Indonesian Journal of Electrical Engineering and Informatics (IJEEI)
Not a member yet
776 research outputs found
Sort by
Analysis on performances of the optimization algorithms in CNN speech noise attenuator
Get PDFIn this paper, we studied the effect of the optimization algorithm of weight coefficients on the performance of the CNN(Convolutional Neural Network) noise attenuator. This system improves the performance of the noise attenuation by a deep learning algorithm using the neural network adaptive predictive filter instead of using the existing adaptive filter. Speech is estimated from a single input speech signal containing noise using 64-neuron, 16-filter CNN filters and an error back propagation algorithm. This is to use the quasi-periodic nature of the voiced sound section of the voice signal. In this study, to verify the performance of the noise attenuator for the optimization, a test program using the Keras library was written and training was performed. As a result of simulation, this system showed the smallest MSE value when using the Adam algorithm among the Adam, RMSprop, and Adagrad optimization algorithms, and the largest MSE value in the Adagrad algorithm. This is because the Adam algorithm requires a lot of computation but it has an exellent ability to estimate the optimal value by using the advantages of RMSprop and Momentum SGD
Evaluation on the Effect of EEG Pre-processing and Hyper parameters Tuning to the Performance of Convolutional Neural Network Motor Execution Classification
Get PDFElectroencephalogram (EEG) based classification has achieved a promising performance using deep learning models like Convolutional Neural Network. Various pre-processing strategies such as smoothing the EEG data or filtering are commonly used to pre-process the captured EEG signal before the subsequent feature extraction and classification while hyperparameters tuning might help to improve the classification performance. As well, the number of layers used in the CNN can affect the performance of the classification. In this paper, the number of layers needed for the CNN to classify the EEG data correctly, the effect of apply smoothing to pre-process the EEG signal for modern end-to-end CNN and the effect of enabling hyperparameters tuning during the training phase of CNN is investigated and analyzed. Two CNN models, namely Deep CNN with 5 layers and Shallow CNN with 1 layer, with convincing classification accuracy on motor execution classification as reported in the literature were chosen for this study. Both the CNN models are trained on EEG motor execution dataset with different training strategies and dataset pre-processing. Based on the obtained training and test classification accuracy, Shallow CNN trained with enabling hyper parameters tuning and without smoothing the EEG data achieved the best classification accuracy with average training accuracy of 99.9% and test accuracy of 96.87%. This indicates that CNN does not need to have many layers to correctly classify the motor execution data and the EEG data does not require smoothing
Optimum Switching Angle Of Switched Reluctance Motor Using Response Surface Methodology
Get PDFSwitched Reluctance Motor has numerous advantages compared to another electric motor. Simple structure, low-cost production, robustness, and high fault tolerance have been remarkable milestones. Still, the problem of excitation angle at power converter becomes crucial, especially for traction use, requiring higher torque at low speed for starting and acceleration. Therefore, this research emphasized finding the optimum excitation angle at low speed using Response Surface Methodology, a practical application to achieve the highest torque, as indicated by the best speed in the constant torque region. As a result, using Matlab simulation, the adaptive combination of optimum angles reached 2691 rpm quicker than a single excitation angle with 2568 rpm, an increase of 4.79% higher speed using RSM optimization. According to the experimental data, the adaptive combination of optimum angle achieved 2475 rpm better than the single excitation angle reached 2340 rpm, an increase of 5.77% higher speed using the Response Surface Methodology
A Comprehensive Study of Capacitive Loaded Resonant Converter Topologies for Charging Applications
Get PDFResonant converters (RCs) are perceiving global interests of the research community for its eminent contribution in design of many industrial and commercial applications. Rich literature and well-established technology is available to define the role of RCs in such applications where the load is predominantly passive and resistive. However in applications like charging, the nature of load is often interpreted as capacitive and the knowledge on how a RC reciprocates to such variable, non linear load is limited. Motivated by this, the paper investigates about 25 capacitive loaded resonant structures and each of them is thoroughly analyzed to evaluate various key parameters like the output current, peak input current, and current gain. A comparative study is done to categorize and organize these topologies in regard to each of the said parameters. This provides a quick overview of various resonant converter topologies and helps designers to choose a structure that may fit their application. To this base knowledge, the study is further narrowed down to find suitable topology for charging application and accordingly proposed a novel fourth-order RC topology called LA7. A hardware prototype was built to compare and validate the simulated and measured performances
Classification of EEG Signal by Using Optimized Quantum Neural Network
Get PDFIn recent years the algorithms of machine learning was used for brain signals identifing which is a useful technique for diagnosing diseases like Alzheimer's and epilepsy. In this paper, the Electroencephalogram (EEG) signals are classified using an optimized Quantum neural network (QNN) after normalizing these signals, wavelet transform (WT) and the independent component analysis (ICA), were utilized for feature extraction. These algorithms used to reduces the dimensions of the data, which is an input to the optimized QNN for the purpose of performing the classification process after the feature extraction process. This research uses an optimized QNN, a form of feedforward neural network (FFNN), to recognize (EEG) signals. The Particle swarm optimization (PSO) algorithm was used to optimize the quantum neural network, which improved the training process of the system's performance. The optimized (QNN) provided us with somewhat faster and more realistic results. According to simulation results, the total classification for (ICA) is 82.4 percent, while the total classification for (WT) is 78.43 percent; from these results, using the ICA for feature extraction is better than using WT
