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.

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.

Dynamic Hand Gesture Recognition Based on Short-Term Sampling Neural Networks

Hand gestures are a natural way for human-robot interaction.Vision based dynamic hand gesture recognition has become a hot research topic due to its various applications.This paper presents a novel deep learning network for hand gesture recognition.The network integrates several well-proved modules together to learn both short-term and long-term features from video inputs and meanwhile avoid intensive computation.To learn short-term features,each video input is segmented into a fixed number of frame groups.A frame is randomly selected from each group and represented as an RGB image as well as an optical flow snapshot.These two entities are fused and fed into a convolutional neural network(Conv Net)for feature extraction.The Conv Nets for all groups share parameters.To learn longterm features,outputs from all Conv Nets are fed into a long short-term memory(LSTM)network,by which a final classification result is predicted.The new model has been tested with two popular hand gesture datasets,namely the Jester dataset and Nvidia dataset.Comparing with other models,our model produced very competitive results.The robustness of the new model has also been proved with an augmented dataset with enhanced diversity of hand gestures.

Short-Term Relay Quality Prediction Algorithm Based on Long and Short-Term Memory

The fraction defective of semi-finished products is predicted to optimize the process of relay production lines, by which production quality and productivity are increased, and the costs are decreased. The process parameters of relay production lines are studied based on the long-and-short-term memory network. Then, the Keras deep learning framework is utilized to build up a short-term relay quality prediction algorithm for the semi-finished product. A simulation model is used to study prediction algorithm. The simulation results show that the average prediction absolute error of the fraction is less than 5%. This work displays great application potential in the relay production lines.

Short-TermWind Power Prediction Based on Combinatorial Neural Networks

Wind power volatility not only limits the large-scale grid connection but also poses many challenges to safe grid operation.Accurate wind power prediction can mitigate the adverse effects of wind power volatility on wind power grid connections.For the characteristics of wind power antecedent data and precedent data jointly to determine the prediction accuracy of the prediction model,the short-term prediction of wind power based on a combined neural network is proposed.First,the Bi-directional Long Short Term Memory(BiLSTM)network prediction model is constructed,and the bi-directional nature of the BiLSTM network is used to deeply mine the wind power data information and find the correlation information within the data.Secondly,to avoid the limitation of a single prediction model when the wind power changes abruptly,the Wavelet Transform-Improved Adaptive Genetic Algorithm-Back Propagation(WT-IAGA-BP)neural network based on the combination of the WT-IAGA-BP neural network and BiLSTM network is constructed for the short-term prediction of wind power.Finally,comparing with LSTM,BiLSTM,WT-LSTM,WT-BiLSTM,WT-IAGA-BP,and WT-IAGA-BP&LSTM prediction models,it is verified that the wind power short-term prediction model based on the combination of WT-IAGA-BP neural network and BiLSTM network has higher prediction accuracy.

State of Health Estimation of Lithium-Ion Batteries Using Support Vector Regression and Long Short-Term Memory

Lithium-ion batteries are the most widely accepted type of battery in the electric vehicle industry because of some of their positive inherent characteristics. However, the safety problems associated with inaccurate estimation and prediction of the state of health of these batteries have attracted wide attention due to the adverse negative effect on vehicle safety. In this paper, both machine and deep learning models were used to estimate the state of health of lithium-ion batteries. The paper introduces the definition of battery health status and its importance in the electric vehicle industry. Based on the data preprocessing and visualization analysis, three features related to actual battery capacity degradation are extracted from the data. Two learning models, SVR and LSTM were employed for the state of health estimation and their respective results are compared in this paper. The mean square error and coefficient of determination were the two metrics for the performance evaluation of the models. The experimental results indicate that both models have high estimation results. However, the metrics indicated that the SVR was the overall best model.

基于ARIMA-LSTM的矿区地表沉降预测方法

煤矿开采安全问题尤其是采空区地表沉降现象会对人员安全及工程安全造成威胁,研究合适的矿区地表沉降预测方法具有很大意义。矿区地表沉降影响因素复杂,单一的深度学习模型对矿区地表沉降数据拟合效果差且现有的地表沉降预测研究多是单独进行概率预测或考虑时序特性进行点预测,难以在考虑数据的时序特征的同时对其随机性进行定量描述。针对此问题,在对数据本身性质进行观察分析后选择差分整合移动平均自回归(ARIMA)模型进行时序特征的概率预测,结合长短时记忆(LSTM)网络模型来学习复杂的且具有长期依赖性的非线性时序特征。提出基于ARIMA-LSTM的地表沉降预测模型,利用ARIMA模型对数据的时序线性部分进行预测,并将ARIMA模型预测的残差数据辅助LSTM模型训练,在考虑时序特征的同时对数据的随机性进行描述。研究结果表明,相较于单独采用ARIMA或LSTM模型,该方法具有更高的预测精度(MSE为0.262 87,MAE为0.408 15,RMSE为0.512 71)。进一步的对比结果显示,预测结果与雷达卫星影像数据(经SBAS-INSAR处理后)趋势一致,证实了该方法的有效性。

