Sliding Time Window Research Articles

A Gray Target Calculation–Cloud Gravity Center Health Assessment Method for Gas Turbine Engine

Abstract Since the health status of gas turbine engine is difficult to quantify, which brings great challenges to health assessment and remaining useful life predictions. To solve this problem, a gray target calculation and cloud gravity center (GTC–CGC) health assessment method is proposed. The three-dimensional data normalization process is to characterize the different initial states between samples. The empirical signal-to-noise ratio and entropy weight are used to calculate the subjective and objective weights of health indicators, and convex optimization is used to realize the fusion assessment. The sliding time window method is used to calculate the real-time health state of the gas turbine engine. Finally, health assessment and remaining useful life prediction tests were conducted on turbofan engines using the NASA C-MAPSS dataset. The experimental results show that compared with the self-organizing neural network, the monotonicity, robustness, and trend of the health assessment results were improved by 0.3216, 0.0843, and 0.0355, respectively. The third-order linear regression algorithm was used for remaining useful life prediction, the prediction score of this model is 265, and root mean squared error is 26.1542, which is equivalent to the prediction accuracy of mainstream intelligent prediction methods such as neural network.

Identification of Optimal Starting Time Instance to Forecast Net Blotch Density in Spring Barley with Meteorological Data in Finland

The performance of meteorological data-based methods to forecast plant diseases strongly depends on temporal weather information. In this paper, a data analysis procedure is presented for finding the optimal starting time for forecasting net blotch density in spring barley based on meteorological data. For this purpose, changes in the information content of typically measured weather variables were systemically quantified in sliding time windows and with additionally generated mathematical transformations, namely with features. Signal-to-noise statistics were applied in a novel way as a metric for identifying the optimal starting time instance and the most important features to successfully distinguish between two net blotch densities during springtime itself. According to the results, the information content of meteorological data used in classifying between nine years with and four years without net blotch reached its maximum in Finnish weather conditions on the 41st day from the beginning of the growing season. Specifically, utilising weather data at 41–55 days from the beginning of the growing season maximises successful forecasting potential of net blotch density. It also seems that this time instance enables a linear classification task with a selected feature subset, since the averages of the metrics in two data groups differ statistically with a minimum 68% confidence level for nine days in a 14-day time window.

Real-Time milling tool breakage monitoring based on multiscale standard deviation diversity entropy

Tool breakage monitoring (TBM) during milling is vital for ensuring product quality, equipment and operator safety. This paper proposes a novel real-time TBM method based on multiscale standard deviation diversity entropy (MSDDE). Unlike existing similar works that only validate the effectiveness of the extracted features while ignoring the requirements of real-time detection, the proposed method can extract effective features from short data samples containing only two spindle rotation periods and achieve computational efficiency in milliseconds. Consequently, it is more suitable for application in the real-time detection of milling tool breakage. MSDDE quantifies the dynamic complexity of vibration signals at different time scales, which is capable of reducing the requirement for the number of spindle periods in the sliding time window while comprehensively describing the tool breakage characteristics. The proposed algorithm exhibits a time complexity of O(MN) where M depends on the scale parameter and embedding dimension. Experimental investigations show that the proposed method is able to complete the calculation of feature values within two milliseconds and realize real-time detection of milling tool breakage.

Online health assessment method based on belief rule base with sliding time window considering input correlation and redundancy

This paper develops a new health assessment method based on belief rule base with sliding time window considering correlation and redundancy of input information (BRB-WCR). The new proposed method aims to solve four problems in health assessment of dynamic complex system: lack of system fault data, uncertainty of expert knowledge, correlation of characteristic and high real-time requirement of system. The traditional BRB model can address the first two problems while it assumes that the input characteristics are completely independent. In order to solve the two remaining problems, a new BRB-WCR model is developed, where the correlation and redundancy of characteristics are considered and the sliding time window for online training data is introduced to improve the real time performance of the model. Moreover, to address the uncertainty of expert knowledge, a new optimization model is developed to train the parameters of the BRB-WCR model. To illustrate the effectiveness of the developed method, a case study of inertial navigation platform is conducted.

