Together with the fast advancement of the Internet of Things (IoT), smart healthcare applications and systems are equipped with increasingly more wearable sensors and mobile devices. These sensors ...are used not only to collect data but also, and more importantly, to assist in daily activity tracking and analyzing of their users. Various human activity recognition (HAR) approaches are used to enhance such tracking. Most of the existing HAR methods depend on exploratory case-based shallow feature learning architectures, which struggle with correct activity recognition when put into real-life practice. To tackle this problem, we propose a novel approach that utilizes the convolutional neural networks (CNNs) and the attention mechanism for HAR. In the presented method, the activity recognition accuracy is improved by incorporating attention into multihead CNNs for better feature extraction and selection. Proof of concept experiments are conducted on a publicly available data set from wireless sensor data mining (WISDM) lab. The results demonstrate a higher accuracy of our proposed approach in comparison with the current methods.
Cyber Physical Systems and Internet of Things have grown significant attention from industry and academia during the past decade. The main reason behind this interest is the capabilities of such ...technologies to revolutionize human life since they appear as seamlessly integrating classical networks, networked objects and people to create more efficient environments. However, enhancing these technologies with intelligent skills becomes an even more interesting and enticing scenario. In this paper, we propose and illustrate through a number of case studies how Decisional DNA, a multi-domain knowledge structure based on experience, can be implemented as a comprehensive embedded knowledge representation for Internet of Things and Cyber Physical Systems. Decisional DNA gathers explicit experiential knowledge based on formal decision events and uses this knowledge to support decision-making processes. The main advantages of using Decisional DNA are as follows: (i) offers a standardized form of the collected knowledge and experience, (ii) provides versatility and dynamicity of the knowledge structure, (iii) stipulates storage of day-to-day explicit experience in a single configuration, (iv) delivers transportability and shareability of the knowledge, and (v) provides predicting capabilities based on the collected experience. Consequently, test and results of the presented implementation of Decisional DNA case studies support it as a technology that can improve and be applied to the aforementioned technologies enhancing them with intelligence by predicting capabilities and facilitating knowledge engineering processes.
Low damping characteristics have always been a key sticking points in the development of gas bearings. The application of squeeze film dampers can significantly improve the damping performance of gas ...lubricated bearings. This paper proposed a novel hermetic diaphragm squeeze film damper (HDSFD) for oil-free turbomachinery supported by gas lubricated bearings. Several types of HDSFDs with symmetrical structure were proposed for good damping performance. By considering the compressibility of the damper fluid, based on hydraulic fluid mechanics theory, a dynamic model of HDSFDs under medium is proposed, which successfully reflects the frequency dependence of force coefficients. Based on the dynamic model, the effects of damper fluid viscosity, bulk modulus of damper fluid, thickness of damper fluid film and plunger thickness on the dynamic stiffness and damping of HDSFDs were analyzed. An experimental test rig was assembled and series of experimental studies on HDSFDs were conducted. The damper fluid transverse flow is added to the existing HDSFD concept, which aims to make the dynamic force coefficients independent of frequency. Although the force coefficient is still frequency dependent, the damping coefficient at high frequency excitation with damper fluid supply twice as that without damper fluid supply. The results serve as a benchmark for the calibration of analytical tools under development.
Knowledge representation and engineering techniques are becoming useful and popular components of hybrid integrated systems used to solve complicated practical problems in different disciplines. ...These techniques offer features such as: learning from experience, handling noisy and incomplete data, helping with decision making, and predicting capabilities. In this paper, we present a multi-domain knowledge representation structure called Decisional DNA that can be implemented and shared for the exploitation of embedded knowledge in multiple technologies. Decisional DNA, as a knowledge representation structure, offers great possibilities on gathering explicit knowledge of formal decision events as well as a tool for decision making processes. Its applicability is shown in this paper when applied to different decisional technologies. The main advantages of using the Decisional DNA rely on: (i) versatility and dynamicity of the knowledge structure, (ii) storage of day-to-day explicit experience in a single structure, (iii) transportability and shareability of the knowledge, and (iv) predicting capabilities based on the collected experience. Thus, after analysis and results, we conclude that the Decisional DNA, as a unique multi-domain structure, can be applied and shared among multiple technologies while enhancing them with predicting capabilities and facilitating knowledge engineering processes inside decision making systems.
Knowledge and experience engineering techniques are becoming increasingly useful and popular components of hybrid integrated systems used to solve complex real-life problems in different disciplines. ...These techniques offer features such as learning from experience, handling noisy and incomplete data, helping with decision making, and predicting capabilities. In this article, we present a number of different applications of a multidomain knowledge representation structure called decisional DNA that can be implemented and shared for the exploitation of embedded knowledge within different technologies.
