State-Space Identification of a Nonlinear Photovoltaic System Using a Bidirectional Long Short-Term Memory (BiLSTM) Recurrent Neural Network

The advancement of wireless sensor networks (WSNs) improves various smart home automation services and home users’ living standards. However, efficiently collecting data and automating smart home services require the extensive deployment of the sensors. Thus, one of the crucial and challenging tasks is to minimize the sensors’ energy consumption for monitoring and automating various activities in a smart home. In this article, we present a solution to control the excessive energy consumption of sensors used to detect various activities of daily living (ADL) of a smart home resident. The sensors within a smart home network are divided into various groups employing the recurrent neural network (RNN) and dynamic time warping (DTW) techniques to predict the activities with high accuracy and less energy consumption. The smart home users’ future activities are forecast with bidirectional long short-term memory (BLSTM) RNN model to select those sensors that are likely to predict the upcoming activities. Similarly, to predict the home users’ unusual activities, a guard sensor is elected among sensors with high similarities with each other using DTW. The sensor’s role is evenly switched between different modes to maintain a fair tradeoff between energy and accuracy. An extensive set of simulations is performed to validate the proposed scheme’s work integrating datasets from authentic sources. Finally, the proposed system significantly reduces the sensors’ energy consumption and prolongs the battery lifetime to approximately 137 days. <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">Note to Practitioners</i> —This article presented an energy-efficient duty-cycling scheme for automating sensors’ operations in a smart home scenario. The traditional duty-cycling schemes mainly provide solutions based on appointing sentries or predicting smart home users’ next activity using models, such as Bayesian networks. We design a system that integrates the advantages of both sentry and prediction-based schemes to reduce the amount of energy required by sensors to detect and automate smart home users’ activities with high accuracy and precision. The active sensors are appointed using a bidirectional long short-term memory recurrent neural network. Similarly, the guard sensors are assigned to detect unusual activities using the similarities among idle sensors. This study could be used to automate the smart home sensors for detecting home user’s activities with less energy, which ultimately prolongs the battery lifetime of the sensors.

This paper proposes a multi-layered anomaly detection scheme to train feature extraction and to test anomaly prediction by using Convolutional Neural Networks (CNNs) layer, Bidirectional and Unidirectional Long Short-Term Memory (LSTM) Recurrent Neural Networks (RNNs), which is one of a novel deep architecture named stacked convolutional bidirectional LSTM network (SCB-LSTM). In the proposed model, the stacked CNNs perform feature extraction of vibration sensor signal patterns, and the result is used to feature learning with the stacked bidirectional LSTMs (SB-LSTMs). After this procedure, the stacked unidirectional LSTMs (SU-LSTMs) enhance the feature learning, and a regression layer finally predicts anomaly detections. The experimental results of bearing data not only show the accuracy of the proposed model in anomaly detection for rotating machinery diagnostics, but also suggest the better performance than other state-of-the-art algorithms such as a plain uni-LSTM or Bi-LSTM.

Bidirectional long short-term memory (BLSTM) recurrent neural networks are powerful acoustic models in terms of recognition accuracy. When BLSTM acoustic models are used in decoding, the speech decoder needs to wait until the end of a whole sentence is reached, such that forward-propagation in the backward direction can then be performed. The nature of BLSTM acoustic models makes them inappropriate for real-time online speech recognition because of the latency issue. Recently, the context-sensitive-chunk BLSTM and latency-controlled BLSTM acoustic models have been proposed, both chop a whole sentence into several overlapping chunks. By appending several left and/or right contextual frames, forward-propagation of BLSTM can be down within a controlled time delay, while the recognition accuracy is maintained when comparing with conventional BLSTM models. In this paper, two improved versions of latency-controlled BLSTM acoustic models are presented. By using different types of neural network topology to initialize the BLSTM memory cell states, we aim at reducing the computational cost introduced by the contextual frames and enabling faster online recognition. Experimental results on a 320-hour Switchboard task have shown that the improved versions accelerate from 24% to 61% in decoding without significant loss in recognition accuracy.

Adaptation of deep neural network (DNN) based language identification models is still a challenging area of research. Recently, state-of-the-art approaches to short duration language identification task have made use of bidirectional long short-term memory (BLSTM) recurrent neural network (RNN) language identification models. Although this enables the effective modelling of sequential information, significant mismatch due to different conditions such as speaker, channel, duration and background noise between training and testing data still exists. An adaptation of BLSTM systems can help to reduce such mismatches between training and testing data. In this paper, a transformation to the existing BLSTM layer is proposed, using learning of a second order factorization matrix called a compensation layer. The condition-dependent parameters of the factorization matrix are estimated to adapt the BLSTM layer weights. Experiments on the AP17-OLR database show that utterance level adaptation helps to achieve relative improvements of 28% in terms of Cavg over a traditional BLSTM for utterances of ‘1s’ duration.

