This article analyzes the stability‐related properties of long short‐term memory (LSTM) networks and investigates their use as the model of the plant in the design of model predictive controllers ...(MPC). First, sufficient conditions guaranteeing the Input‐to‐State stability (ISS) and Incremental Input‐to‐State stability (δISS) of LSTM are derived. These properties are then exploited to design an observer with guaranteed convergence of the state estimate to the true one. Such observer is then embedded in a MPC scheme solving the tracking problem. The resulting closed‐loop scheme is proved to be asymptotically stable. The training algorithm and control scheme are tested numerically on the simulator of a pH reactor, and the reported results confirm the effectiveness of the proposed approach.
•A new deep learning based percussion method is developed to detect bolt looseness.•A new 1D-MACLSTM networks is developed to process percussion-induced sound signal.•Compared to current methods, ...1D-MACLSTM has better performance.•A set of experiments were conducted to verify the proposed method in this paper.
In the past decade, bolt looseness detection has attracted much attention. Compared to common approaches that require the implementation of constant-contact sensors, several percussion-based methods have demonstrated their superiorities, including low-cost and easy-to-operate, in detecting bolt looseness. However, some drawbacks may impede the further real-world application of percussion-based methods in detecting bolt looseness. First, current percussion-based methods depend on hand-crafted features, which require the extensive experience of operators. In addition, the ability of current percussion-based methods in anti-noising and adaptability is unknown, since no related investigation has been conducted. Moreover, only single-bolt looseness is considered in the current percussion-based investigation. With these deficiencies in mind, in this paper, we propose a novel percussion-based method that uses a newly developed one-dimensional memory augmented convolutional long short-term memory (1D-MACLSTM) networks. Via the convolutional operation in the 1D-MACLSTM, we can avoid manual feature extraction, and the long short-term memory (LSTM) controller backed by external memory can enhance the ability of anti-noising and adaptability. Finally, three case studies are conducted on a pair of typical multi-bolt connections to verify the effectiveness of the proposed method, which has better performance than current percussion-based methods, particularly in a noisy environment and new scenarios.
Accurate load forecasting is an important issue for the reliable and efficient operation of a power system. This study presents a hybrid algorithm that combines similar days (SD) selection, empirical ...mode decomposition (EMD), and long short-term memory (LSTM) neural networks to construct a prediction model (i.e., SD-EMD-LSTM) for short-term load forecasting. The extreme gradient boosting-based weighted k-means algorithm is used to evaluate the similarity between the forecasting and historical days. The EMD method is employed to decompose the SD load to several intrinsic mode functions (IMFs) and residual. Separated LSTM neural networks were also employed to forecast each IMF and residual. Lastly, the forecasting values from each LSTM model were reconstructed. Numerical testing demonstrates that the SD-EMD-LSTM method can accurately forecast the electric load.
Wind power forecast is playing a significant role in the operation and dispatch of modern power systems. Compared with traditional point forecast methods, interval forecast is able to quantify ...uncertainties effectively. Unfortunately, the stochastic and intermittent nature of wind power has brought significant challenges to get high quality prediction intervals (PIs). In this paper, a novel interval forecast model based on long short-term memory neural networks (LSTM) is proposed to construct PIs with the lower and upper bound estimation method. Besides, an improved PI evaluation criterion is proposed by considering the estimation error of PIs. Moreover, a multi-objective optimization framework is proposed to investigate the relationship between PI estimation error and average width. To tune the parameters of LSTM, the widely used non-dominated fast sort genetic algorithm is further improved by introducing the competitive learning mechanism. The effectiveness of the proposed model and algorithm is demonstrated by a series of experiments based on a real world wind power dataset.
