Neural Networks (RNN). To that end, the number of confirmed cases (CC) of COVID-19 in three states of India (Maharashtra, Tamil Nadu and Gujarat) is taken as a case study. It has been noted that stationary and nonstationary parts of the features improved the prediction of the stationary and non-stationary trends of the number of confirmed cases, respectively. The new platform has general application and can be used for pandemic time series forecasting.

Vibration Monitoring of Gas Turbine Engines: Machine-Learning Approaches and Their Challenges

In this paper, condition monitoring strategies are examined for gas turbine engines using vibration data. The focus is on data-driven approaches, for this reason a novelty detection framework is considered for the development of reliable data-driven models that can describe the underlying relationships of the processes taking place during an engine's operation. From a data analysis perspective, the high dimensionality of features extracted and the data complexity are two problems that need to be dealt with throughout analyses of this type. The latter refers to the fact that the healthy engine state data can be nonstationary. To address this, the implementation of the wavelet transform is examined to get a set of features from vibration signals that describe the nonstationary parts. The problem of high-dimensionality of the features is addressed by compressing them using the kernel principal component analysis so that more meaningful, lower-dimensional features can be used to train the pattern recognition algorithms. For feature discrimination, a novelty detection scheme that is based on the one-class support vector machine algorithm is chosen for investigation. The main advantage, when compared to other pattern recognition algorithms, is that the learning problem is being cast as a quadratic program. The developed condition monitoring strategy can be applied for detecting excessive vibration levels that can lead to engine component failure. Here, we demonstrate its performance on vibration data from an experimental gas turbine engine operating on different conditions. Engine vibration data that are designated as belonging to the engine’s “normal” condition correspond to fuels and air-to-fuel ratio combinations, in which the engine experienced low levels of vibration. Results demonstrate that such novelty detection schemes can achieve a satisfactory validation accuracy through appropriate selection of two parameters of the one-class support vector machine, the kernel width γ and optimization penalty parameter ν. This selection was made by searching along a fixed grid space of values and choosing the combination that provided the highest cross-validation accuracy. Nevertheless, there exist challenges that are discussed along with suggestions for future work that can be used to enhance similar novelty detection schemes.

Allocation of reliability--redundancy and spares inventory under Poisson fleet expansion

ABSTRACTThis article proposes an integrated product-service model to ensure the system availability by concurrently allocating reliability, redundancy, and spare parts for a variable fleet. In the literature, reliability and inventory allocation models are often developed based on a static installed base. The decision becomes really challenging during new product introduction, as the demand for spare parts is nonstationary due to the fleet expansion. Under the system availability criteria, our objective is to minimize the fleet costs associated with design, manufacturing, and after-sales support. We tackle this reliability--inventory allocation problem in two steps. First, to accommodate the fleet growth effects, the nonstationary spare parts demand stream is modeled as a sum of randomly delayed renewal processes. When the component's failure time is exponential, the mean and variance of the lead time inventory demand are explicitly derived. Second, we propose an adaptive base stock policy against the time-varying parts demand rate. A bisection search combined with metaheuristics is used to find the optimal solution. Numerical examples show that spare parts inventory results in a lower fleet cost under short-term performance-based contracts, whereas reliability--redundancy is preferred for long-term service programs.

Causal and Anti-Causal Segmentation of Voltage Dips in Power Distribution Networks

This paper presents the performance of a joint causal and anti-causal (CaC) segmentation method for automatic location of nonstationary parts of power quality (PQ) events. Accurate time allocation of events and sequences of events is an important to automatically analyze disturbances in the future (smart) grid. The new method developed is based on the cumulative sum (CUSUM) algorithm and is applied to a wide set of power quality events to analyze its performance. The main advantage of CaC segmentation is that the location in time of underlying transitions in the power system is accurately estimated.

Improving BCI performance using SSA and GA

Brain Computer Interface (BCI) systems based on motor imagery tasks have significant usage for disabled people for their life style. Electroencephalogram (EEG) is one of the best approach to obtain human brain signals. The EEG signal requires three stages of preprocessing, feature extraction, and classification to increase signal analysis accuracy. Due to different task of brain, EEG distribution fluctuates (non-stationary), and therefore challenges BCI for many researchers. Recently a method named stationary subspace analysis (SSA) applied to some BCI data by some research teams in order to separate stationary and non-stationary parts of EEG. However they did not obtain significantly results. This method factorizes EEG into its stationary and non-stationary components by dividing signal into number of epochs, and compares their data distributions. In this study, we applied SSA in preprocessing stage into train and test data of the BCI competition dataset of nine healthy people. We applied the Genetic Algorithm (GA) to train our Artificial Neural Network (ANN) classifier. We also inspect different parameters for SSA to improve the performance. Our results indicate significant growth especially for subjects with worse results in other techniques (improving ~40% to ~70%). In addition, the mean of accuracies improves 5% in regard to the winner of competition.