Non-verbal speech cues play an important role in human communication such as expressing emotional states or maintaining the conversational flow. In this paper we investigate the effect of applying deep bidirectional Long Short-Term Memory (BLSTM) recurrent neural networks to the Interspeech 2013 Computational Paralinguistics Social Signals Sub-Challenge dataset requiring frame-wise, speaker-independent detection and classification of laughter and filler vocalizations in speech. BLSTM networks tend to prevail over conventional neural network architectures whenever the recognition or regression task relies on an intelligent exploitation of temporal context information. We introduce deep BLSTM models by stacking several BLSTMs and by combining non-recurrent deep neural networks with BLSTMs. We demonstrate that this new approach achieves significant improvements over previous attempts and we increase the current state-of-the-art unweighted average area-under-the-curve (UAAUC) value of 92.4% to 94.0%. This is the best result on this task reported in the literature so far.

Water inrush and mud burst disasters pose severe challenges to the safe and efficient construction of underground engineering. Water inflow prediction is closely related to drainage design, disaster prevention and control, and the safety of the surrounding ecological environment. Thus, assessing the water inflow accurately is of importance. This study proposes a Bayesian Optimization-Long Short-Term Memory (BOA-LSTM) recurrent neural network for predicting tunnel water inflow. The model is based on four input parameters, namely tunnel depth (H), groundwater level (h), rock quality designation (RQD), and water-richness (W), with water inflow (WI) as the single-output variable. The model first processes and analyzes the data, quantitatively characterizing the correlations between input parameters. The tunnel water inflow is predicted using the long short-term memory (LSTM) recurrent neural network, and the Bayesian optimization algorithm (BOA) is employed to select the hyperparameters of the LSTM, primarily including the number of hidden layer units, initial learning rate, and L2 regularization coefficient. The modeling process incorporates a five-fold cross-validation strategy for dataset partitioning, which effectively mitigates overfitting risks and enhances the model’s generalization capability. After a comprehensive comparison among a series of machine learning models, including a long short-term memory recurrent neural network (LSTM), random forest (RF), back propagation neural network (BP), extreme learning machine (ELM), radial basis function neural network (RBFNN), least squares support vector machine (LIBSVM), and convolutional neural network (CNN), BOA-LSTM performed excellently. The proposed BOA-LSTM model substantially surpasses the standard LSTM and other comparative models in tunnel water inflow prediction, demonstrating superior performance in both accuracy and generalization. Hence, it provides a reference basis for tunnel engineering water inflow prediction.

As the stock market is an important part of the national economy, more and more investors have begun to pay attention to the methods to improve the return on investment and effectively avoid certain risks. Many factors affect the trend of the stock market, and the relevant information has the nature of time series. This paper proposes a composite model CNN‐BiSLSTM to predict the closing price of the stock. Bidirectional special long short‐term memory (BiSLSTM) improved on bidirectional long short‐term memory (BiLSTM) adds 1 − tanh(x) function in the output gate which makes the model better predict the stock price. The model extracts advanced features that influence stock price through convolutional neural network (CNN), and predicts the stock closing price through BiSLSTM after the data processed by CNN. To verify the effectiveness of the model, the historical data of the Shenzhen Component Index from July 1, 1991, to October 30, 2020, are used to train and test the CNN‐BiSLSTM. CNN‐BiSLSTM is compared with multilayer perceptron (MLP), recurrent neural network (RNN), long short‐term memory (LSTM), BiLSTM, CNN‐LSTM, and CNN‐BiLSTM. The experimental results show that the mean absolute error (MAE), root‐mean‐squared error (RMSE), and R‐square (R2) evaluation indicators of the CNN‐BiSLSTM are all optimal. Therefore, CNN‐BiSLSTM can accurately predict the closing price of the Shenzhen Component Index of the next trading day, which can be used as a reference for the majority of investors to effectively avoid certain risks.

In recent times, there is an increasing interest in employing technology to process natural language with the aim of providing information that can benefit society. Language identification refers to the process of detecting which speech a speaker appears to be using. This paper presents an audio-based Ethio-semitic language identification system using Recurrent Neural Network. Identifying the features that can accurately differentiate between various languages is a difficult task because of the very high similarity between characters of each language. Recurrent Neural Network (RNN) was used in this paper in relation to the Mel-frequency cepstral coefficients (MFCCs) features to bring out the key features which helps provide good results. The primary goal of this research is to find the best model for the identification of Ethio-semitic languages such as Amharic, Geez, Guragigna, and Tigrigna. The models were tested using an 8-h collection of audio recording. Experiments were carried out using our unique dataset with an extended version of RNN, Long Short Term Memory (LSTM) and Bidirectional Long Short Term Memory (BLSTM), for 5 and 10 s, respectively. According to the results, Bidirectional Long Short Term Memory (BLSTM) with a 5 s delay outperformed Long Short Term Memory (LSTM). The BLSTM model achieved average results of 98.1, 92.9, and 89.9% for training, validation, and testing accuracy, respectively. As a result, we can infer that the best performing method for the selected Ethio-Semitic language dataset was the BLSTM algorithm with MFCCs feature running for 5 s.

