Attention-based long short-term memory fully convolutional network for chemical process fault diagnosis

A correct and timely fault diagnosis is important for improving the safety and reliability of chemical processes. With the advancement of big data technology, data-driven fault diagnosis methods are being extensively used and still have considerable potential. In recent years, methods based on deep neural networks have made significant breakthroughs, and fault diagnosis methods for industrial processes based on deep learning have attracted considerable research attention. Therefore, we propose a fusion deeplearning algorithm based on a fully convolutional neural network(FCN) to extract features and build models to correctly diagnose all types of faults. We use long short-term memory(LSTM) units to expand our proposed FCN so that our proposed deep learning model can better extract the time-domain features of chemical process data. We also introduce the attention mechanism into the model, aimed at highlighting the importance of features, which is significant for the fault diagnosis of chemical processes with many features. When applied to the benchmark Tennessee Eastman process, our proposed model exhibits impressive performance, demonstrating the effectiveness of the attention-based LSTM FCN in chemical process fault diagnosis.

Predicting and Curing Depression Using Long Short Term Memory and Global Vector

In today’s world, there are many people suffering from mentalhealth problems such as depression and anxiety. If these conditions are notidentified and treated early, they can get worse quickly and have far-reachingnegative effects. Unfortunately, many people suffering from these conditions,especially depression and hypertension, are unaware of their existence until theconditions become chronic. Thus, this paper proposes a novel approach usingBi-directional Long Short-Term Memory (Bi-LSTM) algorithm and GlobalVector (GloVe) algorithm for the prediction and treatment of these conditions.Smartwatches and fitness bands can be equipped with these algorithms whichcan share data with a variety of IoT devices and smart systems to betterunderstand and analyze the user’s condition. We compared the accuracy andloss of the training dataset and the validation dataset of the two modelsnamely, Bi-LSTM without a global vector layer and with a global vector layer.It was observed that the model of Bi-LSTM without a global vector layer hadan accuracy of 83%,while Bi-LSTMwith a global vector layer had an accuracyof 86% with a precision of 86.4%, and an F1 score of 0.861. In addition toproviding basic therapies for the treatment of identified cases, our model alsohelps prevent the deterioration of associated conditions, making our methoda real-world solution.

AN INFORMATIC APPROACH TO A LONG MEMORY STATIONARY PROCESS

Long memory is an important phenomenon that arises sometimes in the analysis of time series or spatial data.Most of the definitions concerning the long memory of a stationary process are based on the second-order properties of the process.The mutual information between the past and future I_(p−f) of a stationary process represents the information stored in the history of the process which can be used to predict the future.We suggest that a stationary process can be referred to as long memory if its I_(p−f) is infinite.For a stationary process with finite block entropy,I_(p−f) is equal to the excess entropy,which is the summation of redundancies that relate the convergence rate of the conditional(differential)entropy to the entropy rate.Since the definitions of the I_(p−f) and the excess entropy of a stationary process require a very weak moment condition on the distribution of the process,it can be applied to processes whose distributions are without a bounded second moment.A significant property of I_(p−f) is that it is invariant under one-to-one transformation;this enables us to know the I_(p−f) of a stationary process from other processes.For a stationary Gaussian process,the long memory in the sense of mutual information is more strict than that in the sense of covariance.We demonstrate that the I_(p−f) of fractional Gaussian noise is infinite if and only if the Hurst parameter is H∈(1/2,1).

LÉVY AREA ANALYSIS AND PARAMETER ESTIMATION FOR FOU PROCESSES VIA NON-GEOMETRIC ROUGH PATH THEORY

This paper addresses the estimation problem of an unknown drift parameter matrix for a fractional Ornstein-Uhlenbeck process in a multi-dimensional setting.To tackle this problem,we propose a novel approach based on rough path theory that allows us to construct pathwise rough path estimators from both continuous and discrete observations of a single path.Our approach is particularly suitable for high-frequency data.To formulate the parameter estimators,we introduce a theory of pathwise Itôintegrals with respect to fractional Brownian motion.By establishing the regularity of fractional Ornstein-Uhlenbeck processes and analyzing the long-term behavior of the associated Lévy area processes,we demonstrate that our estimators are strongly consistent and pathwise stable.Our findings offer a new perspective on estimating the drift parameter matrix for fractional Ornstein-Uhlenbeck processes in multi-dimensional settings,and may have practical implications for fields including finance,economics,and engineering.