A secure forecasting-aided state estimation framework for power distribution systems against false data injection attacks

As a critical session of cyber-energy system operation, the state estimation (SE) is becoming more vulnerable to various malicious cyber-attacks like false data injection attacks (FDIAs), which can circumvent the detection mechanism and tamper with the data transmission to cause abnormal state deviation. Driven by this motivation, this paper tries to address the specific issue and propose a novel secure forecasting-aided SE (FASE) framework for the power distribution system with high penetration of renewable energy. The unbalanced distribution system FASE model is constructed first, and then the modeling process of FDIAs with incomplete system information considering limited attacking costs is illustrated. Afterward, a modified noise statistics estimator is combined with an unscented Kalman filter (UKF) to perform FASE and establish a historical state database beforehand. Moreover, a novel security enhancement strategy for FASE against FDIAs based on sliding time window (STW) theory and optimal state forecasting is innovatively proposed to correct the compromised measurements introduced by potential attacks. Comprehensive case studies are conducted on two modified distribution systems and the results demonstrate the effectiveness and credibility of the proposed secure FASE framework.

A two-stage estimation method based on Conceptors-aided unsupervised clustering and convolutional neural network classification for the estimation of the degradation level of industrial equipment

In practical applications, degradation level estimation is often facing the challenge of dealing with unlabeled time series characterized by long-term temporal dependencies, which are typically not properly represented using sliding time windows. Inspired by the idea of representing temporal patterns by a mechanism of neurodynamical pattern learning, called Conceptors, a two-stage method for the estimation of the equipment degradation level is developed. In the first stage, clusters of Conceptors representing similar patterns of degradation within complete run-to-failure trajectories are identified; in the second stage, the obtained clusters are used to supervise the training of a convolutional neural network classifier of the equipment degradation level. The proposed method is applied to a synthetic case study and to two literature case studies regarding bearings degradation level estimation. The obtained results show that the proposed method provides more accurate estimation of the equipment degradation level than other state-of-the-art methods.

The resilience of cryptocurrency market efficiency to COVID-19 shock

We examine the price disorder and informational efficiency of five cryptocurrencies (Bitcoin, BNB, Cardano, Ethereum, and XRP) before and during the COVID-19 pandemic. In this sense, we estimate the permutation entropy and Fisher information measure (FIM). We use these complexity measures to construct the Shannon–Fisher causality plane (SFCP) to map these cryptocurrencies and their respective locations in a two-dimensional plane and then apply the sliding time window approach to study the temporal evolution of informational efficiency. All cryptocurrencies exhibit high but slightly varying informational efficiency during both periods. Cardano was the most efficient cryptocurrency. These results might point to the increasing maturity and lower potential for price predictability, which matter to cryptocurrencies’ usage for liquidity risk diversification strategy.

Disruption of dynamic functional connectivity in children with developmental dyslexia

ABSTRACT Developmental dyslexia (DD) is the most common learning disorder. The alternations of intrinsic functional connectivity (FC) have been frequently reported to be associated with DD. However, previous research has predominately focused on static FC features. Using functional magnetic resonance imaging and a sliding time window approach, this study examined dynamic FC during rest in 22 children with DD and 38 age-matched controls. Results showed that compared to controls, dyslexic children exhibited increased temporal variability of FC in the left inferior frontal gyrus (IFG). Moreover, the higher dynamics of FC was correlated with poorer reading performance, confirming the role of atypical dynamic FC in DD. We found that the altered dynamic FC in dyslexics was associated with problematic regional activity and functional-structural coupling. This study for the first time reveals the aberrant dynamic FC in a pivotal language region, providing novel insights into the neural signature of DD.

