Search

In short-term load forecasting, deep learning models with recurrent units are widely used. For Long Short-term Memory (LSTM) models, in terms of the input features at one particular time-step, there is a weight series for each input feature. Hence, an input feature with a larger value of weight can be regarded as more important to the final forecasting result. Since the weights of the features are time-invariant, the model considers the input features as equally important at each time step, thus lack of the capability of reflecting the importance fluctuation through the time span. To tackle this issue, this paper proposes a forecasting scheme, which integrates mutual information (MI) into bi-directional long short-term memory (BILSTM) network. The MI method is used to extract the importance value of input features at different time-steps, and constitute the fluctuation matrix of the importance values for input features, which is used as coefficients to correct the original input feature. The corrected features are substituted into the BILSTM network, which further improves the forecasting accuracy. Numerical experiments have been carried out based on real-world load data. The results demonstrate the superior performance of the proposed scheme compared with a series of state-of-the-art methods and thus indicate the importance and effectiveness of capturing the fluctuation of the importance value.

Deep learning (DL) based localization and Simultaneous Localization and Mapping (SLAM) has recently gained considerable attention demonstrating remarkable results. Instead of constructing hand-crafted algorithms through geometric theories, DL based solutions provide a data-driven solution to the problem. Taking advantage of large amounts of training data and computing capacity, these approaches are increasingly developing into a new field that offers accurate and robust localization systems. In this work, the problem of global localization for unmanned aerial vehicles (UAVs) is analyzed by proposing a sequential, end-to-end, and multimodal deep neural network based monocular visual-inertial localization framework. More specifically, the proposed neural network architecture is three-fold; a visual feature extractor convNet network, a small IMU integrator bi-directional long short-term memory (LSTM), and a global pose regressor bi-directional LSTM network for pose estimation. In addition, by fusing the traditional IMU filtering methods instead of LSTM with the convNet, a more time-efficient deep pose estimation framework is presented. It is worth pointing out that the focus in this study is to evaluate the precision and efficiency of visual-inertial (VI) based localization approaches concerning indoor scenarios. The proposed deep global localization is compared with the various state-of-the-art algorithms on indoor UAV datasets, simulation environments and real-world drone experiments in terms of accuracy and time-efficiency. In addition, the comparison of IMU-LSTM and IMU-Filter based pose estimators is also provided by a detailed analysis. Experimental results show that the proposed filter-based approach combined with a DL approach has promising performance in terms of accuracy and time efficiency in indoor localization of UAVs.

The prognostics and health management (PHM) plays the main role to handle the risk of failure before its occurrence. Next, it has a broad spectrum of applications including utility networks, energy storage systems (ESS), etc. However, an accurate capacity estimation of batteries in ESS is mandatory for their safe operations and decision making policy. ESS comprises of different storage mechanisms such as batteries, capacitors, etc. Consequently, the measurement of different charging profiles (CPs) has a strong relation to battery capacity. These profiles include temperature (T), voltage (V), and current (I) where the CPs patterns vary as the battery ages with cycles. Consequently, estimating a battery capacity, the conventional methods practice single channel charging profile (SCCP) and hop multiple channel CPs (MCCPs) that cause incorrect battery health estimation. To tackle these issues, this article proposes MCCPs based battery management system (BMS) to estimate batteries health/capacity through the deep learning (DL) concept where the patterns in these CPs are changed as the battery ages with time and cycles. Thus, we deeply investigate both machine learning (ML) and DL based methods to provide a concrete comparative analysis of our method. The adaptive boosting (AB) and support vector regression (SVR) are widely compared with long short-term memory (LSTM), multi-layer perceptron (MLP), bi-directional LSTM (BiLSTM), and convolutional neural network (CNN) to attain the appropriate approach for battery capacity and state of health (SOH) estimation. These approaches have a high learning capability of inter-relation between the battery capacity and variation in CPs patterns. To validate and verify the proposed technique, we use NASA battery dataset and experimentally prove that BiLSTM outperforms all the approaches and obtains the smallest error values for MAE, MSE, RMSE, and MAPE using MCCPs compared to SCCP.