•A novel nonlinear-learning ensemble of deep learning time series prediction is proposed for wind speed forecasting.•A cluster of LSTMs with diverse hidden layers and neurons are introduced to ...explore and exploit the wind speed time series.•One nonlinear-learning regression top-layer composed of SVRM is developed to perform ensemble prediction.•The extremal optimization algorithm is employed to search for the optimal parameters of top-layer SVRM.•The effectiveness of proposed EnsemLSTM is validated on two case studies data collected from a wind farm in China.
As an essential issue in wind energy industry, wind speed forecasting plays a vital role in optimal scheduling and control of wind energy generation and conversion. In this paper, a novel method called EnsemLSTM is proposed by using nonlinear-learning ensemble of deep learning time series prediction based on LSTMs (Long Short Term Memory neural networks), SVRM (support vector regression machine) and EO (extremal optimization algorithm). First, in order to avert the drawback of weak generalization capability and robustness of a single deep learning approach when facing diversiform data, a cluster of LSTMs with diverse hidden layers and neurons are employed to explore and exploit the implicit information of wind speed time series. Then predictions of LSTMs are aggregated into a nonlinear-learning regression top-layer composed of SVRM and the EO is introduced to optimize the parameters of the top-layer. Lastly, the final ensemble prediction for wind speed is given by the fine-turning top-layer. The proposed EnsemLSTM is applied on two case studies data collected from a wind farm in Inner Mongolia, China, to perform ten-minute ahead utmost short term wind speed forecasting and one-hour ahead short term wind speed forecasting. Statistical tests of experimental results compared with other popular prediction models demonstrated the proposed EnsemLSTM can achieve a better forecasting performance.
Accurate estimating the machine health indicator is an essential part of industrial intelligence. Despite having considerable progress, remaining useful life (RUL) prediction based on deep learning ...still confronts the following two challenges. Firstly, the length of condition monitoring data obtained from sensors is inconsistent, and the existing fixed window data processing method cannot fully adapt to all individual samples. Secondly, it is challenging to extract local and global features for long-series prediction tasks. To address these issues, this paper proposes a Multi-task Spatio-Temporal Augmented Net(MTSTAN) for industrial RUL prediction, which enhances the local features of different sensors data through channel attention mechanism, and proposes a skip connected causal augmented convolution network to enhance the global feature extraction in time series. For the industrial scenario of inconsistent data lengths, a multi-window multi-task sharing mechanism is set up to capture various time dependencies among different time scales. The robustness and universality of the model are increased by sharing information among tasks and multi-task window mechanism. Finally, a large number of experiments were carried out on the turbofan aircraft engine run-to-failure prognostic benchmark dataset (C-MAPSS) to evaluate the proposed model, and compared with the existing 14 state-of-the-art approaches. The results show that the enhancement of local and global time series features can effectively improve the prediction accuracy. The Multi-task learning strategy has excellent applicability in dealing with the problem of inconsistent data length.
•A novel prediction framework is proposed.•Three new hybrid models based on the framework are put forward.•Compared to normal methods, the proposed models yield a better prediction accuracy.
In this ...paper, a novel framework for wind speed forecasting is proposed. In the new prediction framework, wavelet transform is firstly adopted to decompose original wind speed history into several sub-series. Then, for low-frequency sub-series, recurrent neural networks are used to extract deeper features, which are fed into suitable machine learning methods for predicting, while others are still predicted by normal methods. Meanwhile, three new hybrid models are established, where support vector machine is taken as the predictor, and the standard recurrent neural network and its variant version: long short term memory neural networks and gated recurrent unit neural networks are employed to extract the deeper features. Four experiments from the real world are conducted through the proposed methods as well as normal algorithms. The results demonstrate that the three new proposed hybrid models based on the novel framework yield more accurate predictions.
