Status Set Sequential Pattern Mining Considering Time Windows and Periodic Analysis of Patterns

The traditional sequential pattern mining method is carried out considering the whole time period and often ignores the sequential patterns that only occur in local time windows, as well as possible periodicity. Therefore, in order to overcome the limitations of traditional methods, this paper proposes status set sequential pattern mining with time windows (SSPMTW). In contrast to traditional methods, the item status is considered, and time windows, minimum confidence, minimum coverage, minimum factor set ratios and other constraints are added to mine more valuable rules in local time windows. The periodicity of these rules is also analyzed. According to the proposed method, this paper improves the Apriori algorithm, proposes the TW-Apriori algorithm, and explains the basic idea of the algorithm. Then, the feasibility, validity and efficiency of the proposed method and algorithm are verified by small-scale and large-scale examples. In a large-scale numerical example solution, the influence of various constraints on the mining results is analyzed. Finally, the solution results of SSPM and SSPMTW are compared and analyzed, and it is suggested that SSPMTW can excavate the laws existing in local time windows and analyze the periodicity of the laws, which solves the problem of SSPM ignoring the laws existing in local time windows and overcomes the limitations of traditional sequential pattern mining algorithms. In addition, the rules mined by SSPMTW reduce the entropy of the system.     

GIS Partial Discharge Pattern Recognition Based on a Novel Convolutional Neural Networks and Long Short-Term Memory

Distinguishing the types of partial discharge (PD) caused by different insulation defects in gas-insulated switchgear (GIS) is a great challenge in the power industry, and improving the recognition accuracy of the relevant models is one of the key problems. In this paper, a convolutional neural network and long short-term memory (CNN-LSTM) model is proposed, which can effectively extract and utilize the spatiotemporal characteristics of PD input signals. First, the spatial characteristics of higher-level PD signals can be obtained through the CNN network, but because CNN is a deep feedforward neural network, it does not have the ability to process time-series data. The PD voltage signal is related to the time dimension, so LSTM saves and analyzes the previous voltage signal information, realizes the modeling of the time dependence of the data, and improves the accuracy of the PD signal pattern recognition. Finally, the pattern recognition results based on CNN-LSTM are given and compared with those based on other traditional analysis methods. The results show that the pattern recognition rate of this method is the highest, with an average of 97.9%, and its overall accuracy is better than that of other traditional analysis methods. The CNN-LSTM model provides a reliable reference for GIS PD diagnosis.     

Deep Learning for Walking Behaviour Detection in Elderly People Using Smart Footwear

The increase in the proportion of elderly in Europe brings with it certain challenges that society needs to address, such as custodial care. We propose a scalable, easily modulated and live assistive technology system, based on a comfortable smart footwear capable of detecting walking behaviour, in order to prevent possible health problems in the elderly, facilitating their urban life as independently and safety as possible. This brings with it the challenge of handling the large amounts of data generated, transmitting and pre-processing that information and analysing it with the aim of obtaining useful information in real/near-real time. This is the basis of information theory. This work presents a complete system aiming at elderly people that can detect different user behaviours/events (sitting, standing without imbalance, standing with imbalance, walking, running, tripping) through information acquired from 20 types of sensor measurements (16 piezoelectric pressure sensors, one accelerometer returning reading for the 3 axis and one temperature sensor) and warn the relatives about possible risks in near-real time. For the detection of these events, a hierarchical structure of cascading binary models is designed and applied using artificial neural network (ANN) algorithms and deep learning techniques. The best models are achieved with convolutional layered ANN and multilayer perceptrons. The overall event detection performance achieves an average accuracy and area under the ROC curve of 0.84 and 0.96, respectively.     

Increased Entropic Brain Dynamics during DeepDream-Induced Altered Perceptual Phenomenology

In recent years, the use of psychedelic drugs to study brain dynamics has flourished due to the unique opportunity they offer to investigate the neural mechanisms of conscious perception. Unfortunately, there are many difficulties to conduct experiments on pharmacologically-induced hallucinations, especially regarding ethical and legal issues. In addition, it is difficult to isolate the neural effects of psychedelic states from other physiological effects elicited by the drug ingestion. Here, we used the DeepDream algorithm to create visual stimuli that mimic the perception of hallucinatory states. Participants were first exposed to a regular video, followed by its modified version, while recording electroencephalography (EEG). Results showed that the frontal region’s activity was characterized by a higher entropy and lower complexity during the modified video, with respect to the regular one, at different time scales. Moreover, we found an increased undirected connectivity and a greater level of entropy in functional connectivity networks elicited by the modified video. These findings suggest that DeepDream and psychedelic drugs induced similar altered brain patterns and demonstrate the potential of adopting this method to study altered perceptual phenomenology in neuroimaging research.     

Application of Euler Neural Networks with Soft Computing Paradigm to Solve Nonlinear Problems Arising in Heat Transfer

In this study, a novel application of neurocomputing technique is presented for solving nonlinear heat transfer and natural convection porous fin problems arising in almost all areas of engineering and technology, especially in mechanical engineering. The mathematical models of the problems are exploited by the intelligent strength of Euler polynomials based Euler neural networks (ENN’s), optimized with a generalized normal distribution optimization (GNDO) algorithm and Interior point algorithm (IPA). In this scheme, ENN’s based differential equation models are constructed in an unsupervised manner, in which the neurons are trained by GNDO as an effective global search technique and IPA, which enhances the local search convergence. Moreover, a temperature distribution of heat transfer and natural convection porous fin are investigated by using an ENN-GNDO-IPA algorithm under the influence of variations in specific heat, thermal conductivity, internal heat generation, and heat transfer rate, respectively. A large number of executions are performed on the proposed technique for different cases to determine the reliability and effectiveness through various performance indicators including Nash–Sutcliffe efficiency (NSE), error in Nash–Sutcliffe efficiency (ENSE), mean absolute error (MAE), and Thiel’s inequality coefficient (TIC). Extensive graphical and statistical analysis shows the dominance of the proposed algorithm with state-of-the-art algorithms and numerical solver RK-4.     

