Search

Fault diagnosis for rolling elements in rotating machinery persistently receives high research interest due to the said machinery's prevalence in a broad range of applications. State-of-the-art methods in such setups focus on effective identification of faults that usually involve a single component while rejecting noise from limited sources. This article studies the data-based diagnosis of mixed faults coming from multiple components with an emphasis on model robustness against a wide spectrum of external perturbation. A dataset is collected on a rotor and bearing system by varying the levels and types of faults in both the rotor and bearing, which results in 48 machine health conditions. A duplet classifier is developed by combining two 1-D convolutional neural networks (CNNs) that are responsible for the diagnosis of the rotor and bearing faults, respectively. Experimental results show that the proposed classifier can reliably identify the onset and nature of mixed faults. In addition, one-vs-all classifiers are built using the features generated by the developed 1-D CNNs as predictors to recognize previously unlearned fault types. The effectiveness of such classifiers is demonstrated using data collected from four new fault types. Finally, the robustness and ability to reject external perturbation of the duplet classification model are analyzed using kernel density estimation. The code for the proposed classifiers is available at https://github.com/siyuanc2/machine-fault-diag .

In traditional brain-computer interfaces (BCIs), using only a certain type of feature or a simple mixture of different features cannot meet the requirements for high performance in classification. Moreover, a simple mixture of various features might lead to information redundancies and thus increase the computational complexity. In this paper, we studied the feasibility of integrating two kinds of features, which showed opposite variation trends as the memory load levels increase, into a single fusion feature. We also proposed a feature fusion framework based on non-invasive electroencephalography to classify the memory load levels and estimate the workload for a series of challenging working memory (WM) tasks (involving delayed match-to-sample tasks) on a single-trial basis. A novel fusion feature called spectral entropy/Lempel-Ziv complexity ( SEn / LZC ) was proposed to classify three memory load levels. The results showed that the generalization of the support vector machine (SVM) with SEn / LZC was significantly higher than the generalization of an SVM with four other types of feature, namely SEn , LZC, SEn & LZC and LZC / SEn . The findings suggested that the proposed fusion feature could act as a biomarker to successfully distinguish different load levels and that the constructed framework could achieve consistency between optimal cognitive performance and fusion features. In addition, the proposed fusion framework could provide a new method of successfully promoting the classification generalization of BCI and implicitly evaluating the mental workload.

Disclosing the complex relationships effectively between paired brain regions played a significant role in measuring the brain functional connectivity and exploring brain topological structures. Even though Pearson correlation coefficient (PCC) has been widely used to construct functional brain networks in the previous studies, it was mainly sensitive to linear associations. Therefore, maximal information coefficient (MIC) was first utilized to make up this weakness of PCC to construct electroencephalography (EEG) connectivity in the current study. The simulation results showed that MIC could capture certain relationships which PCC failed to detect. Furthermore, brain network properties changed with various thresholds under the resting-state EEG, and the comparison analysis of network properties illustrated that MIC and PCC could capture different aspects of connections between paired brain regions. These findings indicated that MIC could be a complementary method of PCC for the construction of scalp resting-state EEG connectome and provided a novel tool to reveal the potential mechanisms of brain networks.

N2pc is defined as a negative event-related potential component that appears after about 250 ms at posterior electrodes contralateral to a target’s location in visual search, which can be used to measure attentional shifts between bilateral visual hemifields and locate the spatial location of lateral targets. However, the waves between the left and right hemispheres elicited by lateral targets usually exhibit a small amplitude difference and strong synchronicity, which may lead to low classification performance. Therefore, the present study explored the feasibility of a multiple correlated components analysis (MCORCA) methods to identify the lateral targets in visual search tasks with a single trial, which could weight the target signals by spatial filters to enlarge the amplitude difference between bilateral hemispheres and extract the linear combinations of multiple channels across trials with an optimal subset of correlated components to avoid the loss of efficient information. The classification rate achieved 82% with a single short-duration trial when using the proposed method with Leave-one-out-cross-validation (LOOCV). The findings demonstrated that the MCORCA-based methods could be used to improve the classification performance for the N2pc-based brain-computer interfaces (BCI) in visual search.