Comparative study on the performance of ConvLSTM and ConvGRU in classification problems-taking early warning of short-duration heavy rainfall as an example

卷积长短期记忆单元ConvLSTM和卷积门控循环单元ConvGRU是两种广泛应用的深度学习单元,通过将循环机制与卷积运算相结合,常常用于时空序列的预测.为了明确上述两种模型的收敛速度和分类能力,需要使用相同的模型架构对相同的分类问题进行预测.本研究将北京短时强降水区级预警问题看作深度学习中的二分类问题,使用京津冀雷达网的组合反射率数据和北京区域内的自动气象站降雨数据进行深度学习模型的训练和评估.结果表明,ConvGRU的收敛速度比ConvLSTM快约25%.ConvLSTM和ConvGRU的预警性能随地区,时间,降雨强度的变化趋势相似,但大部分ConvLSTM的得分较高,少数情况下ConvGRU的得分较高.

Eye on the Sky: A UAP Research and Field Study off New York’s Long Island Coast

A ten-month field research study was meticulously conducted at Robert Moses State Park (RMSP) on the south shore of Long Island, NY. The objective was to determine if aerial phenomena of an unknown nature exist over a coastal location and to characterize their properties and behaviors. Primary and secondary field observation methods were utilized in this data-centric study. Forensic engineering principles and methodologies guided the study. The challenges set forward were object detection, observation, and characterization, where multispectral electro-optical devices and radar were employed due to limited visual acuity and intermittent presentation of the phenomena. The primary means of detection utilized a 3 cm X-band radar operating in two scan geometries, the X- and Y-axis. Multispectral electro-optical devices were utilized as a secondary means of detection and identification. Data was emphasized using HF and LF detectors and spectrum analyzers incorporating EM, ultrasonic, magnetic, and RF field transducers to record spectral data in these domains. Data collection concentrated on characterizing VIS, NIR, SWIR, LWIR, UVA, UVB, UVC, and the higher energy spectral range of ionizing radiation (alpha, beta, gamma, and X-ray) recorded by Geiger-Müller counters as well as special purpose semiconductor diode sensors.

一种基于long short-term memory的唇语识别方法

唇动视觉信息是说话内容的重要载体。受嘴唇外观、背景信息和说话习惯等影响,即使说话者说相同的内容,唇动视觉信息也会相差很大。为解决唇语视觉信息多样性的问题,提出一种基于long short-term memory(LSTM)的新的唇语识别方法。以往大多数的方法从嘴唇外表信息入手。本方法用嘴唇关键点坐标描述嘴唇形变信息作为唇语视频的特征,它具有类内一致性和类间区分性的特点。然后利用LSTM对特征进行时序编码,它能学习具有区分性和泛化性的空间-时序特征。在公开的唇语数据集GRID、MIRACL-VC和Oulu VS上对本方法做了针对分割的单词或短语的说话者独立的唇语识别评估。在GRID和MIRACL-VC上,本方法的准确率比传统方法至少高30%;在Oulu VS上,本方法的准确率接近于最优结果。以上实验结果表明,本文提出的基于LSTM的唇语识别方法有效地解决了唇语视觉信息多样性的问题。

Respective Roles of Short-and Long-Range Interactions in Protein Folding

A new method was presented to discuss the respective roles of short- and long-range interactions in protein folding. It's based on an off-lattice model, which is also being called as toy model. Simulated annealing algorithm was used to search its native conformation. When it is applied to analysis proteins 1agt and 1aho, we find that helical segment cannot fold into native conformation without the influence of long-range interactions. That's to say that long-range interactions are the main determinants in protein folding. Key words toy model - protein folding - simulated annealing algorithm - short and long range interactions CLC number O 242.28 - Q71 Foundation item: Supported by the National Natural Science Foundation of China((60301009)Biography: WANG Long-hui (1976-), female, Ph. D candidate, research direction: machine learning, bioinformatics.