Reducing micro‐Doppler effect in compressed sensing ISAR imaging for aircraft using limited pulses

In compressed sensing (CS)-based inverse synthetic aperture radar (ISAR) imaging for aircraft, the image of the target can be reconstructed using fewer pulses with random pulse repetition intervals than with the conventional range-Doppler (RD) method. However, the micro-Doppler (mD) effect induced by the non-stationary parts of aircraft still causes defocusing as in RD imaging. A method is proposed to reduce the mD effect in CS ISAR imaging. The CS-based short-time Fourier transform is deployed to reconstruct the time-frequency (TF) spectrogram of echoes. An L-statistics-based algorithm is applied to separate the non-stationary scatterers from a rigid main body in the TF domain. Furthermore, the CS algorithm is used to reconstruct the cross-range image of the main body after mD separation. Compared with direct CS imaging without mD removal, a better image can be obtained. The results of both simulated and real data processing demonstrate the validity of the proposed method.

Discriminative Non-Linear Stationary Subspace Analysis for Video Classification.

Low-dimensional representations are key to the success of many video classification algorithms. However, the commonly-used dimensionality reduction techniques fail to account for the fact that only part of the signal is shared across all the videos in one class. As a consequence, the resulting representations contain instance-specific information, which introduces noise in the classification process. In this paper, we introduce non-linear stationary subspace analysis: a method that overcomes this issue by explicitly separating the stationary parts of the video signal (i.e., the parts shared across all videos in one class), from its non-stationary parts (i.e., the parts specific to individual videos). Our method also encourages the new representation to be discriminative, thus accounting for the underlying classification problem. We demonstrate the effectiveness of our approach on dynamic texture recognition, scene classification and action recognition.

Unusually cold and dry winters increase mortality in Australia

Seasonal patterns in mortality have been recognised for decades, with a marked excess of deaths in winter, yet our understanding of the causes of this phenomenon is not yet complete. Research has shown that low and high temperatures are associated with increased mortality independently of season; however, the impact of unseasonal weather on mortality has been less studied. In this study, we aimed to determine if unseasonal patterns in weather were associated with unseasonal patterns in mortality. We obtained daily temperature, humidity and mortality data from 1988 to 2009 for five major Australian cities with a range of climates. We split the seasonal patterns in temperature, humidity and mortality into their stationary and non-stationary parts. A stationary seasonal pattern is consistent from year-to-year, and a non-stationary pattern varies from year-to-year. We used Poisson regression to investigate associations between unseasonal weather and an unusual number of deaths. We found that deaths rates in Australia were 20–30% higher in winter than summer. The seasonal pattern of mortality was non-stationary, with much larger peaks in some winters. Winters that were colder or drier than a typical winter had significantly increased death risks in most cities. Conversely summers that were warmer or more humid than average showed no increase in death risks. Better understanding the occurrence and cause of seasonal variations in mortality will help with disease prevention and save lives.

Hydrodynamic and Diffusion Problems in the Theory of Nonstationary Diffusophoresis of Spherical Particles

The theory of the nonstationary diffusiophoresis of a large spherical solid particle in a viscous gaseous medium is beginning to be constructed. Hydrodynamic and diffusion problems are divided into stationary and strictly nonstationary parts. The solution of the stationary parts of these problems has yielded the final formula for determining the stationary component of the diffusiophoretic velocity of the particle considered. A corresponding formula in the space of Laplace images has been obtained for determining the nonstationary component of the diffusiophoretic velocity of this particle.

The theory of nonstationary thermophoresis of a solid spherical particle

The theory of nonstationary thermophoresis of a solid spherical particle in a viscous gaseous medium is presented. The theory is constructed on the solutions of fluid-dynamics and thermal problems, each of which is split into stationary and strictly nonstationary parts. The solution of the stationary parts of the problems gives the final formula for determining the stationary component of the thermophoretic velocity of this particle. To determine the nonstationary component of the thermophoretic velocity of the particle, the corresponding formula in the space of Laplace transforms is derived. The limiting value theorems from operational calculus are used for obtaining the dependence of the nonstationary component of the thermophoretic velocity of the spherical particle on the strictly nonstationary temperature gradient for large and small values of time. The factors determining the thermophoretic velocity of the particle under investigation are determined.