According to the separability criterion of overall scale Class Distance there will be a fault cluster indiscernibility problem when this diagnosis is used in the process of subway bearing weak fault diagnosis. Aiming at this problem, this paper presents a diagnosis method based on Multi-Scale Class Distance (MSCD). In the basis of traditional criterion which focused on the classes distance within single overall scale class, we use the distance measure of multi-scale class as the separability criterion, extract the class of fault from the sample and calculate the euclidean distance between classes. Then classify the sample into several sub-samples, and use these sub-samples as a whole sample to carry on the separability criterion calculation of the next scale. Repeat this process until the multi-scale classification is finished and obtain the sub-samples with best separability. We can get the vibration signals of bearing weak fault in different working conditions from the subway gearbox test bench, and use the method proposed by this paper to carry on weak fault diagnosis. The result shows that this method can achieve the classification of bearing weak fault. Comparing with the traditional PCA classification result based on single scale, this method can obtain a fault classification result with high separability.

In the current study, we utilized EEG coherence and complex brain network to study the changes of working memory (WM) and constructed differential statistical networks under alpha rhythms before and after training. The results showed that the long-range frontoparietal and frontooccipital interactions during WM retention involved in the alpha frequency network. The findings revealed that the connections between neurons varied to complete the efficient transmission and processing of information, indicating the neural plasticity before and after WM training from the network level.

Previous studies have shown that different frequency band oscillations are associated with cognitive processing such as working memory (WM). Electroencephalogram (EEG) coherence and graph theory can be used to measure functional connections between different brain regions and information interaction between different clusters of neurons. At the same time, it was found that better cognitive performance of individuals indicated stronger small-world characteristics of resting-state WM networks. However, little is known about the neural synchronization of the retention stage during ongoing WM tasks (i.e., online WM) by training on the whole-brain network level. Therefore, combining EEG coherence and graph theory analysis, the present study examined the topological changes of WM networks before and after training based on the whole brain and constructed differential networks with different frequency band oscillations (i.e., theta, alpha, and beta). The results showed that after WM training, the subjects’ WM networks had higher clustering coefficients and shorter optimal path lengths than before training during the retention period. Moreover, the increased synchronization of the frontal theta oscillations seemed to reflect the improved executive ability of WM and the more mature resource deployment; the enhanced alpha oscillatory synchronization in the frontoparietal and fronto-occipital regions may reflect the enhanced ability to suppress irrelevant information during the delay and pay attention to memory guidance; the enhanced beta oscillatory synchronization in the temporoparietal and frontoparietal regions may indicate active memory maintenance and preparation for memory-guided attention. The findings may add new evidence to understand the neural mechanisms of WM on the changes of network topological attributes in the task-related mode.

Accurate prediction traffic flow is one of the most critical works of the intelligent transport system (ITS). Accurate prediction results can provide better conditions for traffic guidance, management, and control. However, many existing traffic flow prediction methods are not particularly satisfactory in practical applications. In this paper, the stack auto-encoder (SAE) and long short-term memory network (LSTM) are combined for traffic flow prediction, in which SAE is used to obtain spatial features, while LSTM extracts temporal features of traffic flow. Then, the features from SAE and LSTM are combined to predict the traffic flow state. The real-time traffic flow data from Beijing is used to evaluate the performance of the proposed method. Experimental results show that the performance of the proposed method is better than some well-known prediction models.

In this paper, we present FigExplorer, a novel general system that supports the retrieval and exploration of research article figures. Specifically, FigExplorer can support 1) figure retrieval using keyword queries, 2) exploration of related figures of a given figure, 3) exploration of a figure topic using the citation network, and 4) search result re-ranking using an example figure. The different functions were implemented using either classical IR models or neural network-based figure embeddings. Finally, the system was designed to facilitate the collection of user data for training and test purposes and it is flexible enough such that it can be extended to include new functions and algorithms. As an open-source system, FigExplorer can help advance the research, evaluation, and development of applications in this area.