Tool Health Condition Recognition Method for High Speed Milling of Titanium Alloy Based on Principal Component Analysis (PCA) and Long Short Term Memory (LSTM)

The healthy condition of the milling tool has a very high impact on the machining quality of the titanium components.Therefore,it is important to recognize the healthy condition of the tool and replace the damaged cutter at the right time.In order to recognize the health condition of the milling cutter,a method based on the long short term memory(LSTM)was proposed to recognize tool health state in this paper.The various signals collected in the tool wear experiments were analyzed by time-domain statistics,and then the extracted data were generated by principal component analysis(PCA)method.The preprocessed data extracted by PCA is transmitted to the LSTM model for recognition.Compared with back propagation neural network(BPNN)and support vector machine(SVM),the proposed method can effectively utilize the time-domain regulation in the data to achieve higher recognition speed and accuracy.

Conditional Random Field Tracking Model Based on a Visual Long Short Term Memory Network

In dense pedestrian tracking,frequent object occlusions and close distances between objects cause difficulty when accurately estimating object trajectories.In this study,a conditional random field tracking model is established by using a visual long short term memory network in the three-dimensional(3D)space and the motion estimations jointly performed on object trajectory segments.Object visual field information is added to the long short term memory network to improve the accuracy of the motion related object pair selection and motion estimation.To address the uncertainty of the length and interval of trajectory segments,a multimode long short term memory network is proposed for the object motion estimation.The tracking performance is evaluated using the PETS2009 dataset.The experimental results show that the proposed method achieves better performance than the tracking methods based on the independent motion estimation.

Comparison between the preparation,structure and mechanical properties of long fiber reinforced thermoplastics and short fiber reinforced thermoplastic

This article summarizes the comparison between the preparation, structure and mechanical properties of long fiber reinforced thermoplastics （LFT） and short fiber reinforced thermoplastics （SFT）. Both of the experiment and theory results showed that the mechanical properties of long glass fiber reinforced thermoplastics pellets （LGFRT） have been enhanced better than that of short glass fiber reinforced thermoplastics pellets （SGFRT） manufactured by molding procession. After regulation of the relative humidity by 50 % , the mechanical properties of 30 % （ weight percent） short glass fiber content in SFT （ SFT-PA6-SGF30 ） are similar to that of 40 % long glass fiber content in LFT. Howev- er, the density of the latter is about 17 % lower than that of the former. Thus, the corresponding weight of products is reduced by 13 % ;output rate is increased by 21% , and the cost is therefore significantly lowered. And it has the fol- lowing advantages： impact strength is increased by 87 % ; the proportion is reduced by 20 % ; molding cycle is short- ened by 10 % ;materials cost is saved by 20 % -30 % and the final total cost is saved by 30 % -40 %. So LFT （LFT-PP-LGF40） can replace SFT （SFT-PA6-SGF30） with the similar basic mechanical properties under normal tem- perature or 160 ℃ lower.

Growth process and short chain alcohol separation performance of fluoride-containing NaY zeolite membrane

Growth process of the NaY zeolite membranes was investigated by fluoride-containing precursor synthesis gel.Compared with the fluoride-free precursor synthesis gel,the irregular NaY zeolite crystals were dissolved into amorphous by the fluoride-containing precursor synthesis gel initially,the amorphous contained the Y-type zeolite characteristic bands by the IR characterization.The fine square NaY zeolite crystals arose from the amorphous,which were accumulated and gradually grew into a dense NaY zeolite layer on the support surface after 6.5 h.Because the excessive NaY zeolites were dissolved by the strong alkaline and fluoride-containing precursor synthesis gel,there was plenty of amorphous on NaY zeolites layer for prolonging the crystallization time.The assynthesized NaY zeolite membranes had a good separation performance and repeatability for separation of 10 wt%methanol(MeOH)/methyl methacrylate(MMA) mixture by pervaporation,the flux and separation factor were(1.27 ± 0.07) kg·M^(-2)·h^(-1) and(4900 ± 1500) at 323 K,respectively.Besides,the NaY zeolite membranes were applied to separate the other short chain alcohol from the various alcohol/organic ester and alcohol/organic ether mixtures,the NaY zeolite membranes showed high short chain alcohol perm-selectivity.