Recent studies indicate that bidirectional Long Short-Term Memory (BLSTM) recurrent neural networks are well-suited for automatic emotion recognition systems and may lead to better results than systems applying other widely used classifiers such as Support Vector Machines or feedforward Neural Networks. The good performance of BLSTM emotion recognition systems could be attributed to their ability to model and exploit contextual information self-learned via recurrently connected memory blocks which allows them to incorporate information about how emotion evolves over time. However, the actual amount of bidirectional context that a BLSTM classifier takes into account when classifying an observation has not been investigated so far. This paper presents a methodology to systematically investigate the number of past and future utterance-level observations that are considered to generate an emotion prediction for a given utterance, and to examine to what extent this temporal bidirectional context contributes to the overall BLSTM performance.

It has been shown in [1, 2] that improved performance can be achieved by formulating the keyword spotting as a non-uniform error automatic speech recognition problem. In this work, we discriminatively train a deep bidirectional long short-term memory (BLSTM) – hidden Markov model (HMM) based acoustic model with non-uniform boosted minimum classification error (BMCE) criterion which imposes more significant error cost on the keywords than those on the non-keywords. By introducing the BLSTM, the context information in both the past and the future are stored and updated to predict the desired output and the long-term dependencies within the speech signal are well captured. With non-uniform BMCE objective, the BLSTM is trained so that the recognition errors related to the keywords are remarkably reduced. The BLSTM is optimized using backpropagation through time and stochastic gradient descent. The keyword spotting system is implemented within weighted finite state transducer framework. The proposed method achieves 5.49% and 7.37% absolute figure-of-merit improvements respectively over the BLSTM and the feedforward deep neural network baseline systems trained with cross-entropy criterion for the keyword spotting task on Switchboard-1 Release 2 dataset.

Reservoir classification is an important component of reservoir geological modelling and reservoir evaluation and identification. Using a single conventional logging curve to identify complex heterogeneous reservoir types has always been a difficult task in logging interpretation. For the first time, this study reveals the advantages of recurrent neural networks in the identification of heterogeneous reservoirs and proposes an optimal parameter bidirectional long short-term memory (Bi-LSTM) recurrent neural network reservoir classification model with optimal parameters that can make full use of logging sequence information. The data used in this work originate from 3 wells in the BZ gas field in China. First, the rationality of the data set and the generation of sequence data were studied in detail, and the logging curve response sequence data, which can fully characterize the reservoir characteristics, were obtained. Then, through multiple simulation experiments, the optimal network structure and hyperparameters were determined, and a Bi-LSTM network model with 5 hidden layers and the optimal network parameters was established. The model was used to predict fractured, pore-fracture and fracture-pore reservoirs in the buried hill metamorphic rock buried beneath the BZ gas field. A comparison with the prediction results of 5 classic machine learning methods and baseline models shows that the Bi-LSTM model with the optimal parameters is superior to the other machine learning methods, especially regarding the prediction accuracy of pore-fracture reservoirs, and the overall accuracy is 92.69%. The method proposed in this paper can accurately identify the strata developed in different types of storage space and significantly improves the reservoir identification accuracy.

Wind speed prediction in the South China Sea is crucial for enhancing maritime safety, supporting operational planning, and optimizing economic activities in sectors such as offshore energy, shipping, and disaster preparedness. In recent years, the statistical auto-regressive integrated moving average (ARIMA) model and advanced deep learning models such as recurrent neural networks (RNN), long short-term memory (LSTM) networks, and Bidirectional LSTM (BiLSTM) have shown strong potential for time series forecasting due to their capacity to model temporal dependencies. However, these models often face limitations in simultaneously capturing rapid short-term fluctuations and long-term temporal patterns in meteorological data. To address this challenge, we propose a novel hybrid architecture, h-RNN-BiLSTM, which integrates the short-term dynamic modeling capability of RNN with the long-range bidirectional dependency modeling of BiLSTM. This fusion enables multi-scale temporal pattern learning, thereby improving forecasting accuracy. The model is evaluated using two widely recognized spatiotemporal datasets: the European Centre for Medium-Range Weather Forecasts (ECMWF) and the Global Forecast System (GFS). Data preprocessing, including missing value imputation and standardization, was applied to ensure data consistency and improve model convergence. Experiments were conducted in two settings: (i) short-term datasets from GFS and ECMWF, and (ii) long-term ECMWF datasets. The performance of h-RNN-BiLSTM was compared against baseline RNN, LSTM, BiLSTM, and the ARIMA model using root mean square error (RMSE) and mean absolute percentage error (MAPE) as evaluation metrics. Results demonstrate that the proposed model consistently outperforms the deep learning baselines and ARIMA, with the most significant gains observed for the long-term ECMWF dataset. Specifically, the model reduced error by 99.7% compared with ARIMA, 70.3% compared with RNN, 30.7% compared with LSTM, and 37.6% compared with BiLSTM. For MAPE, the improvements were 84.3% over ARIMA, 38.8% over RNN, 40.3% over LSTM, and 32.1% over BiLSTM. To the best of our knowledge, this is the first study to integrate RNN and BiLSTM for multi-scale wind speed prediction in the South China Sea, demonstrating improved predictive accuracy over both deep learning and statistical baselines. These findings highlight the model’s operational potential for energy planning, navigation safety, and weather risk management.