Learning Adaptive Grasping From Human Demonstrations

This article studied a learning-based approach to learn grasping policies from teleoperated human demonstrations, which can achieve adaptive grasping using three different neural network (NN) structures. To transfer human grasping skills effectively, we used multisensing state within a sliding time window to learn the state–action mapping. By teleoperating an anthropomorphic robotic hand using human hand tracking, we collected training datasets from representative grasping of various objects, which were used to train grasping policies with three proposed NN structures. The learned policies can grasp objects with varying sizes, shapes, and stiffness. We benchmarked the grasping performance of all policies, and experimental validations showed significant advantages of using the sequential history states, compared to the instantaneous feedback. Based on the benchmark, we further validated the best NN structure to conduct extensive experiments of grasping hundreds of unseen objects with adaptive motions and grasping forces.

NEURAL NETWORK MODELING OF RELAY PROTECTION WITH A TIME DELAY

Modern electric power facilities – stations and high-voltage substations – have become digital objects with the active use of high-speed local networks directly involved in the technological process. Management, analysis and control of information exchange in the digital substation of the power system require the development of new means and approaches. For these purposes, machine learning methods can be used, in particular the apparatus of artificial neural networks (ANN). The paper shows the possibilities of using direct propagation ANNs (multilayer perceptrons) for modeling and identifying anomalies in the operation modes of relay protection with a time delay. The results of training and testing of the ANN are presented on the example of analyzing the operation of the over current protection in the “sliding time window” mode in a three-phase electrical network. The proposed neuroalgorithm and configuration of the ANN can be used to control the modes and accuracy of relay and cybernetic defenses.

RECURRENT USE OF A PERCEPTRON FOR SIGNAL STRUCTURAL ANALYSIS

The paper is devoted to the use of an artificial neural network (ANN) of direct propagation (multilayer perceptron) for signal processing in electrical engineering and electric power industry. It is proposed to use such simple neural networks instead of ANN with a more complex structure (convolutional, recurrent), but within the framework of a sequential recurrent algorithm. This allows checking and controlling the quality of signal processing at each stage of calculations. The proposed algorithm is tested on the example of structural analysis of a signal with nonlinear distortions in a sliding time window. It is shown that the amplitude, frequency and phase of an industrial frequency signal with a high level of harmonics and an aperiodic component can be isolated with an accuracy of units of percent for a time not exceeding units of milliseconds. To increase the accuracy at each step of the calculations, traditional methods can be used, in addition to the ANN: averaging, median smoothing, etc.

Design a regional and multistep air quality forecast model based on deep learning and domain knowledge

Air pollution is an issue across the world. It not only directly affects the environment and human health, but also influences the regional and even global climate by changing the atmospheric radiation budget, resulting in extensive and serious adverse effects. It is of great significance to accurately predict the concentration of pollutant. In this study, the domain knowledge of Atmospheric Sciences, advanced deep learning methods and big data are skillfully combined to establish a novel integrated model TSTM, derived from its fundamental features of Time, Space, Type and Meteorology, to achieve regional and multistep air quality forecast. Firstly, Expectation Maximization and Min-Max algorithms are used for the interpolation and normalization of data. Secondly, feature selection and construction are accomplished based on domain knowledge and correlation coefficient, and then Sliding Time Window algorithm is employed to build the supervised learning task. Thirdly, the features of pollution source and meteorological condition are learned and predicted by CNN-BiLSTM-Attention model, the integrated model of convolutional neural network and Bidirectional long short-term memory network based on Sequence to Sequence framework with Attention mechanism, and then Convolutional Long Short-Term Memory Neural Network (ConvLSTM) integrates the two determinant features to obtain predicted pollutant concentration. The multiple-output strategy is also employed for the multistep prediction. Lastly, the forecast performance of TSTM for pollutant concentration, air quality and heavy pollution weather is tested systematically. Experiments are conducted in Beijing-Tianjin-Hebei Air Pollution Transmission Channel (“2+26” cities) of China for multistep prediction of hourly concentration of six conventional air pollutants. The results show that the performance of TSTM is better than other benchmark models especially for heavy pollution weather and it has good robustness and generalization ability.