基于IWOA-LSTM算法的预应力钢筋混凝土梁损伤识别

为准确识别桥梁结构的损伤程度,制作了桥梁的关键构件——预应力钢筋混凝土梁,进行三点弯曲加载试验.收集了损伤破坏全过程的声发射(AE)信号,通过AE信号参数分析,将梁的损伤破坏过程划分为4个典型阶段.构建了长短时记忆神经网络(LSTM)模型,根据经验设置LSTM模型的超参数容易导致网络陷入局部最优而影响了分类结果,提出采用Sine混沌映射和自适应权重来改进鲸鱼优化算法(WOA),对LSTM进行超参数寻优.设计了IWOA-LSTM算法模型,训练识别试验梁各损伤阶段的AE信号特征参数.定型网络结构,并识别同种工况下其他梁的AE信号.结果表明:IWOA-LSTM算法模型识别准确率均超过或接近92%,相较于普通LSTM模型,IWOA-LSTM模型识别准确率提高了约7%.

基于BP-DCKF-LSTM的锂离子电池SOC估计

电池荷电状态(SOC)的准确估计是电池管理系统(BMS)的核心功能之一。为了提高锂电池SOC估算精度,提出了一种将反向传播神经网络(BP)、双容积卡尔曼滤波(DCKF)和长短期记忆神经网络(LSTM)相结合的SOC估计方法。针对多温度条件下传统多项式拟合法在拟合开路电压(OCV)与SOC时效果较差的问题,提出了一种基于BP神经网络的拟合方法,通过验证表明该方法能有效提高拟合精度。针对单独使用模型法或数据驱动法估计SOC各自存在的优缺点,提出了一种将DCKF与LSTM相结合的估计方法,在提高估计精度的同时,可以减少参数调节时间和训练成本。实验验证表明,BP-DCKF-LSTM算法的均方根误差(RMSE)和平均绝对误差(MAE)分别小于0.5%和0.4%,具有较高的SOC估算精度和鲁棒性。

基于改进LSTM的数码雷管模组印刷质量预测

由于数码雷管模组印刷过程中生产工艺复杂、强时序性等特点,其质量的精准预测已成为提高产品质量管理水平的关键。基于此提出一种改进长短期记忆(long short-term memory,LSTM)网络的数码雷管模组印刷质量预测模型。首先根据数码雷管模组印刷过程提炼机器运行参数、环境参数与检测参数作为印刷产品质量的原始特征,并对关键检测参数进行时序特征重构以增强特征表达能力;其次基于改进的LSTM网络建立数码雷管模组印刷特征提取框架,采用卷积神经网络提取空间特征避免LSTM挖掘高维印刷特征时隐含关系的不足,通过全局注意力机制自适应学习不同时刻印刷特征对印刷产品质量的贡献度,为LSTM提取的深层时序特征分配不同权值;最后以深层特征作为输入,通过全连接网络实现数码雷管模组印刷产品的质量预测。实验结果表明,相较于BP神经网络、门控循环单元网络、LSTM等预测方法,改进的LSTM网络有效提高了数码雷管模组印刷产品质量的预测精度。

基于ASFF-AAKR和CNN-BILSTM滚动轴承寿命预测

针对滚动轴承寿命预测精度低,构建健康指标困难的问题。提出了一种基于自适应特征融合(adaptively spatial feature fusion,ASFF)和自联想核回归模型(auto associative kernel regression,AAKR)与卷积神经网络(convolutional neural networks,CNN)和双向长短期记忆网络(bi-directional long-short term memory,BILSTM)的轴承剩余寿命预测模型。首先,在时域、频域和时频域提取多维特征,利用单调性和趋势性筛选敏感特征;其次利用ASFF-AAKR对敏感特征进行特征融合构建健康指标;最后,将健康指标输入到CNN和BILSTM中,实现对滚动轴承的寿命预测。结果表明:所构建的寿命预测模型优于其他模型,该方法具有更低的误差、寿命预测精度更高。

基于Bi-LSTM和改进残差学习的风电功率超短期预测方法

现有的方法在以风电功率时间序列拟合功率曲线时,难以表达风电功率数据所包含的趋势性和周期性等时间信息而出现性能退化问题,从而导致预测精度下降。为了解决性能退化问题从而提高风电功率时间序列预测的精度,提出了基于双向长短时记忆(Bi-LSTM)和改进残差学习的风电功率预测方法。方法由两个部分组成,第一部分是以Bi-LSTM为主的多残差块上,结合稠密残差块网络(DenseNet)与多级残差网络(MRN)的残差连接方式,并且在残差连接上使用一维卷积神经网络(1D CNN)来提取风电功率值中时序的非线性特征部分。第二部分是Bi-LSTM与全连接层(Dense)组成的解码器,将多残差块提取到的功率值时序非线性特征映射为预测结果。方法在实际运行的风电功率数据上进行实验,并与常见的残差网络方法和时间序列预测方法进行对比。方法相比于其他模型方法有着更高的预测精度以及更好的泛化能力。