PurposeThis paper aims to develop predictive methods, based on recurrent neural networks, useful to support facility managers in building maintenance tasks, by collecting information coming from a computerized maintenance management system (CMMS).Design/methodology/approachThis study applies data-driven and text-mining approaches to a CMMS data set comprising more than 14,500 end-users’ requests for corrective maintenance actions, collected over 14 months. Unidirectional long short-term memory (LSTM) and bidirectional LSTM (Bi-LSTM) recurrent neural networks are trained to predict the priority of each maintenance request and the related technical staff assignment. The data set is also used to depict an overview of corrective maintenance needs and related performances and to verify the most relevant elements in the building and how the current facility management (FM) relates to the requests.FindingsThe study shows that LSTM and Bi-LSTM recurrent neural networks can properly recognize the words contained in the requests, thus correctly and automatically assigning the priority and predicting the technical staff to assign for each end-user’s maintenance request. The obtained global accuracy is very high, reaching 93.3% for priority identification and 96.7% for technical staff assignment. Results also show the main critical building elements for maintenance requests and the related intervention timings.Research limitations/implicationsThis work shows that LSTM and Bi-LSTM recurrent neural networks can automate the assignment process of end-users’ maintenance requests if trained with historical CMMS data. Results are promising; however, the trained LSTM and Bi-LSTM RNN can be applied only to different hospitals adopting similar categorization.Practical implicationsThe data-driven and text-mining approaches can be integrated into the CMMS to support corrective maintenance management by facilities management contractors, i.e. to properly and timely identify the actions to be carried out and the technical staff to assign.Social implicationsThe improvement of the maintenance of the health-care system is a key component of improving health service delivery. This work shows how to reduce health-care service interruptions due to maintenance needs through machine learning methods.Originality/valueThis study develops original methods and tools easily integrable into IT workflow systems (i.e. CMMS) in the FM field.

Recurrent neural network language models have enjoyed great success in speech recognition, partially due to their ability to model longer-distance context than word n-gram models. In recurrent neural networks (RNNs), contextual information from past inputs is modeled with the help of recurrent connections at the hidden layer, while Long Short-Term Memory (LSTM) neural networks are RNNs that contain units that can store values for arbitrary amounts of time. While conventional unidirectional networks predict outputs from only past inputs, one can build bidirectional networks that also condition on future inputs. In this paper, we propose applying bidirectional RNNs and LSTM neural networks to language modeling for speech recognition. We discuss issues that arise when utilizing bidirectional models for speech, and compare unidirectional and bidirectional models on an English Broadcast News transcription task. We find that bidirectional RNNs significantly outperform unidirectional RNNs, but bidirectional LSTMs do not provide any further gain over their unidirectional counterparts.

Predicting driver steering intention enables intelligent vehicles to optimize its assistance and collaborative strategies with the human driver in advance, which contribute to an intelligent mutual-understanding system for driver-vehicle collaboration. In this study, a deep time-series learning-enabled driver steering intention prediction system is developed based on the Electromyography (EMG) signal processing. Specifically, the connection between the upper limb EMG signals from different muscles and the steering torque is established using a deep bi-directional long short-term memory (BiLSTM) recurrent neural network (RNN). The deep time-series model is trained to predict the future steering torque with historical EMG signals, and the prediction horizon is selected as 200 ms in this study. Moreover, three different steering postures with different hand positions on the steering wheel are studied. A joint BiLSTM network with shared temporal pattern extraction layers is developed to investigate the impact of the hand positions on the steering intention prediction. It is found that based on the joint BiLSTM network, the most accurate steering intention can be achieved with both hands on 3-clock positions. The experiments are conducted on a driving simulator environment with 21 participants. The proposed system can be used for precise driver steering intention prediction system towards a better mutual-understanding module on the intelligent and automated driving vehicles.