Variasi spasial dan temporal nilai-b pada gempa bumi di wilayah Sulawesi Tengah, Gorontalo, dan sekitarnya menggunakan metode robust fitting

This study discusses variation in seismic and tectonic modeled by a Gutenberg-Richter relationship for earthquakes in the Central Sulawesi, Gorontalo, and surrounding areas using the Robust Fitting Method (RFM) with the weight function of Tukey’s bisquare. The declustering process on earthquake data is carried out using the Reasenberg equation. The values for both parameters are analyzed spatially and temporally. In the spatial analysis, the research area is divided into 43 grids. In the temporal analysis, the research area is divided into zone A and zone B. The data grouping is done using a sliding time window method, i.e., grouping 50 earthquake catalogs with 5 overlapping events. The results according to spatial analysis show that the b-values range from 0.38 – 1.19. Areas with low b-values (0.38 – 0.7) occur around the Palu-Koro Fault, i.e., Palu city, Malacca strait, and to Toli-Toli, and also in the northern region of Gorontalo, i.e., the subduction plate of the Sulawesi Sea. Meanwhile, high b-values (0.71 – 1.19) are in the Tomini Bay area which is an area with frequent occurrence of earthquakes but has the small potential to generate large-scale earthquakes. The results of the temporal b-value estimation in zones A and B range between values of 0.38 - 1.25. The b-values appear to decrease before the occurrence of major earthquakes in 1996 and 2018 in zone A. The b-values decreased before the occurrence of major earthquakes in 1990, 1991, 2000, and 2008 in zone B. However, the b-values cannot be used as a precursor before the big earthquake in 1997.
 Keywords: Tukey’s bisquare, Reasenberg equation, Gutenberg-Richter relationship, sliding time window, Robust Fitting Method. MSC2020: 86A15

Altered dynamics of brain spontaneous activity in betel quid dependence chewers

Objective: To investigate the dynamic characteristics of brain spontaneous activity in betel quid dependence (BQD) chewers and its relationship with clinical indexes. Method: This study was conducted in Hainan General Hospital from April to December 2019 and the data of 53 BQD chewers (37 males and 16 females, aged 20 to 58(38±11) years) and 37 healthy controls (HC) (24 males and 13 females, aged 23-57(42±12) years) were collected. All these subjects underwent resting-state functional magnetic resonance imaging (rs-fMRI) scan. The dynamic characteristics of resting fMRI indexes, including dynamic amplitude of low-frequency fluctuations (ALFF), regional homogeneity (ReHo) and degree centrality (DC) of these subjects were calculated using the sliding time window method, parameters such as age and dynamic local consistency were analyzed and compared between the two groups. Pearson correlation was used to analyze the relationship between dynamic indexes, betel quid dependence score (BQDS) and disease duration in BQD group. Results: BQD chewers showed decreased dynamic ALFF in the left orbital prefrontal cortex (0.138±0.041 vs 0.171±0.070), the right temporal pole superior temporal gyrus (0.277±0.070 vs 0.319±0.086) and the right inferior parietal lobule (0.223±0.052 vs 0.259±0.088) than HC (all P<0.05). For regional homogeneity, BQD chewers showed a decrease dynamic ReHo in the right inferior temporal gyrus (0.055±0.008 vs 0.061±0.009), the orbital prefrontal cortex (0.058±0.005 vs 0.063±0.008), the right inferior frontal gyrus (0.081±0.006 vs 0.087±0.011), the right superior occipital gyrus (0.056±0.007 vs 0.062±0.008), the left precentral gyrus (0.068±0.008 vs 0.074±0.008), and the left superior frontal gyrus (0.058±0.008 vs 0.064±0.009) than HC (all P<0.05). BQD chewers showed an increase dynamic ReHo in the right precuneus (0.095±0.009 vs 0.089±0.008) (P<0.05). There was no significant difference in DC between the two groups (all P>0.05). The relationships between three dynamic ALFF and BQDS (r=-0.104, -0.015, -0.119), seven dynamic ReHo and BQDS (r=-0.099, -0.141, -0.055, -0.078, -0.027, -0.111, -0.090), three dynamic ALFF and disease duration (r=-0.122, -0.095, -0.171), and seven dynamic ReHo and disease duration (r=0.242, -0.252, 0.035, 0.056, 0.047, 0.081, 0.169) were not statistically significant (all P>0.05). Conclusions: BQD chewers showed a decrease dynamic ReHo and/or ALFF in multiple brain regions dominated by prefrontal cortex, and an increase dynamic ReHo in the right precuneus.