With the development of smart agricultural technology, traditional agriculture has also begun to shift to intelligent control. This article integrates wireless sensor network, wireless video monitoring network with Internet. From the collection of farm environmental parameters to real-time video monitoring of crop growth, data is eventually transmitted to the monitoring center. The environmental parameters of farmland and crop growth can be monitored in real time.

Objective. A deep convolutional neural network (CNN) is a method for deep learning (DL). It has a powerful ability to automatically extract features and is widely used in classification tasks with scalp electroencephalogram (EEG) signals. However, the small number of samples and low signal-to-noise ratio involved in scalp EEG with low spatial resolution constitute a limitation that might restrict potential brain-computer interface (BCI) applications that are based on the CNN model. In the present study, a novel CNN model with source-spatial feature images (SSFIs) as the input is proposed to decode auditory attention tracking states in a cocktail party environment. Approach. We first extract SSFIs using rhythm entropy and weighted minimum norm estimation (WMNE). Next, we develop a CNN model with three convolutional layers. Furthermore, we estimate the performance of the proposed model via generalized performance, alternative models that deleted or replaced a model's component, and loss curves. Finally, we use a deep transfer model with fine-tuning for a low (poor) behavioral performance group (L-group). Main results. Based on cortical activity reconstructions from the scalp EEGs, the classification accuracy (CA) of the proposed model is 80.4% (chance level: 52.5%), which is superior to that achieved by scalp EEG. Additionally, the performance of the proposed model is more stable when compared to alternative models that delete or replace specific model components. The proposed model identifies the difference between two auditory attention tracking states (successful versus unsuccessful) at an early stage with a short time window (250 ms after target offset). Furthermore, we propose a deep transfer learning model to improve the classification for the L-group. With this model, the CA of the L-group significantly increase by 5.3%. Significance. Our proposed model improves the performance of a decoder for auditory attention tracking, which could be suitable for relieving the difficulty with the attentional modulation of individual's neural responses. It provides a novel communication channel with auditory cognitive BCI for patients with attention and hearing impairment.

The advent of natural language understanding (NLU) benchmarks for English, such as GLUE and SuperGLUE allows new NLU models to be evaluated across a diverse set of tasks. These comprehensive benchmarks have facilitated a broad range of research and applications in natural language processing (NLP). The problem, however, is that most such benchmarks are limited to English, which has made it difficult to replicate many of the successes in English NLU for other languages. To help remedy this issue, we introduce the first large-scale Chinese Language Understanding Evaluation (CLUE) benchmark. CLUE is an open-ended, community-driven project that brings together 9 tasks spanning several well-established single-sentence/sentence-pair classification tasks, as well as machine reading comprehension, all on original Chinese text. To establish results on these tasks, we report scores using an exhaustive set of current state-of-the-art pre-trained Chinese models (9 in total). We also introduce a number of supplementary datasets and additional tools to help facilitate further progress on Chinese NLU. Our benchmark is released at https://www.CLUEbenchmarks.com, Comment: Accepted by COLING2020; 10 pages, 4 figures

Problem-based learning (PBL) is a student-centered teaching and learning model, in which teachers play a guiding and inspiring role, students need to cooperate with each other to solve problems. The computer network course covers a wide range of knowledge and needs to integrate theory with practice. It is very suitable for PBL teaching. In view of the general situation of teaching and learning for teachers and students in China, i.e., the passivity of teaching and learning, this paper attempts to apply PBL to the teaching of computer network course. Finally, the effectiveness of PBL is evaluated through examination and questionnaire for students.

This paper addresses electric motor fault diagnosis using supervised machine learning classification. A total of 15 distinct fault types are classified and multilabel strategies are used to classify concurrent faults. we explored, developed, and compared the performance of different types of binary (fault/non-fault), multi-class (fault type) and multi-label (single fault versus combination fault) classifiers. To evaluate the effectiveness of fault identification and classification, we used different supervised machine learning methods, including Random forest classification, support vector machine and neural network classification. Through experiment, we compared these methods over 4 classification regimes and finally summarize the most suitable machine learning algorithms for different aspects of health diagnosis in traction motors area.

