Accurate lithium-ion battery state estimation is crucial for dynamic applications, such as Electric Vehicles (EVs). Artificial Neural Networks (ANNs) are widely explored in recent years due to its high adaptability for any non-linear system. However, ANNs take longer to train with large datasets, especially with complex architectures. This can be solved by using optimization algorithms prior to ANN training that sets the hyper-parameters in the optimal region, leading to high convergence rate, which is desired for EV based battery state estimators. This paper studies some of the recently used population-based optimization algorithms. This includes Firefly, Bat, Gravitation Search, Particle-Swarm, Antlion and Backtracking Search algorithms along with their variants. The performance of each of these algorithms are compared, in terms of optimizing the weights and biases for a Feedforward Neural Network used for lithium-ion battery State of Charge (SoC) estimation. The obtained results show that Antlion optimization Algorithm with Double-Step Search (ALO-DSS) produced low estimation error (0.0194 MSE), with less variation in the performance when run multiple times and a shorter computation time (343.6s) as compared to Firefly Algorithm (605.3s). Thus, ALO-DSS may be considered viable for optimization of ANNs for EV battery state estimation, especially for SoC estimation. Moreover, using optimization algorithms can save training time and computational costs, leading to higher feasibility in deploying ANNs for EV applications.
This research addresses the economic and ecological losses caused by diseases that damage potato leaves, such as Late Blight, Early Blight, Septoria leaf spots, curly leaves, and bacterial wilt. The study utilizes machine learning and deep learning techniques to swiftly and accurately identify these diseases at an early stage, reducing damage and losses to farmers. The research focuses on four categorization groups, including three leaf illnesses and one healthy leaf, with the main goal of providing early detection of disease. The study employs three deep learning models, VGGNet16, RenNet101, and modified AlexNet, with modified AlexNet proving to be the most accurate, achieving a training accuracy of 99.97% and a testing accuracy of 61%.
Edge computing (EC) is an act of bringing computational and storage capability near data sources. It helps to reduce response times and bandwidth requirements. However, the rapid proliferation of edge devices has expanded the attack surface and opportunity for adversaries to penetrate corporate networks. The limited computational abilities of edge devices and the heterogeneous nature of communication protocols further increase the security challenges of EC. Also, the trustworthiness of hardware devices is challenged due to security and privacy threats like trojan insertion, IP cloning, and hardware counterfeits. The application of Machine Language (ML) models in the edge computing paradigm creates a distributed intelligence architecture. Also, Field Programmable Gate Arrays (FPGAs) can exploit Physical Unclonable Functions (PUFs) characteristics to generate and store authentication keys. The PUF structure deployed with ML models in the edge layer can learn its complex input-output mapping from the Challenge and Response pairs (CRPs) to identify the suspicious and unknown responses. This article discusses the security and privacy issues in various layers of the EC architecture and proposes intrusion detection systems through the integration of FPGA-based edge sever and ML models. A PUF-assisted ML framework of the intrusion detection system is proposed to authenticate and detect potential attacks on the network.
The capacity of solar energy worldwide has grown significantly, from 40.277 to 580.159 MW over the last 9 years. The operation of solar panels is prone to defects due to changes in weather or the environment. Different types of defects can occur depending on the phase of the module, such as infant, midlife, or wear-out failure. Several types of defect images can be used to identify photovoltaic panel (PV) defects such as RGB, thermography, and Electroluminescence (EL) images. Recently, many researchers used EL images for PV defect classification due to their availability to create high-contrast patterns on PV modules, making it easier to detect defects, such as cracks or broken cells, that might be difficult to spot with the naked eye. Therefore, this work aims to analyze and classify PV defects in EL images using ShuffleNet, a Convolution Neural Network (CNN) based architecture. For comparison purposes, other CNN architectures, namely MobileNet and SqueezeNet will be implemented. The results show that the ShuffleNet architecture outperforms MobileNet and SqueezeNet architectures in terms of precision (92.53%), recall (92.24%), and F1-score (93.17%) in classifying EL PV module defect images.