A noise robust feature extraction algorithm using joint wavelet packet subband decomposition and AR modeling of speech signals

This paper presents a noise robust feature extraction algorithm NRFE using joint wavelet packet decomposition (WPD) and autoregressive (AR) modeling of a speech signal. In opposition to the short time Fourier transform (STFT)-based time–frequency signal representation, wavelet packet decomposition can lead to better representation of non-stationary parts of the speech signal (e.g. consonants). The vowels are well described with an AR model as in LPC analysis. The proposed Root-Log compression scheme is used to perform the computation of the wavelet packet parameters. The separately extracted WPD and AR-based parameters are combined together and then transformed with the usage of linear discriminant analysis (LDA) to finally produce a lower dimensional output feature vector. The noise robustness is improved with the application of proposed wavelet-based denoising algorithm with a modified soft thresholding procedure and time–frequency adaptive threshold. The proposed voice activity detector based on a skewness-to-kurtosis ratio of the LPC residual signal is used to effectively perform a frame-dropping principle. The speech recognition results achieved on Aurora 2 and Aurora 3 databases show overall performance improvement of 44.7% and 48.2% relative to the baseline MFCC front-end, respectively.

A Stochastic Model for the Random Impact Series Method in Modal Testing

A novel stochastic model is developed to describe a random series of impacts in modal testing that can be performed manually or by using a specially designed random impact device. The number of the force pulses, representing the impacts, is modeled as a Poisson process with stationary increments. The force pulses are assumed to have an arbitrary, deterministic shape function, and random amplitudes and arrival times. The force signal in a finite time interval is shown to consist of a wide-sense stationary part and two nonstationary parts. The expectation of the force spectrum is obtained from two approaches. The expectations of the average power densities associated with the entire force signal and the stationary part of it are determined and compared. The analytical expressions are validated by numerical solutions for two different types of shape functions. A numerical example is given to illustrate the advantages of the random impact series over a single impact and an impact series with deterministic arrival times of the pulses in estimating the frequency response function. The model developed can be used to describe a random series of pulses in other applications.

Climate oscillations and abrupt changes in C14 data

The radiocarbon series are analysed by a method of non-linear spectral analysis to detect time intervals of appearance of non-stationary oscillations of large amplitude, and the times of abrupt changes of their oscillation regime. The analysis shows that the most powerful cycles of the spectra can be interpreted in terms of periods (and their respective higher harmonics) of astronomical origin. An intense stationary sinusoid from the spectrum with period T∼6500 yr, the 4th harmonic of the period of equinox precession, correlates with the time variations of the geomagnetic dipole moment. The most powerful non-stationary sinusoid with mean period T=2230 yr, reflects oscillations in C14 data related to the non-dipole part of the geomagnetic field, and correlates with periods of climate warming/cooling. The apparent regularities that can be inferred in the interaction of such two powerful cycles (i.e. stationary and non-stationary parts of the uniform mechanism of the geomagnetic field generation) permit to forecast a tendency of the climate changes. A possible physical mechanism is presented based on a possible transformation, of some signals caused by perturbation of the tidal forces of astronomical origin (that can arise along the orbit of the Earth), into effects that control geophysical systems through small variations of the dissipative parameters of a dynamo system.

Central bank interventions and exchange rates: an analysis with high frequency data

Abstract We use high frequency data for the mark–dollar exchange rate for the period 1992–1995 to evaluate the effects of central bank interventions on the foreign exchange market. We estimate an unobserved component model that decomposes volatility into non-stationary and stationary parts. Stationary components in turn are decomposed into seasonal and non-seasonal intra-day parts. Our results confirm the view that interventions are not particularly effective. The exchange rate moves in the desired direction for only about 50% of the time, and often with a substantial increase in volatility. The model suggests that the two components, which are affected the most by interventions, are the permanent and the stochastic intra-day.

Robust recursive AR speech analysis

In this paper a new robust recursive method of estimating the linear prediction parameters of an auto-regressive speech signal model using weighted least squares with variable forgetting factors (VVFs) is described. The proposed robust recursive least-squares (RRLS) method differs from the conventional recursive least-squares (RLS) method by the insertion of a suitably chosen nonlinear transformation of the prediction residuals. The RRLS algorithm takes into account the contaminated Gaussian nature of the excitation for voiced speech, and the effect of nonlinearity is to assign less weight to the small portions of large residuals so that the spiky excitation will not greatly influence the final AR parameter estimates, while giving unity weight to the bulk of small to moderate residuals generated by the nominal Gaussian distribution. In addition, the VFF is adapted to a nonstationary speech signal by a generalized likelihood ratio algorithm, which accounts for the nonstationarity of a speech signal. The proposed method has a good adaptability to the nonstationary parts of a speech signal, and gives low bias and low variance at the stationary signal segments. The feasibility of the robust approach is demonstrated with both synthesized and natural speech.