Known for their capabilities in automated fluid manipulation, microfluidic devices integrated with pneumatic valves are broadly used for researches in life science and clinical practice. The application is, however, hindered by the high cost and overly complex fabrication procedure. Here, we present an approach for fabricating molds of active fluidic devices using a benchtop 3D printer and a simple 2-step protocol (i.e. 3D printing and polishing). The entire workflow can be completed within 6 h, costing less than US$ 5 to produce all necessary templates for PDMS replica molding, which have smooth surface and round-shaped pneumatic valve structures. Moreover, 3D printing can create unique bespoke on-off objects of a wide range of dimensions. The millimeter- and centimeter-sized features allow examination of large-scale biological samples. Our results demonstrate that the 3D-printed active fluidic device has valve control capacities on par with those made by photolithography. Controlled nutrients and ligands delivery by on-off active valves allows generation of dynamic signals mimicking the ever-changing environmental stimuli, and combinatorial/sequential drug inputs for therapeutic screening on liver tumor spheroid. We believe that the proposed methodology can pave the way for integration of active fluidic systems in research labs, clinical settings and even household appliances for a broad range of application.

Many existing studies for the mapping of function brain networks impose an implicit assumption that the networks' spatial distributions are constant over time. However, the latest research reports reveal that functional brain networks are dynamical and have time-varying spatial patterns. Furthermore, how these functional networks evolve over time has not been elaborated and explained in sufficient details yet. In this paper, we aim to discover and characterize the dynamics of functional brain networks via a windowed group-wise dictionary learning and sparse coding approach. First, we aggregated the sampled subjects' fMRI signals into one big data matrix, and learned a common dictionary for all individuals via a group-wise dictionary learning step. Second, we obtained the dynamic time-varying functional networks by using the windowed time-varying sparse coding approach. Experimental results demonstrated that our windowed group-wise dictionary learning and sparse coding method can effectively detect the task-evoked networks and also characterize how these networks evolve over time. This work sheds novel insights on the dynamics mechanism of functional brain networks.

The failure of the railway station or line will cause the reduction of the railway network efficiency and bring travel inconvenience to the passengers, especially for the key stations and tracks. Reasonable invulnerability assessment is beneficial to maintain the system operation ability and prevent passengers from abandoning railway when the uncertain environmental hazard risk is appeared. At present, larger numbers of research about the railway network invulnerability assessment are purely based on complex network theory, while ignored the nature of railways. In this paper, the Railway Network Invulnerability Model (RNIM) is introduced from the perspective of infrastructure network and service network, and the efficiency of the proposed RNIM is verified according to compare the structural connectively and functional accessibility using different attack strategies. Furthermore, the RNIM framework is applied on the China high-speed railway network, and the results show the validity of the proposed model for the invulnerability of railway network assessment in accurately and efficiently. Finally, some protection suggestions are proposed to reduce the negative effects after the inevitable disruptions emerged.

Aiming at the disadvantage of high availability based on node redundant, a high availability scheme based on storage devices redundant was researched and designed in large-scale cluster system with network storage devices by working on system kernel. The advantages of low-cost, low-complexity of maintenance and high-utilization of resources were received. Besides, an asynchronous prototype system of high availability based on log was implemented. Test proved that anticipated goals in system performance, data consistency and availability were achieved.

In order to meet the needs of large number of overlapping samples in sequential fine feature extraction, this work proposes a frame by frame (segment) iterative algorithm based on the “sparse” feature of incremental data to achieve efficient real-time calculation of large FFT, which breaks through the hardware bottleneck of real-time computing platform such as DSP platform in high-resolution spectral analysis. The proposed algorithm makes use of the sparseness of incremental data to make the computation burden of segment iteration lighter than that of standard FFT, and has the potential of parallel computation. The simulation results show that the new algorithm is more efficient than the standard FFT when the number of segments is large.

Fault diagnosis is vital for normal operation of the rotating motor. An effective and reliable deep learning method known as stacked denoising autoencoder (SDAE) is investigated in this paper, which can extract the features from the pending signals with disturbances. Deep adaptive networks are designed to extract features automatically from time domain data and frequency domain data of motor vibration signal, respectively. Then, the network parameters of the SDAE are trained to reconstruct the signal features, and clustering results are investigated. Finally, a classification layer is added to the top layer of the SDAE network for the fault isolation. It is shown that, the diagnosis accuracy with input of vibratory frequency signal is higher than that of time domain signal. The features extracted by SDAE can represent complex mapping relationships between signal and various running status, and the accuracy is improved comparing with traditional fault diagnosis methods.