Artificial Intelligence (AI) and computer vision have provided various ways to solve problems in our daily life. In this paper, a web-based safety eyewear detection system to detect the presence of safety eyewear in input images or video streams is developed using OpenCV, TensorFlow/Keras, and deep learning. This detection system is necessary to be deployed at risky workplaces such as construction sites to help reduce the risks of accidents and facilitate supervisors to detect workers who do not adhere to the regulations of wearing safety eyewear before entering a construction site. This paper uses a combination of transfer learning techniques using a pre-trained MobileNet architecture and Single Shot Detection framework to build a fast and efficient deep learning-based method for safety eyewear detection. With the help of Streamlit, the model is deployed into a web application to provide a user-friendly interface for the users. This web application can detect faces instantly by applying the safety eyewear classifier efficiently and quickly. Experimental results on the dataset collected demonstrated the superior performance of the proposed model with 98% accuracy.
Many people need to see the pest with their own eyes to know what species the pest is to know if it could destroy the harvest or not. However, there are so many pests out there that are unknown to many people, so people need to look up the pest which cost more effort and time. Technology development in the machine learning field could help to detect pests in real time. The research method proposed in this study is the classification of pests on plants using a convolutional neural network with Google Inception-v3 architecture as image embedding. This study classifies aphids, armyworms, beetles, bollworms, grasshoppers, mites, mosquitos, sawflies, and stem borers with image input. This research also compares Multi-Layer Perceptron (MLP) and Support Vector Machine (SVM) as the final classifier for the architecture. The results of this study show that the best model can detect with 99.87% accuracy.
We investigated the performance of multiwavelength random fiber laser (MWRFL) using a bidirectional semiconductor optical amplifier (SOA) and Lyot filter inside a ring cavity. The bidirectional nature of linear SOA acts as the gain medium while the Lyot filter functions as the wavelength-selective device. The input SOA current, half-wave plate (HWP) angles of polarization controllers (PCs), and the stability of the laser were measured to determine the best optimized lasing line. The best MWRFL performance of 57 lasing lines within the 5 dB spectral range, 27.5 dB extinction ratio (ER), and 11.4 nm multiwavelength bandwidth are achieved through 550 mA of SOA current. The HWP angles of 90° and 120° for PC1 and PC2 respectively aid in the best polarization state of the output spectrum compared to other HWP angles measured at every interval of 30°. The laser had good stability with a maximum peak value deviation of 0.35 dBm at a wavelength range from 1541.0 nm to 1549.0 nm.
This paper presents a Kalman filter based approach in order to solve the problem of tracking a noisy cosinusoidal signal with constant amplitude in the presence of noise. The objective is to estimate the state of the signal accurately, considering the inherent challenges posed by noise corruption. The Kalman filter is utilized as the core algorithm for state estimation, leveraging its ability to combine noisy measurements and a dynamic model to provide optimal estimates. The filter is initialized with zero states and covariance, and the state and covariance estimates are iteratively updated using time updates and measurement update equations. Through extensive simulations, the performance of the proposed Kalman filter-based algorithm is evaluated. The results demonstrate its effectiveness in accurately tracking cosinusoidal signals and mitigating the impact of noise. the Kalman Filter algorithm in this system produces low MSE at about 0.021 and MAE at about 0.111. The metrics results signify the algorithm’s ability to filter noise and estimate the actual state of the system, reflecting its robust tracking performance. The simulation results validate the effectiveness of the proposed approach and highlight its potential to enhance signal tracking accuracy in the presence of noise. Further research can explore the algorithm’s performance in various scenarios and investigate additional modifications to increase its robustness in challenging environments.
Education is the bedrock of any society, it provides a foundation for development; therefore, this foundation must be properly built so that it can provide the desired development. The multilayer perceptron is characterized by a series of layers of interconnected neurons and feedforward neural networks that use standard backpropagation algorithms in their training. As a class of supervised learning model, they always require a desired response to be trained. This paper used student records collected from schools across Nasarawa State, to derive insights from the dataset. Multilayer perceptron architectures were used to predict students’ performance. The model consists of 8 layers; an input layer, six hidden layers, and an output layer. Accuracy, precision, recall, and confusion metrics were used to measure the performance of the model. The student dataset was analyzed to address the problem of mass failure in senior secondary schools in Nigeria. The result of the analysis showed that only 11% of students passed with marks greater than 50 in their core courses. Most students that passed with good facilities were from private schools. These results show that there was a massive failure, and it is mostly from government-owned schools that lack portable drinking water, stable electricity, and other infrastructures. The model presented in this paper predicted student performance 97% correctly. This paper made recommendations on what government and the society need to do to improve the quality of education in Nigeria.