With the development of vehicle intelligence technology, the combination of network and vehicle becomes inevitable, which brings much convenience to people. At the same time, hackers can also use ...technical vulnerabilities to attack vehicles, leading to severe traffic accidents and even vehicle crash. Based on this situation, the vehicle information security protection techniques have drawn great attention from researchers. This paper studies the vehicle intrusion detection system (IDS) based on the neural network algorithm in deep learning, and uses gradient descent with momentum (GDM) and gradient descent with momentum and adaptive gain (GDM/AG) to improve the efficiency and accuracy of IDS. The accuracy and efficiency of the proposed model are validated and evaluated by using real vehicles at the end of the paper. Experiments show that the GDM/AG algorithm can achieve faster convergence in comparison with the GDM algorithm in vehicle anomaly detection, and can detect anomaly data at the level of milliseconds. At the same time, the proposed model can adapt itself to detect unknown attacks. The veracity rate ranges from 97% to 98% in directing the adaptation when facing unknown attack types.
Time series classification (TSC) is the problem of categorizing time series data by using machine learning techniques. Its applications vary from cybersecurity and health care to remote sensing and ...human activity recognition. In this paper, we propose a novel multi-process collaborative architecture for TSC. The propositioned method amalgamates multi-head convolutional neural networks and capsule mechanism. In addition to the discovery of the temporal relationship within time series data, our approach derives better feature extraction with different scaled capsule routings and enhances representation learning. Unlike the original CapsNet, our proposed approach does not need to reconstruct to increase the accuracy of the model. We examine our proposed method through a set of experiments running on the domain-agnostic TSC benchmark datasets from the UCR Time Series Archive. The results show that, compared to a number of recently developed and currently used algorithms, we achieve 36 best accuracies out of 128 datasets. The accuracy analysis of the proposed approach demonstrates its significance in TSC by offering very high classification confidence with the potential of making inroads into plentiful future applications.
•Salt-tolerant activated sludge system was successfully acclimated under the salinity of 7.85 g/L.•Coupling induced MAP crystallization unit significantly improved the removal efficiency of TN and ...TP.•Optimal performance of SRNR-SDS process was obtained under the HRT of 20 h.•SRNR-SDS process has obvious environmental, social, and economic benefits.
Marine pollution caused by the untreated and substandard discharge of ship domestic sewage has received widespread attention. A novel integrated process for struvite recovery and nutrient removal from ship domestic sewage (SRNR-SDS) based on seawater magnesium source was developed in this study. Removal efficiencies of the total nitrogen (TN) and total phosphorus (TP) for the activated sludge unit in SRNR-SDS process were approximately 67.61% and 41.35%, respectively, under the salinity of 7.85 g/L. The coupling-induced struvite crystallization unit significantly improved the removal efficiency of TN and TP, and the scanning electron microscopy and X-ray diffraction demonstrated that magnesium ammonium phosphate (MAP) crystals were successfully formed on the surface of zeolite. The SRNR-SDS process had an ideal performance for pollutant removal and MAP recovery under the optimal hydraulic retention time of 20 h. The effluent concentrations of COD, NH4+-N, TN and TP in SRNR-SDS process were approximately 34.73 mg/L, 4.31 mg/L, 10.07 mg/L and 0.23 mg/L, respectively, which meet the Chinese and international ship sewage discharge standards. SRNR-SDS process has obvious environmental, social and economic benefits, which could save 6.20%∼57.14% of the operation cost of ship domestic sewage treatment via MAP recovery. The results could provide theoretical and technical support for the development and application of ship sewage treatment process with the functions of pollutant removal and resource recovery.
Display omitted
To improve the performance and expand the application range of gas-lubricated bearings to mainstream high-power devices, load capacity and damping characteristics are the key sticking points. ...Combining hermetic squeeze film dampers (HSFDs) with porous tilting pad gas bearings (PTPGBs) is a potential option to address this critical issue. A comprehensive bearing rotor test rig was assembled to experimentally investigate the dynamic response of a rigid rotor supported by two HPTPBs (or PTPGB with HSFDs). The rotor with a diameter of 70 cm and a weight of 23.6 kg was successfully accelerated to 20 krpm. Based on the test rig, the effects of bearing parameters such as supply pressure ratio, unbalanced mass, installation of HSFDs and viscosity of the damping fluid on the nonlinear dynamic behaviors are explored. The presence of HSFDs significantly improves the vibration characteristics of bearing rotor systems. This work presents a numerical model of a rotor system supported by PTPGBs with HSFDs to predict the dynamic responses of rotor, and the predictions are in good agreement with the test results. Understanding the effect of different parameters on the dynamic response of the rotor can provide guidance for the optimal design of bearings.
Text classification is a fundamental part of natural language processing and can help with many downstream tasks, such as emotion analysis, question and answer systems, and recommendation systems. ...The graph convolution neural network has the natural superiority in the non - Euclidean space data. For Chinese text data, there is a lot of correlation between the data, using the graph convolutional neural network for text classification can achieve good results. In our experiment, we use a simple one-hot encoding of the word vector to process our words and use of the graph convolutional neural network can achieve 94.24% accuracy in our data set.