Enhancements of the Gaussian network model in describing nucleotide residue fluctuations for RNA

Gaussian network model (GNM) is an efficient method to investigate the structural dynamics of biomolecules. However, the application of GNM on RNAs is not as good as that on proteins, and there is still room to improve the model. In this study, two novel approaches, named the weighted GNM (wGNM) and the force-constant-decayed GNM (fcdGNM), were proposed to enhance the performance of ENM in investigating the structural dynamics of RNAs. In wGNM, the force constant for each spring is weighted by the number of interacting heavy atom pairs between two nucleotides. In fcdGNM, all the pairwise nucleotides were connected by springs and the force constant decayed exponentially with the separate distance of the nucleotide pairs. The performance of these two proposed models was evaluated by using a non-redundant RNA structure database composed of 51 RNA molecules. The calculation results show that both the proposed models outperform the conventional GNM in reproducing the experimental B-factors of RNA structures. Compared with the conventional GNM, the Pearson correlation coefficient between the predicted and experimental B-factors was improved by 9.85% and 6.76% for wGNM and fcdGNM, respectively. Our studies provide two candidate methods for better revealing the dynamical properties encoded in RNA structures.     

Adaptive synchronization of a class of fractional-order complex-valued chaotic neural network with time-delay

This paper is concerned with the adaptive synchronization of fractional-order complex-valued chaotic neural networks (FOCVCNNs) with time-delay. The chaotic behaviors of a class of fractional-order complex-valued neural network are investigated. Meanwhile, based on the complex-valued inequalities of fractional-order derivatives and the stability theory of fractional-order complex-valued systems, a new adaptive controller and new complex-valued update laws are proposed to construct a synchronization control model for fractional-order complex-valued chaotic neural networks. Finally, the numerical simulation results are presented to illustrate the effectiveness of the developed synchronization scheme.     

基于PCA-BP神经网络的光谱消光法颗粒粒径反演算法研究

光谱消光法广泛应用于颗粒粒径测量领域,在利用光谱消光法对颗粒粒径进行反演的过程中,由于颗粒的消光系数存在理论复杂、计算繁琐、收敛速度慢以及求解不稳定等问题,很大程度上影响了整个反演过程的快速性和准确性。且在众多波长的消光数据中,存在较多重复冗余的信息,也很大程度上增加了反演算法的时间。针对光谱消光法粒径反演算法计算繁琐、反演效率低的问题,提出了基于主成分分析（PCA）和BP神经网络的光谱消光颗粒粒径分析方法。基于Mie散射理论对不同粒径、不同波长下的光谱消光值进行了仿真计算,通过对光谱消光数据集的主成分分析及各个波长综合载荷系数的计算,实现了最优特征波长的选取,利用降维后的光谱消光数据训练了PCA-BP神经网络模型,并利用该网络模型计算了粒径颗粒分布。通过仿真计算,比较了PCA-BP神经网络模型与传统的BP神经网络模型的预测精度,并分析了波长数目对两种神经网络模型预测结果的影响。针对训练得到的PCA-BP神经网络模型开展光谱消光法粒径参数反演算法的验证实验,搭建了光谱消光法颗粒粒径参数测量实验系统,测量了粒径范围在0.5～9.7 μm内的6种不同粒径参数的聚苯乙烯标准颗粒。仿真和实验结果表明：基于主成分分析方法可确定各个波长向量之间的相关性,利用综合载荷系数选取最优特征波长对应的消光值对整体的光谱数据具有较好的代表性,可实现光谱数据的降维。相比传统的BP神经网络模型,基于PCA-BP神经网络模型的颗粒粒径分布的分析方法预测精度更高,对于较分散颗粒系的分布参数的预测有更加明显的优势。而且,被选取的波长数较少时,PCA-BP神经网络模型依然有较高的预测精度。利用训练好的PCA-BP神经网络模型对颗粒粒径参数进行实验验证,预测结果可瞬时输出,颗粒粒径分布误差在5%以内,验证了该算法的可行性。     

基于逆卷积神经网络的等离子体辐射分布断层重建方法

发展了一种基于逆卷积神经网络的图像级重建方法用于聚变等离子体辐射分布的断层反演.通过引入结构相似度(SSIM)作为损失函数,该方法在模拟数据实验中表现出了较好的重建效果.模拟实验结果表明,在弦积分信号噪声强度为10％、15％及20％时,该方法的重建结果依然具有良好的精确度和鲁棒性.     

基于精确时间协议的HL-2M装置定时同步网络研究

基于HL-2M装置对时序精度的要求,参考ITER的设计方案,设计了基于精确时间协议(PTP)的HL-2M分布式时间通讯网络,使HL-2M装置的时钟同步和事件触发的精度从微秒级提高到亚微秒级,最终优于100ns.