Dry cough has been recognized as a common symptom of coronavirus respiratory diseases, emphasizing the importance of accurately identifying and classifying cough types to mitigate the spread of the disease. The study employs various acoustic features and a Python-based data processing algorithm to extract and analyze the Energy Envelope Peaks, Crest Factors, Zero-Crossings, and Formant Frequencies 1-4 from a dataset of 870 cough samples. The analysis of 347 wet cough sound samples and 523 dry cough samples reveals distinctive characteristics. Wet coughs exhibit a higher number of peaks and zero-crossings, while dry coughs display a slightly higher crest factor on average. Moreover, the F1 and F2 formant frequencies are higher in wet coughs, whereas the F3 and F4 formant frequencies are higher in dry coughs. To classify the cough types, both Support Vector Machine (SVM) and Logistic Regression Method (LRM) classifiers are trained using the identified features. The SVM classifier achieves an average accuracy of 71.26%, sensitivity of 72.73%, specificity of 70.87%, and F1-score of 67.94% during testing. Similarly, the LRM classifier achieves an accuracy of 71.26%, sensitivity of 70.59%, specificity of 71.55%, and F1-score of 68.45%. Such automated classification systems have the potential to aid in the early detection and monitoring of respiratory diseases in enclosed spaces.
Using multiple-input multiple-output (MIMO) technology has revolutionized the radar field by providing higher resolution, improved target detection, and reduced interference. However, interrupting signals can significantly degrade the Performance of MIMO radar systems to transmit and receive signals simultaneously; MIMO radar systems employ multiple antennas at the transmitter and receiver. This allows for improved spatial resolution and increased target detection capabilities. However, MIMO radar systems are susceptible to fading and interference, degrading performance. STC is a technique to mitigate these effects by transmitting multiple signals simultaneously from different antennas. This article evaluates the performance of space-time coding (STC) with coherent multiple-input multiple-output (MIMO) radar. The paper investigates the impact of STC on the radar system's detection and estimation capabilities in various scenarios, including target detection in cluttered environments and target tracking in the presence of interference. The simulation results demonstrate that STC can significantly improve the radar system's performance by reducing interference effects and increasing the signal-to-noise ratio. Furthermore, the study shows that different STC schemes have varying degrees of effectiveness depending on the specific scenario. Overall, this research provides valuable insights into the potential benefits of using STC with coherent MIMO radar for various applications.
The limited efficiency of a photovoltaic (PV) system, which is affected by solar radiation and operating temperature, hinders their widespread adoption. This study concentrates on developing a user-friendly MATLAB/Simulink model to investigate the effect of temperature and irradiance on the performance of solar cells and aims to improve the comprehension of solar cell behavior and to develop a method for accurately predicting output under different conditions. The investigation commences with a theoretical analysis of temperature and irradiance dependencies, including their impact on current-voltage and power-voltage characteristics. The Simulink model is then used to simulate and analyze these correlations, while experimental data gathered using a dual-axial solar tracker is used to provide independent validation. The validation data for open circuit voltage only deviates by 5% from the simulation value, while the deviation for short circuit current is 18%. These findings facilitate the optimization of solar energy systems and contribute to a deeper comprehension of solar cell behavior.
Waste generation is a significant challenge exacerbated by factors such as population growth, industrialization, and urbanization, particularly in densely populated areas like Metro Manila. Consequently, effective waste management becomes increasingly crucial. This study focused on the development of a waste management device designed to address waste generation at its source, specifically targeting the residential sector. The researchers have created a semi-automated kitchen waste composter to facilitate the conversion of kitchen waste into valuable fertilizers, even for individuals with limited composting knowledge. The system incorporates a sensor network that monitors key parameters of the composting process, including temperature and moisture levels. An actuator network ensures that these parameters remain within optimal ranges. Additionally, an image processing algorithm has been implemented to detect compost maturity. By implementing this waste management device, households in urban areas can actively contribute to waste reduction efforts. It empowers individuals to participate in composting without requiring extensive expertise. This technology represents a promising solution to mitigate waste generation and promote sustainable practices in residential settings, ultimately contributing to a cleaner and more environmentally friendly urban environment.