With the rapid development of traction motor, the mechanical health monitoring and fault diagnosis field have entered the era of big data. Definite harmonic signals of the line current are located by a popular method known as motor current signature analysis. Different faults of an induction motor such as rotor, stator, bearing, vibration, air gap eccentricity and their different diagnosis techniques are also explored. In fact, the actual fault detection also has a deep development in the artificial intelligence. It is truly evident that the scope of this area is vast. Hence, acknowledging the need for future research, this review paper presents a birds eye view on different types of traction induction faults and their diagnostics schemes.

This paper presents a Ka-band secondary surveillance radar transceiver system module. Temperature compensated technique is adopted to stable the transceiver's performance. A MEMS filter and four multifunction chips based on GaAs technique are designed in this work. The transceiver system consists of these multifunction chips and MEMS filter beside with some other commercial chips. Mixers and filters are reused to reduce the system's size, which makes the system be suitable for the large-scale system integration. Furthermore, the system only has two signal terminals and one control bus to reduce the interconnection complexity. The transceiver is able to work form −55oC to +75oC, benefiting from the designed temperature compensation circuits on the chips. The noise figure of the receiver is less than 3.7 dB with more than 23.9 dB power gain. The maximum output power of the transmitter is larger than 30.3 dBm.

Using the scalp time-varying network method, the present study is the first to investigate the temporal influence of the reference on N170, a negative event-related potential component (ERP) appeared about 170 ms that is elicited by facial recognition, in the network levels. Two kinds of scalp electroencephalogram (EEG) references, namely, AR (average of all recording channels) and reference electrode standardization technique (REST), were comparatively investigated via the time-varying processing of N170. Results showed that the latency and amplitude of N170 were significantly different between REST and AR, with the former being earlier and smaller. In particular, the information flow from right temporal-parietal P8 to left P7 in the time-varying network was earlier in REST than that in AR, and this phenomenon was reproduced by simulation, in which the performance of REST was closer to the true case at source level. These findings indicate that reference plays a crucial role in ERP data interpretation, and importantly, the newly developed approximate zero-reference REST would be a superior choice for precise evaluation of the scalp spatio-temporal changes relating to various cognitive events.

Using single-trial cortical signals calculated by weighted minimum norm solution estimation (WMNE), the present study explored a feature extraction method based on rhythm entropy to classify the scalp electroencephalography (EEG) signals of error response from that of correct response during performing auditory-track tasks in cocktail party environment. The classification rate achieved 89.7% with single-trial (≈700 ms) when using support vector machine(SVM) with the leave-one-out-cross-validation (LOOCV). And high discriminative regions mainly distributed at the medial frontal cortex (MFC), the left supplementary motor area (lSMA) and the right supplementary motor area (rSMA). The mean entropy value for error trials was significantly lower than that for correct trials in the discriminative cortices. By time-varying network analysis, different information flows changed among these discriminative regions with time, i.e. error processing showed a left-bias information flow, and correct processing presented a right-bias information flow. These findings revealed that the rhythm information based on single cortical signals could be well used to describe characteristics of error-related EEG signals and further provided a novel application about auditory attention for brain computer interfaces (BCIs).