Short-term photovoltaic (PV) power forecasting is essential for integrating renewable energy sources into the grid as it provides accurate and timely information on the expected output of PV systems. ...Deep learning (DL) networks have shown promising results in this area, but depending on the weather conditions and the particularities of each PV system, different DL architectures may perform best. This paper proposes a meta-learning method to improve one-hour-ahead deterministic forecasts of PV systems by dynamically blending the base forecasts of multiple DL models to learn under what conditions each model performs best. Four base models of different long short-term memory architectures are used to produce PV production forecasts without using numerical weather predictions, with the objective to enhance the generalizability of the proposed solution. The accuracy of the meta-learner is evaluated using three rooftop PV systems in Lisbon, Portugal. Results indicate that different base models perform best at different PV plants, and meta-learning can improve accuracy by up to 5% over the most accurate base model per plant and up to 4.5% over the equal-weighted combination of the base forecasts. These improvements are statistically significant and even larger during peak production hours.
•Four LSTM models for hour-ahead PV production forecasting.•A meta-learning model to optimally blend the forecasts of the four LSTM models.•Focus on NWP independent approaches, resulting in faster and more reliable models.•Evaluation on three PV plants located in Portugal using representative benchmarks.
Atrial fibrillation (AF) is a common arrhythmia, and its incidence increases with age. Many methods have been developed to identify AF, including both the hand-picked features by experts and the ...recent emerging artificial intelligent (AI) methods. As the traditional hand-picked features have almost reached the boundary of their capability, the AI methods have shown their great potentials to achieve high accuracy for the AF identification. However, some common AI methods, especially deep learning methods, do not provide good properties of interpretability, making it difficult to explore the internal relationship between input and prediction results. In addition, most of the reported methods are only for the intra-patient test of AF and Non-AF. In this study, we try to develop an AF detector based on a twin-attentional convolutional long short-term memory neural network (TAC-LSTM), which can not only generate results with high accuracy but also enable a human-friendly function to provide the possible explanations of the automated extracted features by AI. TAC-LSTM was applied to extract multi-domain features of ECG signals for AF detection and to mine the influence of different input segments on the final prediction results. Finally, the proposed method is validated on the MIT-BIH Atrial Fibrillation Database (AFDB) with intra-patient test and inter-patient test and the results also have shown that multi-domain features extracted by TAC-LSTM can provide more useful information. Collectively, TAC-LSTM can be used for clinicians as an auxiliary diagnostic tool.
•A novel deep neural network with multi-domain input layers is proposed.•CNN and LSTM are employed to extract short-term and long-term dependence features.•In AF detection, multi-domain information is characterized by complementary.•Fusing signal time and time-frequency domain information improves model performance.•Our method mines the relationship between the input segment and prediction results.
•We design and validate various deep learning systems to improve diagnosis of infant cry records.•Considered deep learning systems are deep feedforward neural networks (DFFNN), long short-term memory ...(LSTM) neural networks, and convolutional neural networks (CNN).•All deep learning systems are trained with cepstrum analysis-based coefficients.•Compared to existing models, all deep learning systems were found to be more effective in distinguishing between healthy and unhealthy infant cry records.
Nowadays, deep learning architectures are promising artificial intelligence systems in various applications of biomedical engineering. For instance, they can be combined with signal processing techniques to build computer-aided diagnosis systems used to help physician making appropriate decision related to the diagnosis task. The goal of the current study is to design and validate various deep learning systems to improve diagnosis of infant cry records. Specifically, deep feedforward neural networks (DFFNN), long short-term memory (LSTM) neural networks, and convolutional neural networks (CNN) are designed, implemented and trained with cepstrum analysis-based coefficients as inputs to distinguish between healthy and unhealthy infant cry records. All deep learning systems are validated on expiration and inspiration sets separately. The number of convolutional layers and number of neurons in hidden layers are respectively varied in CNN and DFFNN. It is found that CNN achieved the highest accuracy and sensitivity, followed by DFFNN. The latter, obtained the highest specificity. Compared to similar work in the literature, it is concluded that deep learning systems trained with cepstrum analysis-based coefficients are powerful machines that can be employed for accurate diagnosis of infant cry records so as to distinguish between healthy and pathological signals.