Binaural Speech Separation Algorithm Based on Long and Short Time Memory Networks

Speaker separation in complex acoustic environment is one of challenging tasks in speech separation.In practice,speakers are very often unmoving or moving slowly in normal communication.In this case,the spatial features among the consecutive speech frames become highly correlated such that it is helpful for speaker separation by providing additional spatial information.To fully exploit this information,we design a separation system on Recurrent Neural Network(RNN)with long short-term memory(LSTM)which effectively learns the temporal dynamics of spatial features.In detail,a LSTM-based speaker separation algorithm is proposed to extract the spatial features in each time-frequency(TF)unit and form the corresponding feature vector.Then,we treat speaker separation as a supervised learning problem,where a modified ideal ratio mask(IRM)is defined as the training function during LSTM learning.Simulations show that the proposed system achieves attractive separation performance in noisy and reverberant environments.Specifically,during the untrained acoustic test with limited priors,e.g.,unmatched signal to noise ratio(SNR)and reverberation,the proposed LSTM based algorithm can still outperforms the existing DNN based method in the measures of PESQ and STOI.It indicates our method is more robust in untrained conditions.

Short-and long-term behaviors of drifts in the Callovo-Oxfordian claystone at the Meuse/Haute-Marne Underground Research Laboratory

Since 2000, the French National Radioactive Waste Management Agency （ANDRA） has been constructing an Underground Research Laboratory （URL） at Bure （east of the Paris Basin） to perform experiments in order to obtain in situ data necessary to demonstrate the feasibility of geological repository in the Callovo- Oxfordian claystone. An important experimental program is planned to characterize the response of the rock to different drift construction methods, Before 2008, at the main level of the laboratory, most of the drifts were excavated using pneumatic hammer and supported with rock bolts, sliding steel arches and fiber shotcrete. Other techniques, such as road header techniques, stiff and flexible supports, have also been used to characterize their impacts. The drift network is developed following the in situ major stresses. The parallel drifts are separated enough so as they can be considered independently when their hydromechanical （HM） behaviors are compared. Mine-by experiments have been performed to measure the HM response of the rock and the mechanical loading applied to the support system due to the digging and after excavation. Drifts exhibit extensional （mode I） and shear fractures （modes II and III） induced by excavation works. The extent of the induced fracture networks depends on the drift orientation versus the in situ stress field. This paper describes the drift convergence and deformation in the surrounding rock walls as function of time and the impact of different support methods on the rock mass behavior. An observation based method is finally applied to distinguish the instantaneous and time-dependent parts of the rock mass deformation around the drifts.

Long Short-Term Memory Recurrent Neural Network-Based Acoustic Model Using Connectionist Temporal Classification on a Large-Scale Training Corpus

A Long Short-Term Memory(LSTM) Recurrent Neural Network(RNN) has driven tremendous improvements on an acoustic model based on Gaussian Mixture Model(GMM). However, these models based on a hybrid method require a forced aligned Hidden Markov Model(HMM) state sequence obtained from the GMM-based acoustic model. Therefore, it requires a long computation time for training both the GMM-based acoustic model and a deep learning-based acoustic model. In order to solve this problem, an acoustic model using CTC algorithm is proposed. CTC algorithm does not require the GMM-based acoustic model because it does not use the forced aligned HMM state sequence. However, previous works on a LSTM RNN-based acoustic model using CTC used a small-scale training corpus. In this paper, the LSTM RNN-based acoustic model using CTC is trained on a large-scale training corpus and its performance is evaluated. The implemented acoustic model has a performance of 6.18% and 15.01% in terms of Word Error Rate(WER) for clean speech and noisy speech, respectively. This is similar to a performance of the acoustic model based on the hybrid method.