The Audiovisual Speech Recognition (AVSR) most commonly applied to multimodal learning employs both the video and audio information to do Robust Automatic Speech Recognition. Traditionally, AVSR was regarded as the inference and projection, a lot of restrictions on the ability of it. With the in-depth study, DNN becomes an important part of the toolkit in traditional classification tools, such as automatic speech recognition, image classification, natural language processing. AVSR often use some DNN models including Multimodal Deep Autoencoders (MDAEs), Multimodal Deep Belief Network (MDBN) and Multimodal Deep Boltzmann Machine (MDBM), which are always better than the traditional methods. However, such DNN models have several shortcomings: Firstly, they can’t balance the modal fusion and temporal fusion, or even haven’t temporal fusion; Secondly, the architecture of these models isn’t end-to-end. In addition, the training and testing are cumbersome. We designed a DNN model—Aggregate\(\varvec{d}\) Mult\(\varvec{i}\)moda\(\varvec{l}\) Bidirection\(\varvec{a}\)l Recurren\(\varvec{t}\) Mod\(\varvec{e}\)l (DILATE)—to overcome such weakness. The DILATE could be not just trained and tested simultaneously, but alternatively easy to train and prevent overfitting automatically. The experiments show that DILATE is superior to traditional methods and other DNN models in some benchmark datasets.

In the present study, we used the time-varying scalp network analysis method of electroencephalography (EEG) to investigate information flows among different tempos perception in the alpha band. The results showed the network hubs of different tempos existed variously. Only during listening to normal tempo (52 bpm), the strongest out-degree of hubs is distributed in the right hemisphere and the information flow transferred from the left to the right hemisphere. Based on these findings, we proposed that the left hemisphere did not prime the processing until the necessary information had been transferred from the right hemisphere. This study was the first to use time-varying network method based on adaptive directed transfer function (ADTF) to investigate music-related EEG activities and proposed a novel method to reveal the neural mechanisms on music tempo.

Spectral entropy, which is generated by using the power distribution of the Fourier-transformed EEG using the Shannon entropy, was utilized to measure the uniformity of power spectral density underlying EEG when subjects performed the working memory tasks twice, i.e. before and after training. According to Signed Residual Time (SRT) scores that based on response speed and accuracy track-off, twenty subjects were divided into two groups, namely high-performance and low-performance groups, to undertake working memory (WM) tasks. We found that spectral entropy derived from the retention period of WM on channel FC4 exhibited a high correlation with SRT scores. To this end, spectral entropy was used in support vector machine (SVM) classifier with linear kennel to differentiate these two groups. Receiver operating characteristics (ROC) analysis and leave-one out cross-validation (LOOCV) demonstrated that the averaged classification accuracy (CA) was 90.0% and 92.5% for intra-session and inter-session, respectively, indicating that spectral entropy could be used to distinguish these two different WM performance groups successfully. Furthermore, the support vector regression (SVR) prediction model with RBF kernel and the root-mean-square error of prediction (RMSEP) revealed that spectral entropy could be utilized to predict SRT scores on individual WM performance. After testing the changes in SRT scores and spectral entropy for each subject before and after training, we found that 16 in 20 subjects’ SRT scores was clearly promoted after training and 15 in 20 subjects’ SRT scores showed consistent changes with spectral entropy before and after training. The findings revealed that the spectral entropy could be a promising indicator to predict individual’s WM changes by training and further provide a novel application about working memory for Brain-Computer Interfaces (BCIs).

To effectively solve the traffic data problems such as data invalidation in the process of the acquisition of road traffic states, a road traffic states estimation algorithm based on matching of the regional traffic attracters was proposed in this work. First of all, the road traffic running states were divided into several different modes. The concept of the regional traffic attracters of the target link was put forward for effective matching. Then, the reference sequences of characteristics of traffic running states with the contents of the target link’s traffic running states and regional traffic attracters under different modes were established. In addition, the current and historical regional traffic attracters of the target link were matched through certain matching rules, and the historical traffic running states of the target link corresponding to the optimal matching were selected as the initial recovery data, which were processed with Kalman filter to obtain the final recovery data. Finally, some typical expressways in Beijing were adopted for the verification of this road traffic states estimation algorithm. The results prove that this traffic states estimation approach based on matching of the regional traffic attracters is feasible and can achieve a high accuracy.

Magnetic sensors can be applied in vehicle recognition. Most of the existing vehicle recognition algorithms use one sensor node to measure a vehicle's signature. However, vehicle speed variation and environmental disturbances usually cause errors during such a process. In this paper we propose a method using multiple sensor nodes to accomplish vehicle recognition. Based on the matching result of one vehicle's signature obtained by different nodes, this method determines vehicle status and corrects signature segmentation. The co-relationship between signatures is also obtained, and the time offset is corrected by such a co-relationship. The corrected signatures are fused via maximum likelihood estimation, so as to obtain more accurate vehicle signatures. Examples show that the proposed algorithm can provide input parameters with higher accuracy. It improves the av- erage accuracy of vehicle recognition from 94.0% to 96.1%, and especially the bus recognition accuracy from 77.6% to 92.8%.

