Deep neural networks (DNN) can be trained to self-learn useful representative features of arrhythmias from raw ECG waveforms. The superior accuracy of DNNs is achieved at a cost of high computational complexity due to the large number of parameters. This limits its deployment to devices with high computing capabilities. In this paper, a lightweight CNN model based on ShuffleNet architecture is proposed as a solution to make the deployment of deep neural networks on small devices feasible. A novel encoding scheme for the label of the training and test set samples is employed, allowing the model to detect multiple classes in one sample. A loss function named Focal loss that proved to be effective when applied for DNN training on imbalanced dataset was also explored in this work. With 9x less number of trainable parameters, the proposed model has outperformed traditional CNN improving the F1-score by 2%.
Concrete is the most widely used construction material. Its properties are affected by the type and proportion of materials used in the mixing process. Inaccurate mixing affects the structure endurance and poses safety risks. Hence, it is critical to verify that the dried concrete is homogeneous, and consistent throughout the structure. In this work, a novel non-destructive material classification technique is proposed. It is based on passing Wi-Fi signals through several metallic and non-metallic samples, then analyzing the Channel State Information (CSI) amplitude and phase components, which are affected by the channel variations in the form of amplitude attenuation and phase shift. Placing different objects in the channel yields different CSI responses. The collected data are preprocessed using an averaging operation, which generates training examples for the 1-DCNN classifier. Accuracies ranging between 93.33%-100% were achieved. Therefore, the method is validated to be used for concrete classification later in the project.
Electrical Impedance Tomography (EIT) is a non-invasive, non-ionizing imaging technique that incorporates the internal conductivity changes to provide the spatial mapping of a region under test. The motivation to create better tactile sensors that have better skin-like properties has led to exploration of EIT-based tactile sensors that are thin, flexible and stretchable. An EIT-based tactile sensor can be realized by connecting electrodes around a conductive membrane boundary and using current excitation, it is possible to determine the internal conductivity distribution. In this paper, an experiment was conducted on an EIT-based tactile sensor to investigate its ability to detect thermal changes; where two aluminium cups at different temperatures (20.8?C and 65.4?C) were placed on the sensor at different times. The experiment showed that the proposed EIT system was able to recognize the 65.4C stimulus, while room temperature stimulus remained undetected; indicating the proposed EIT system is able to detect thermal changes.
In this paper, the Residue Number System (RNS) is employed for the design of an energy efficient, high throughput DNN accelerator. The proposed system is fully RNS-based, requiring no intermediate conversions to a binary representation. This is made possible due to the sign detection mechanism needed for the ReLU and MaxPooling operations that eliminate the need for converters after each CONV layer, while managing to maintain a small maximum word-length among the residue channels. Moreover, a method for reducing the complexity of the convolutional layers is introduced. By identifying common terms that occur frequently due to the small range of each RNS channel, inside each weight kernel, and by rearranging the order of computations, the number of multiplications required for each convolution is reduced up to 97%. The remaining multiplications are also simplified, as they are implemented through shift operations or fixed-operand multipliers.
The UAE has been a pioneer in regulating renewable energy in the middle east. The purpose of this paper is to define the main factors that affect customers' perception of photovoltaic solar projects for the residential sector as an alternative renewable source of electricity in the UAE. Reliably collected responses were used to build a hypothetical model using confirmatory factor analysis (CFA) and structural equation modeling (SEM). The findings showed that financial and environmental aspects are the main contributors to customers' perception in the UAE. Furthermore, it was apparent that different nationalities in the UAE have a similar average perception of PV projects. We suggest that perception may positively impact the intention towards adopting residential PV systems. The findings can be utilized to enhance the perception of customers and increase the tendency to install such projects in the residential sector.
Wear developments due to traffic volumes can negatively affect the slipperiness of flooring surfaces, hence, provide unsafe environments to users in terms of slip, trip and fall incidents. This study will review the current advancements in the literature associated with floors safety and investigate the latest technologies used for assessing the coefficient of friction of flooring surfaces. This study will contribute to enhancing the existing literature by exploring the development of a three-dimensional simulation model using a finite element method and highlight the benefits associated with such models in terms of safety level assessment of flooring tiles.
Increased demand on electricity calls for newer methods to generate and store energy. Newer methods directly imply cleaner methods, well known as green energy. Therefore micro-grids have known a relatively a higher demand recently. Generating energy is one thing, and using it is another. That is mainly why energy storage systems were introduced. This research is about a technoeconomic assessment of batteries, and reversible fuel cells using the Levelized Cost of Storage (LCOS). The research was run on four states and the results show that the LCOS is higher for RFCs compared to 400-kW LIBs, yet lower at the size increases. Sensitivity analysis highlights how heavily the change in the capital cost, the discount rate, roundtrip efficiency, and the lifetime of the system impacts the LCOS. Different current densities and charge rates were applied to test the reliability and resiliency of the grid.
The pharmaceutical supply chain (PSC) is important for efficient healthcare delivery and requires efficient planning and optimized distribution to improve quality of life and save lives. In addition, it is considered as a complex supply chain since it involves demand uncertainties, product perishability leading to expiration, and harmful emissions. This paper addresses the optimization of a three-echelon (PSC) for manufacturing and distributing medication products. The proposed robust optimization model considers aspects of demand uncertainty, product perishability, and emission reduction while being able to be scaled in size as desired. Two different models were optimized using the proposed approach to test its effectiveness and the size scalability. From the obtained results, it can be concluded that managers can use the proposed robust approach to perform a quick and easy optimization for their pharmaceutical supply chains in order to make well-informed decisions.
Recently, Additive manufacturing (AM) gained substantial attention as one of the most innovative technologies and one of the most disruptive innovations that impact the global supply chain and logistics industry. This research survey analysis was used to investigate the current spare parts supply chain challenges in the UAE. It evaluates the understanding, expectations, and readiness to adopt additive manufacturing and digital spare parts concept.
Today, millions of people around the world are using portable electronics devices in their daily lives. A lot of these devices are using rechargeable lithium-ion batteries as their main source of power. The current study aims to study the environmental and economic opportunities obtained from recycling of Lithium-ion batteries. The main valuable metals that used on these batteries are Lithium (Li), Cobalt (Co), and Manganese (Mn). The hydrometallurgical recycling method was used to recycle LIBs and recover most of the metals. SEM/EDS characterization method was used and confirmed that the battery sample that we used was NMC battery and that because the combined weight of these three elements is almost ` of the total sample weight. This research would be important for recovery of some valuable metals from spent LIBs, and it can be improved for higher metal recovery in the future.