Engineering factors as a decision model for choosing the type of renewable energy power plant establishment in Thailand
Get PDFThe objective of this article is to propose the approach about engineering factors as a decision model for choosing the type of renewable energy power plant establishment in Thailand with supporting information to increase the number of renewable energy power plant establishments. Reducing energy consumption through the government and energy development from renewable energy plays an important role to strengthen the energy security for the country by increasing economic stability. To conserve the environment and reduce the use of energy from fossils, which is one of the main causes of global warming, the literature about acceptance, consisting of attitude related to usage, performance expectations, expectations of use, social influence, environmental support and government policy with analysis of decision models including interest in products, desire for products, product knowledge and the possibility to buy in the future to create energy security for the country, was reviewed. The government is making an effort to encourage the private sector to install electricity generation systems using renewable energy with support in various fields, including tax measures and the providing of various benefits to create investment incentives, such as academic information support, import duty exemption on raw materials for solar panels, support with funding loans, and revolving funds through financial institutions
A Mini Review of Peer-to-Peer (P2P) for Vehicular Communication
Get PDFIn recent times, peer-to-peer (P2P) has evolved, where it leverages the capability to scale compared to server-based networks. Consequently, P2P has appeared to be the future distributed systems in emerging several applications. P2P is actually a disruptive technology for setting up applications that scale to numerous concurrent individuals. Thus, in a P2P distributed system, individuals become themselves as peers through contributing, sharing, and managing the resources in a network. In this paper, P2P for vehicular communication is explored. A comprehensive of the functioning concept of both P2P along with vehicular communication is examined. In addition, the advantages are furthermore conversed for a far better understanding on the implementation
Comparative study of battery energy storage systems in a micro-grid based on real-time simulation
Get PDFThe penetration of the distributed energy resources particularly solar and wind power into the electrical system has been increasing, but the intermittent nature of these resources produces perturbations and instability in the electrical grid. Thus, integrating energy storage system into the power grids is one of the best solutions to improve the reliability with performance of electrical systems; ensure power balance and meet consumer demand. Different technologies of energy storage devices have been used to support the integration of renewable energy resources and contribute to improve either the effectiveness with the management of power operating system during critical situations such as power fluctuations in the electrical grid. The main objective of this work is to test the effectiveness of battery energy storage system in reducing active power fluctuations in presence of a perturbation in a micro-grid. In addition, a comparative study is carried out by comparing the response of different battery technologies which are used to support the electrical grid in order to verify the appropriate battery technology for power systems and particularly for energy management in micro-grids during power fluctuations, also, the evaluation of the behaviour, feasibility, performance and effectiveness of BESS is achieved by using real-time simulation
Agricultural Commodity Price Forecasting using PSO-RBF Neural Network for Farmers Exchange Rate Improvement in Indonesia
Get PDFAgricultural commodity price forecasting becomes important for farmers since the knowledge of agriculture commodity price fluctuation can help the farmers to identify the right selling time. Recently, the absence of such the forecasting system makes the farmers decide to sell their commodities to middlemen which in turn, reduces their exchange rate as the length of distribution flow is complicated. The length of distribution flow is started from farmers, middlemen, wholesalers, retailers, and consumers. To address this problem, a forecasting system based on radial basis function neural network (RBFNN) is proposed. To optimize the network’s learning process, particle swarm optimization (PSO)-based learning technique is applied. The RBFNN is chosen because of its ability to generally track irregular signal changing, good speed in learning process and robustness. Meanwhile, the implementation of PSO aims to improve weight values towards global optimum in RBFNN model
Utilising Target Adjacency Information for Multi-target Prediction
Get PDFIn this paper, we explored how information on the cost of misprediction can be used to train supervised learners for multi-target prediction (MTP). In particular, our work uses depression, anxiety and stress severity level prediction as the case study. MTP describes proposals which results require the concurrent prediction of multiple targets. There is an increasing number of practical applications that involve MTP. They include global weather forecasting, social network users’ interaction and the thriving of different species in a single habitat. Recent work in MTP suggests the utilization of “side information” to improve prediction performance. Side information has been used in other areas, such as recommender systems, information retrieval and computer vision. Existing side information includes matrices, rules, feature representations, etc. In this work, we review very recent work on MTP with side information and propose the use of knowledge on the cost of incorrect prediction as side information. We apply this notion in predicting depression, anxiety and stress of 270,322 anonymous respondents to the DASS-21 psychometric scale in Malaysia. Predicting depression, anxiety and stress based on the DASS-21 fit an MTP problem. Often, a patient experiences anxiety as well as depression at the same time. This is not unusual since it has been discovered that both tend to co-exist at different degrees depending on a patient’s experience. By using existing machine learning algorithms to predict the severity levels of each category (i.e., depression, anxiety and stress), the result shows improved precision with the use of cost matrix as side information in MTP