Eye activity is one of the main sources of artifacts in electroencephalogram (EEG) recordings, however, the ocular artifact can seriously distort the EEG recordings. It is an open issue to remove the ocular artifact as completely as possible without losing the useful EEG information. Independent Component Analysis (ICA) has been one of the correction approaches to correct the ocular artifact in practice. However, ICA based approach may overly or less remove the artifacts when the EEG sources and ocular sources cannot be represented in different independent components (ICs). In this paper, a new approach combining ICA and Auto-Regressive eXogenous (ARX) (ICA-ARX) is proposed for a more robust removal of ocular artifact. In the proposed approach, to lower the negative effect induced by ICA, ARX is used to build the multi-models based on the ICA corrected signals and the reference EEG selected before contamination period for each channel, and then the optimal model will be selected for further artifact removal. The results applied to both the simulated signals and actual EEG recordings demonstrate the effectiveness of the proposed approach for ocular artifact removal, and its potential to be used in the EEG related studies.

Motor imagery (MI) requires subjects to visualize the requested motor behaviors, which involves a large-scale network that spans multiple brain areas. The corresponding cortical activity reflected on the scalp is characterized by event-related desynchronization (ERD) and then by event-related synchronization (ERS). However, the network mechanisms that account for the dynamic information processing of MI during the ERD and ERS periods remain unknown. Here, we combined ERD/ERS analysis with the dynamic networks in different MI stages (i.e. motor preparation, ERD and ERS) to probe the dynamic processing of MI information. Our results show that specific dynamic network structures correspond to the ERD/ERS evolution patterns. Specifically, ERD mainly shows the contralateral networks, while ERS has the symmetric networks. Moreover, different dynamic network patterns are also revealed between the two types of MIs, in which the left-hand MIs exhibit a relatively less sustained contralateral network, which may be the network mechanism that accounts for the bilateral ERD/ERS observed for the left-hand MIs. Similar to the network topologies, the three MI stages also appear to be characterized by different network properties. The above findings all demonstrate that different MI stages that involve specific brain networks for dynamically processing the MI information.

The neural mechanism of steady-state visual evoked potentials (SSVEP) is still not clearly understood. Especially, only certain frequency stimuli can evoke SSVEP. Our previous network study reveals that 8 Hz stimulus that can evoke strong SSVEP response shows the enhanced linkage strength between frontal and visual cortex. To further probe the directed information flow between the two cortex areas for various frequency stimuli, this paper develops a causality analysis based on the inversion of double columns model using particle swarm optimization (PSO) to characterize the directed information flow between visual and frontal cortices with the intracranial rat electroencephalograph (EEG). The estimated model parameters demonstrate that the 8 Hz stimulus shows the enhanced directional information flow from visual cortex to frontal lobe facilitates SSVEP response, which may account for the strong SSVEP response for 8 Hz stimulus. Furthermore, the similar finding is replicated by data-driven causality analysis. The inversion of neural mass model proposed in this study may be helpful to provide the new causality analysis to link the physiological model and the observed datasets in neuroscience and clinical researches.

Mapping human brain networks provides a basis for studying brain function and dysfunction, and thus has gained significant interest in recent years. However, modeling human brain networks still faces several challenges including constructing networks at multiple spatial scales and finding common corresponding networks across individuals. As a consequence, many previous methods were designed for a single resolution or scale of brain network, though the brain networks are multi-scale in nature. To address this problem, this paper presents a novel approach to constructing multi-scale common structural brain networks from DTI data via an improved multi-scale spectral clustering applied on our recently developed and validated DICCCOLs (Dense Individualized and Common Connectivity-based Cortical Landmarks). Since the DICCCOL landmarks possess intrinsic structural correspondences across individuals and populations, we employed the multi-scale spectral clustering algorithm to group the DICCCOL landmarks and their connections into sub-networks, meanwhile preserving the intrinsically-established correspondences across multiple scales. Experimental results demonstrated that the proposed method can generate multi-scale consistent and common structural brain networks across subjects, and its reproducibility has been verified by multiple independent datasets. As an application, these multi-scale networks were used to guide the clustering of multi-scale fiber bundles and to compare the fiber integrity in schizophrenia and healthy controls. In general, our methods offer a novel and effective framework for brain network modeling and tract-based analysis of DTI data.