3D time-dependent scattering about complex shapes using high order difference potentials

We compute the scattering of unsteady acoustic waves about complex three-dimensional bodies with high order accuracy. The geometry of a scattering body is defined with the help of CAD. Its surface is represented as a collection of non-overlapping patches, each parameterized independently by means of high order splines (NURBS). As a specific example, we consider a submarine-like scatterer constructed using five different patches.The acoustic wave equation on the region exterior to the scatterer is solved by first reducing it to a system of Calderon's boundary operator equations. The latter are obtained using the method of difference potentials coupled with a compact fourth order accurate finite difference scheme. When solving the boundary operator equations, we employ Huygens' principle. It allows us to work on a sliding time window of non-increasing duration rather than keep the entire temporal history of the solution at the boundary.The proposed methodology demonstrates grid-independent computational complexity at the boundary and sub-linear complexity with respect to the grid dimension. It efficiently handles complex non-conforming geometries on Cartesian grids with no penalty for either accuracy or stability due to the cut cells. Its performance does not deteriorate over arbitrarily long simulation times. The exact treatment of artificial outer boundaries is inherently built in. Finally, multiple similar problems can be solved efficiently at a low individual cost per problem. This is important when, for example, the boundary condition on the surface changes but the scattering body stays the same.

Wi-Fi-Based Fall Detection Using Spectrogram Image of Channel State Information

Wi-Fi channel state information (CSI)-based fall detection systems have a great potential compared with other alternatives since they are nonintrusive and nonspace limited. However, in the conventional work on Wi-Fi CSI-based fall detection, a phenomenon is commonly observed: the classification performance degrades when data in different environments are used for learning and testing. Nonetheless, when the signal-to-noise-power ratio (SNR) is small, the conventional methods cannot capture features of motion and cannot segment signals accurately. Therefore, there is a need to address these problems in order to build a robust fall detection system. In this article, we propose a spectrogram-image-based fall detection using Wi-Fi CSI. Unlike the conventional method, CSI is segmented with a certain sliding-time window, and then the classifier detects fall by using the spectrogram image generated from the segmented CSI. We use a pretrained convolutional neural network (CNN) optimized for binary classification of the spectrogram images of the fall and nonfall motions. We carried out experiments to evaluate the classification performance of our proposed method against the conventional one by using motion data in two different rooms for learning and testing. As a result, we confirmed that our proposed method outperforms the conventional one and reaches over 0.92 accuracy. In addition, compared with the conventional method, the fall detection performance of our method does not degrade even when using different environment data for learning and testing.

Triple Network Model-Based Analysis on Abnormal Core Brain Functional Network Dynamics in Different Stage of Amnestic Mild Cognitive Impairment.