基于LSTM-DDPG的再入制导方法

针对现有基于深度确定性策略梯度(deep deterministic policy gradient,DDPG)算法的再入制导方法计算精度较差,对强扰动条件适应性不足等问题,在DDPG算法训练框架的基础上,提出一种基于长短期记忆-DDPG(long short term memory-DDPG,LSTM-DDPG)的再入制导方法。该方法采用纵、侧向制导解耦设计思想,在纵向制导方面,首先针对再入制导问题构建强化学习所需的状态、动作空间;其次,确定决策点和制导周期内的指令计算策略,并设计考虑综合性能的奖励函数;然后,引入LSTM网络构建强化学习训练网络,进而通过在线更新策略提升算法的多任务适用性;侧向制导则采用基于横程误差的动态倾侧反转方法,获得倾侧角符号。以美国超音速通用飞行器(common aero vehicle-hypersonic,CAV-H)再入滑翔为例进行仿真,结果表明:与传统数值预测-校正方法相比,所提制导方法具有相当的终端精度和更高的计算效率优势;与现有基于DDPG算法的再入制导方法相比,所提制导方法具有相当的计算效率以及更高的终端精度和鲁棒性。

基于二次分解、LSTM-ELM和误差修正的空气质量指数预测模型

精准预测空气质量指数(Air Quality Index,AQI)对于制定有效的空气污染治理策略至关重要。为了进一步提升AQI的预测精度,提出了一种新的预测模型,并结合了二次分解(Secondary Decomposition,SD)、优化算法、双尺度预测和误差修正的方法。首先,采用改良的自适应白噪声完全集合经验模态分解(Improved Complete Ensemble Empirical Mode Decomposition with Adaptive Noise,ICEEMDAN)和样本熵(Sample Entropy,SE)对原始AQI序列进行分解并重构,获得高频、中频和低频3个频率分量。其次,利用经过北方苍鹰算法(Northern Goshawk Optimization,NGO)优化的变分模态分解(Variational Mode Decomposition,VMD)对高频分量进行二次分解,进一步降低其复杂度。再次,引入向量加权平均算法(Weighed Mean of Vectors Algorithm,INFO)对长短期记忆网络(Long Short-Term Memory,LSTM)和极限学习机(Extreme Learning Machine,ELM)的关键参数进行优化,同时利用INFO-LSTM预测高频分量分解后的子序列,进而利用INFO-ELM分别预测中、低频分量,并将所得预测结果进行线性叠加。最后,利用NGO-VMD和INFO-ELM对误差序列进行分解和预测,并对初次预测结果进行修正,得到最终的AQI预测值。研究选取北京、上海和成都3个典型城市为例进行实证分析,并对比了7个对照试验,发现基于二次分解、LSTM-ELM和误差修正的模型具有最高的预测精度。该模型可为治理空气污染提供理论和技术上的帮助。

融合BiLSTM与CNN的推特黑灰产分类模型

当前推特等国外社交平台,已成为从事网络黑灰产犯罪不可或缺的工具,对推特上黑灰产账号进行发现、检测和分类对于打击网络犯罪、维护社会稳定具有重大意义。现有的推文分类模型双向长短时记忆网络(bi-directional long short-term memory,BiLSTM)可以学习推文的上下文信息,却无法学习局部关键信息,卷积神经网络(convolution neural network,CNN)模型可以学习推文的局部关键信息,却无法学习推文的上下文信息。结合BiLSTM与CNN两种模型的优势,提出了BiLSTM-CNN推文分类模型,该模型将推文进行向量化后,输入BiLSTM模型学习推文的上下文信息,再在BiLSTM模型后引入CNN层,进行局部特征的提取,最后使用全连接层将经过池化的特征连接在一起,并应用softmax函数进行四分类。模型在自主构建的中文推特黑灰产推文数据集上进行实验,并使用TextCNN、TextRNN、TextRCNN三种分类模型作为对比实验,实验结果显示,所提的BiLSTM-CNN推文分类模型在对四类推文进行分类的宏准确率为98.32%,明显高于TextCNN、TextRNN和TextRCNN三种模型的准确率。