Steady-state visual evoked potentials (SSVEP) are the visual system responses to a repetitive visual stimulus flick- ering with the constant frequency and of great importance in the study of brain activity using scalp electroencephalography (EEG) recordings. However, the reference influence for the investigation of SSVEP is generally not considered in previous work. In this study a new approach that combined the canonical correlation analysis with infinite reference (ICCA) was pro- posed to enhance the accuracy of frequency recognition of SSVEP recordings. Compared with the widely used periodogram method (PM), ICCA is able to achieve higher recognition accuracy when extracts frequency within a short span. Further, the recognition results suggested that ICCA is a very robust tool to study the brain computer interface (BCI) based on SSVEP.

This paper proposes a novel method to solve the problem of train communication network design. Firstly, we put forward a general description of such problem. Then, taking advantage of the bilevel programming theory, we created the cost-reliability-delay model (CRD model) that consisted of two parts: the physical topology part aimed at obtaining the networks with the maximum reliability under constrained cost, while the logical topology part focused on the communication paths yielding minimum delay based on the physical topology delivered from upper level. We also suggested a method to solve the CRD model, which combined the genetic algorithm and the Floyd-Warshall algorithm. Finally, we used a practical example to verify the accuracy and the effectiveness of the CRD model and further applied the novel method on a train with six carriages.

In addition to the traditional TCN, both the train communication networks based on Ethernet and WSNs are also applied, to some extent, in metros. The real time is a decisive indicator of communication network in metro trains. To address such index, this paper develops an evaluative model to comprehensively analyze it and compares the Real-Time of different communication networks. Supported by practical examples, we conclude that this model functions effectively and accurately.

The statistics of residents' travel is a very important basic data in traffic planning. Traditional resident trip survey usually uses the methods household survey and roadside monitoring which is time-consuming acquiring process, expensive and low-level automaticity. The method of calculating OD traffic capacity through the traffic volumes of road sections is studied in the preceding years, but it is difficult in parametric inference and algorithm solution, and its accuracy yet to be improved. With the development of mobile communication technology, the method of positioning based on GSM mobile phone network has attracted much attention in the application of traffic area. This paper introduces the mobile positioning technology, and uses the residents' mobile phone location and timestamps date, combined with the technology of map matching, adopting holistic approach and collecting idea, to analysis the OD trips between traffic cells. We use the mobile phone location data provided by the operator in China to confirm this method.

Steady state visual evoked potentials (SSVEP) are assumed to be regulated by multiple brain areas, yet the underlying mechanisms are not well understood. In this study, we utilized multi-channel intracranial recordings together with network analysis to investigate the underlying relationships between SSVEP and brain networks in anesthetized rat. We examined the relationship between SSVEP amplitude and the network topological properties for different stimulation frequencies, the synergetic dynamic changes of the amplitude and topological properties in each rat, the network properties of the control state, and the individual difference of SSVEP network attributes existing among rats. All these aspects consistently indicate that SSVEP response is closely correlated with network properties, the reorganization of the background network plays a crucial role in SSVEP production, and the background network may provide a physiological marker for evaluating the potential of SSVEP generation.

With the increasing of side-scan sonar applications, the studies on side-scan sonar image processing have been reported more and more increasingly. This paper firstly analyzes these current studies in this field to explain the need for this study, then it brings forward the main contents of side-scan sonar image processing and analyzes these contents one by one. Based on SIS-1624 side-scan sonar system, ArcGIS92 and Oracle92 this paper has built a prototype system and implemented the geometric correction, the gray correction, the geo-coding and the image fusion methods.