Amnestic mild cognitive impairment (aMCI) is considered to be a transitional stage between Alzheimer's disease (AD) and normal cognitive state because it has the same clinical symptoms as AD but with lower severity. Studies have confirmed that patients with aMCI are more likely to develop to AD. Although studies on resting state functional connectivity have revealed the abnormal organization of brain networks, the dynamic changes of the functional connectivity across the scans have been ignored. Dynamic functional connectivity is a novel method to reveal the temporal variation of brain networks. This paper aimed to investigate the dynamic characteristics of brain functional connectivity in the early and late phases of aMCI. Based on the "triple network" model, we used the sliding time window approach to construct dynamical functional networks and then analyzed the dynamic characteristics of the functional connectivity across the entire scan. The results showed that patients with aMCI had longer dwell times in weaker network connection than in the strong network. The transitions between different states become more frequent, and the stability of the patient's brain core network deteriorates. This study also found the correlation between the altered dynamic properties of the core functional networks and the patient's clinical Mini-Mental State Examination assessment scale sores. This study revealed that the characteristics of dynamic functional networks constructed by the core cognitive networks varied in distinct ways at different stages of aMCI, which could provide a new idea for exploring the neuro-mechanisms of neurological disorders.

Closed-loop intermittent sensor fault detection for linear stochastic time-delay systems with unknown disturbances

In this paper, the problem of closed-loop intermittent sensor fault (ISF) detection is investigated for a class of linear stochastic time-delay systems with unknown disturbances. A modified unknown input observer (UIO) is proposed to eliminate the influence of delayed errors caused by the discrete-time proportional-integral controller and constant time delay. In order to detect the appearing time and disappearing time of the ISF, a truncated residual is designed by introducing a sliding-time window. Moreover, two hypothesis tests are utilised to set the ISF detection thresholds, and the detectability, false alarm rates as well as missing alarm rates of the ISF are analysed in the framework of statistical analysis. Finally, the effectiveness of the proposed method is validated via a simulation example of a simplified radial flight control system.

Layered monitoring of xylenol tail gas treatment process based on stationary subspace analysis

AbstractTo deal with the inaccurate monitoring caused by the non‐stationary characteristics of the xylenol tail gas treatment process, a layered monitoring scenario utilizing stationary subspace analysis has been presented in the present study. First, principal component analysis (PCA) is applied to establish the upper‐level stationary monitoring model. Then the sliding time window is used to capture the dynamic information of the remaining non‐stationary features, update the model, and establish the lower non‐stationary monitoring model. Next, the Bayesian information criterion is employed to build the global monitoring model according to the detection results of the upper and lower layers. Finally, the industrial data collected from the industrial boiler is used to evaluate the effectiveness and applicability of the proposed method. It is demonstrated that layered monitoring has higher accuracy than conventional non‐stationary process monitoring techniques. Moreover, it can capture the faults in the stationary features masked by the non‐stationary features.

Dynamic Relief Items Distribution Model with Sliding Time Window in the Post-Disaster Environment

In smart cities, relief items distribution is a complex task due to the factors such as incomplete information, unpredictable exact demand, lack of resources, and causality levels, to name a few. With the development of Internet of Things (IoT) technologies, dynamic data update provides the scope of distribution schedule to adopt changes with updates. Therefore, the dynamic relief items distribution schedule becomes a need to generate humanitarian supply chain schedules as a smart city application. To address the disaster data updates in different time periods, a dynamic optimised model with a sliding time window is proposed that defines the distribution schedule of relief items from multiple supply points to different disaster regions. The proposed model not only considers the details of available resources dynamically but also introduces disaster region priority along with transportation routes information updates for each scheduling time slot. Such an integrated optimised model delivers an effective distribution schedule to start with and updates it for each time slot. A set of numerical case studies is formulated to evaluate the performance of the optimised scheduling. The dynamic updates on the relief item demands’ travel path, causality level and available resources parameters have been included as performance measures for optimising the distributing schedule. The models have been evaluated based on performance measures to reflect disaster scenarios. Evaluation of the proposed models in comparison to the other perspective static and dynamic relief items distribution models shows that adopting dynamic updates in the distribution model cover most of the major aspects of the relief items distribution task in a more realistic way for post-disaster relief management. The analysis has also shown that the proposed model has the adaptability to address the changing demand and resources availability along with disaster conditions. In addition, this model will also help the decision-makers to plan the post-disaster relief operations in more effective ways by covering the updates on disaster data in each time period.