LSTM neural network for solar radio spectrum classification

A solar radio spectrometer records solar radio radiation in the radio waveband. Such solar radio radiation spanning multiple frequency channels and over a short time period could provide a solar radio spectrum which is a two dimensional image. The vertical axis of a spectrum represents frequency channel and the horizontal axis signifies time. Intrinsically, time dependence exists between neighboring columns of a spectrum since solar radio radiation varies continuously over time. Thus, a spectrum can be treated as a time series consisting of all columns of a spectrum, while treating it as a general image would lose its time series property. A recurrent neural network(RNN) is designed for time series analysis. It can explore the correlation and interaction between neighboring inputs of a time series by augmenting a loop in a network.This paper makes the first attempt to utilize an RNN, specifically long short-term memory(LSTM), for solar radio spectrum classification. LSTM can mine well the context of a time series to acquire more information beyond a non-time series model. As such, as demonstrated by our experimental results, LSTM can learn a better representation of a spectrum, and thus contribute better classification.

Real-time UAV path planning based on LSTM network

To address the shortcomings of single-step decision making in the existing deep reinforcement learning based unmanned aerial vehicle(UAV)real-time path planning problem,a real-time UAV path planning algorithm based on long shortterm memory(RPP-LSTM)network is proposed,which combines the memory characteristics of recurrent neural network(RNN)and the deep reinforcement learning algorithm.LSTM networks are used in this algorithm as Q-value networks for the deep Q network(DQN)algorithm,which makes the decision of the Q-value network has some memory.Thanks to LSTM network,the Q-value network can use the previous environmental information and action information which effectively avoids the problem of single-step decision considering only the current environment.Besides,the algorithm proposes a hierarchical reward and punishment function for the specific problem of UAV real-time path planning,so that the UAV can more reasonably perform path planning.Simulation verification shows that compared with the traditional feed-forward neural network(FNN)based UAV autonomous path planning algorithm,the RPP-LSTM proposed in this paper can adapt to more complex environments and has significantly improved robustness and accuracy when performing UAV real-time path planning.

基于EEMD与CNN-BiLSTM的噪声环境下滚动轴承故障诊断方法

针对滚动轴承在噪声环境中发生故障时,传统深度神经网络容易出现特征提取不充分,过拟合,泛化能力不足的问题,提出一种集成经验模态分解(EEMD)与卷积神经网络-双向长短时记忆网络(CNN-BiLSTM)的故障诊断方法。在信号预处理阶段使用EEMD将噪声环境下的振动信号分解为一系列固有模态函数,降低噪声的影响;在CNN部分的第1层使用大卷积核与多分支结构获得不同的感受野,在每一个分支中随机丢弃一些数据增强模型的抗干扰能力,从而提取到更具泛化能力的多样化特征信息,后续部分使用残差结构,以免网络较深时发生梯度消失的现象,解决深层次网络退化问题;在BiLSTM部分使用2个并行的分支结构,用于增强模型对时序信息的利用,从而提高模型在不同工况和噪声环境下的准确率。使用凯斯西储大学轴承数据集和西安交通大学轴承数据集对所提方法进行验证,并与其他深度学习方法和传统机器学习方法进行对比,结果表明本文方法在多种工况和噪声环境下均取得了优异的故障诊断性能。

Optimizing the Clinical Decision Support System (CDSS) by Using Recurrent Neural Network (RNN) Language Models for Real-Time Medical Query Processing

This research aims to enhance Clinical Decision Support Systems(CDSS)within Wireless Body Area Networks(WBANs)by leveraging advanced machine learning techniques.Specifically,we target the challenges of accurate diagnosis in medical imaging and sequential data analysis using Recurrent Neural Networks(RNNs)with Long Short-Term Memory(LSTM)layers and echo state cells.These models are tailored to improve diagnostic precision,particularly for conditions like rotator cuff tears in osteoporosis patients and gastrointestinal diseases.Traditional diagnostic methods and existing CDSS frameworks often fall short in managing complex,sequential medical data,struggling with long-term dependencies and data imbalances,resulting in suboptimal accuracy and delayed decisions.Our goal is to develop Artificial Intelligence(AI)models that address these shortcomings,offering robust,real-time diagnostic support.We propose a hybrid RNN model that integrates SimpleRNN,LSTM layers,and echo state cells to manage long-term dependencies effectively.Additionally,we introduce CG-Net,a novel Convolutional Neural Network(CNN)framework for gastrointestinal disease classification,which outperforms traditional CNN models.We further enhance model performance through data augmentation and transfer learning,improving generalization and robustness against data scarcity and imbalance.Comprehensive validation,including 5-fold cross-validation and metrics such as accuracy,precision,recall,F1-score,and Area Under the Curve(AUC),confirms the models’reliability.Moreover,SHapley Additive exPlanations(SHAP)and Local Interpretable Model-agnostic Explanations(LIME)are employed to improve model interpretability.Our findings show that the proposed models significantly enhance diagnostic accuracy and efficiency,offering substantial advancements in WBANs and CDSS.