Excessive Power Consumption (PC) and demand for power is increasing on a daily basis, due to advancements in technology, the rise in electricity-dependent machinery, and the growth of the human population. It has become necessary to predict PC in order to improve power management and co-operation between the energy used in a building and the power grid. State-of-the-art Energy Consumption Prediction (ECP) methods are limited in terms of predicting the energy effectively, due to various challenges such as weather conditions and the dynamic behaviour of occupants. Thus, to overcome the drawbacks of these methods, we present an intelligent hybrid technique that combines a Convolutional Neural Network (CNN) with a Multi-layer Bi-directional Long-short Term Memory (M-BDLSTM) method using three steps. When applied to short-term power ECP, this approach helps to provide efficient power management i.e. it can assist the supplier to produce the optimum amount of power. The first step in our proposed method integrates the pre-processing and data organisation mechanisms to refine the data and remove abnormalities. The second step employs a deep learning network, where the sequence of refined data is fed into the CNN via the M-BDLSTM network to learn the sequence pattern effectively. The third step generates the ECP/PC by comparing actual and predicted data series and evaluates the prediction using error metrics. The proposed method achieves better prediction results than existing techniques, thus demonstrating its effectiveness. Furthermore, it achieved the smallest value of the Mean Square Error (MSE) and Root Mean Square Error (RMSE) for individual household dataset using 10-fold Cross Validation (CV) and a hold-out (CV) method.

Recently, person re-identification has become one of the research hotspots in the field of computer vision and has received extensive attention in the academic community. Inspired by the part-based research of image ReID, this paper presents a novel feature learning and extraction framework for video-based person re-identification, namely, the extended global-local representation learning network (E-GLRN). Given a video sequence of a pedestrian, the holistic and local features are simultaneously extracted using the E-GLRN network. Specifically, for the global feature learning, we adopt the channel attention convolutional neural network (CNN) and the bidirectional long short-term memory (Bi-LSTM) networks, which are responsible for introducing a CNN-LSTM module to learn the features of consecutive frames. The local feature learning module relies on the key local information extraction, which is based on the Bi-LSTM networks. In order to obtain the local feature more effectively, our work defines a concept of “the main image group” by selecting three representative frames. The local feature representation of a video is obtained by exploiting the spatial contextual and appearance information of this group. The local and global features extracted in this paper are complementary and further combined into a discriminative and robust feature representation of the video sequence. Extensive experiments are conducted on three video-based ReID datasets, including the iLIDS-VID, PRID2011 and MARS datasets. The experimental results demonstrate that the proposed method outperforms state-of-the-art video-based re-identification approaches.

For the robotic positioning and navigation, visual odometry (VO) system is widely used. However, the errors of the traditional VO accumulate when the robot moves. Besides, this paper proposes a new framework to solve the problem of monocular VO, called MagicVO. Based on the convolutional neural network (CNN) and the bi-directional LSTM (Bi-LSTM), MagicVO outputs a 6-DoF absolute-scale pose at each position of the camera with a sequence of continuous monocular images as input. It does not only utilize the outstanding performance of CNN in extracting the rich features of image frames fully but also learns the geometric relationship from image sequences pre and post through Bi-LSTM to get a more accurate prediction. A pipeline of the MagicVO is shown in this paper. The MagicVO is an end-to-end system, and the results of the experiments on the KITTI and ETH datasets show that MagicVO has a better performance than the traditional VO systems in the accuracy of pose and the generalization ability.

Sign language recognition aims to recognize meaningful movements of hand gestures and is a significant solution in intelligent communication between the deaf community and hearing societies. However, until now, the current dynamic sign language recognition methods still have some drawbacks with difficulties of recognizing complex hand gestures, low recognition accuracy for most dynamic sign language recognition, and potential problems in larger video sequence data training. In order to solve these issues, this paper presents a multimodal dynamic sign language recognition method based on a deep 3-dimensional residual ConvNet and bi-directional LSTM networks, which is named as BLSTM-3D residual network (B3D ResNet). This method consists of three main parts. First, the hand object is localized in the video frames in order to reduce the time and space complexity of network calculation. Then, the B3D ResNet automatically extracts the spatiotemporal features from the video sequences and establishes an intermediate score corresponding to each action in the video sequence after feature analysis. Finally, by classifying the video sequences, the dynamic sign language is accurately identified. The experiment is conducted on test datasets, including DEVISIGN_D dataset and SLR_Dataset. The results show that the proposed method can obtain state-of-the-art recognition accuracy (89.8% on the DEVISIGN_D dataset and 86.9% on SLR_Dataset). In addition, the B3D ResNet can effectively recognize complex hand gestures through larger video sequence data, and obtain high recognition accuracy for 500 vocabularies from Chinese hand sign language.