Influence of long and short guide vanes on hydraulic performance of water-jet pump

Different guide vane structures will affect the flow inside the pump,and then affect the transformation of the pressure energy and kinetic energy,and change the velocity distribution of the pump outlet.In order to study the influence of long and short guide vanes on the water-jet pump,on the basis of conventional design,eight schemes of guide vane with different vertical heights were designed in the method of computational fluid dynamics for numerical calculation,the performance curve of water-jet pumps with different long and short guide vanes was obtained,and finally the influence of different guide vanes on hydraulic performance and internal flow was analyzed.The results show that all of schemes reducing the height of blade can improve the head and efficiency.In the schemes reducing the height on the shroud,the guide vanes that the height of the blade is equal to the height difference between hub and shroud in impeller have the highest head and efficiency.In all schemes decreasing the blade height,with the increase of the height difference,the velocity increases gradually and the distribution of turbulence kinetic energy becomes more reasonable in the guide vane outlet.The schemes reducing the height on the hub have more reasonable distribution of velocity and turbulence kinetic energy according to schemes reducing the height on the shroud.The guide vanes of long and short blades can be used to stagger the position of the diffusion flow generated by adjacent blades,which can reduce the effect of the velocity circulation and make the flow of the outlet position more stable.

Fluctuations and pseudo long range dependence in network flows: A non-stationary Poisson process model

In the study of complex networks （systems）, the scaling phenomenon of flow fluctuations refers to a certain powerlaw between the mean flux （activity） （Fi） of the i-th node and its variance σi as σi α （Fi）α Such scaling laws are found to be prevalent both in natural and man-made network systems, but the understanding of their origins still remains limited. This paper proposes a non-stationary Poisson process model to give an analytical explanation of the non-universal scaling phenomenon： the exponent α varies between 1/2 and 1 depending on the size of sampling time window and the relative strength of the external/internal driven forces of the systems. The crossover behaviour and the relation of fluctuation scaling with pseudo long range dependence are also accounted for by the model. Numerical experiments show that the proposed model can recover the multi-scaiing phenomenon.

Short-circuit process model in underwater welding using rotating arc sensor

In order to detecting and tracking along the weld seam with rotating arc sensor in underwater welding,the highpressure water environment rotating arc welding hardware platform is established and welding experiments using rotating arc sensor is done. Different radius of rotating arc sensor is used. And the corresponding welding current and voltage is obtained,which is compared with the results of rotating arc sensor short-circuit process simulation model under high-pressure water environment established in this article. The results show that under high-pressure water environment,rotating arc radius should be optimized,otherwise the short-circuit-arcing cycle will transit to a short-circuit-arcing-abruption cycle,making the welding quality poor. At last the critical radius between the short-circuit-arcing cycle and short-circuit-arcing-abruption cycle under high-pressure water environment is obtained.

On the exact solutions to the long short-wave interaction system

The complete discrimination system for polynomial method is applied to the long-short-wave interaction system to obtain the classifications of single traveling wave solutions. Compared with the solutions given by the （G~/G）-expansion method, we gain some new solutions.

Developing a Process Driven Morphological Model for Long Term Evolution of a Dynamic Coastal Embayment

Numerical modelling of coastal morphology is a complex and sometimes unrewarding exercise and often not yielding tangible results. Typically, the underlying drivers of morphology are not properly accounted for in numerical models. Such inaccuracies combined with a paucity of validation data create a difficulty for coastal planners/engineers who are required to interpret such morphological models to develop coastal management strategies. This study develops an approach to long term morphological modelling of a barrier beach system that includes the findings of over 10 years of coastal monitoring on a dynamic coastal system. The novel approach to predicting the long term evolution of the area combines a mix of short term hydrodynamic monitoring and long term morphological modelling to predict future changes in a breached barrier system. A coupled wave, wind, hydrodynamic and sediment transport numerical model was used to predict the coastal evolution in the dynamic barrier beach system of Inner Dingle Bay, Co. Kerry, Ireland. The modelling approach utilizes the schematisation of inputs to reflect observed trends. The approach is subject to two stages of validation both quantitative and qualitative. The study highlights the importance of considering all the parameters responsible for driving coastal evolution and the necessity to have long term monitoring results for trend based validation.