In recent years, music streaming has emerged as the primary method for users to enjoy their favorite songs. Music Streaming Services (MSS) have risen to become influential entities in the music industry, exemplified by Spotify’s impressive 406 million monthly active users by the end of 2021. However, many artists express concerns that the current MSS model fails to adequately compensate them for their music, raising questions about fairness. To tackle this issue, this paper proposes the implementation of IOTA-MSS, a Pay-per-Play Music Streaming System that leverages the secure and scalable IOTA distributed ledger technology (DLT). By harnessing the capabilities of IOTA’s DLT, IOTA-MSS offers a platform that enables seamless microtransactions. This innovative system empowers users to play and distribute music in real-time while ensuring that rights holders receive corresponding payments. Through the utilization of IOTA’s DLT, IOTA-MSS addresses the challenges faced by artists, providing a more equitable solution for music compensation. By adopting a Pay-per-Play model, this system aligns artist remuneration with the actual usage and popularity of their music. Furthermore, IOTA-MSS enhances the user experience, offering a secure and efficient platform for music enthusiasts to engage with their favorite tunes.
Crude oil is the world's most crucial commodity since it is used as an essential material in many sectors and as the state budget's price base. The Indonesian Crude Price (ICP) swings in response to changes in global crude oil prices. A pick increase in crude oil prices will undoubtedly cause economic disruption. Thus, the movement or fluctuation of ICP is critical for business actors in the energy industry, particularly in the domestic market. As a result, crude oil price forecasting is required to aid businesspeople in making energy-related decisions. This research uses the Moving Average, and ARIMA approaches to develop an appropriate forecasting model for Indonesian crude oil prices. Within five years or 63 months, the forecasting process employed ICP time-series data per month for 12 different types of crude oil. We discovered that the fittest models for ICP forecasting are ARIMA models (0,1,1), (1,1,0), (0,1,0), and (1,2,1) with MAPE at 16.0967%. The ICP forecasting results from April to September 2020 ICP have a good and proper interpretation, except the type of BRC oil indicates inaccurate forecasts.
Facial expressions are the reflections of internal feelings of humans towards a situation. It is viewed as an effective tool of non-verbal communication. Analyzing these facial expressions give an insight into the human behaviour. In the education field, for effective tutoring system, understanding the emotional state of the students in the class is very important. But it is very difficult for teachers to observe every student all the time, so this gives the motivation to develop an automatic emotion monitoring system and sends the overall emotional behavior report to the teacher at the end of the class. The main objective of this proposed work is to develop a Real-Time facial emotion recognition (FER) system that keeps track of individual facial emotions. With this proposed work 7 universal human facial emotions can be recognized. FER-2013 dataset is used for training the CNN model. The existing models lack of giving accurate results with a conclusive report to understand the emotion analysis towards a situation in real-time. In this proposed work, we have implemented a deep learning model to monitor and recognize facial emotions in real-time and generates a report of all the emotions detected during the time interval and gives the count of each detected emotion. For face detection in real-time Viola-Jones algorithm is used. Our model gives a very good accuracy of 90.40% and can recognize emotions with very good accuracy in real-time.
In the context of cryptocurrency forecasting, this paper provides a comprehensive analysis of various prediction methods, including financial methods, statistical methods, machine learning, and deep learning. It investigates the causes of the effectiveness of the most well-known and accurate techniques. In addition, the study compares the results of its RMSE, MAE, and MAPE to those of other studies that have used the same dataset with same time period. This study investigates the effect of different evaluation matrices on the accuracy of models and compares the performance of two distinct cryptocurrencies on various deep-learning models. By analyzing the relationship between evaluation metrics and the accuracy of price predictions, the study aims to facilitate the development of more precise models for predicting the prices of cryptocurrencies. This study adds to the literature on cryptocurrency forecasting by evaluating several approaches to see which methods provide the most reliable results. Researchers and practitioners can make informed decisions regarding the development and application of cryptocurrencies if they comprehend the factors that contribute to the accuracy of cryptocurrency prediction models. In addition, the study emphasizes how bi-LSTM and LSTM can be used to forecast various cryptocurrencies and how price fluctuations can be measured and predicted with an accuracy level greater than 80%. Overall, this study contributes to the advancement of knowledge and the development of cryptocurrency price prediction methods, thereby augmenting decision-making processes in cryptocurrency markets.
Although there have been multiple studies related to the tracking of mosquito wingbeat frequencies, not much has been done yet on Philippine mosquitoes. Additionally, most of the current methods for tracking mosquitoes involve actively trapping mosquitoes and evaluating them in a separate laboratory. As a response, an acoustic sensor module was developed using an Arduino microprocessor to identify and classify mosquitoes. Mosquitoes were lured and zapped. The module took sound input using an omnidirectional microphone and a parabolic dish housed in a pipe. The Arduino used digital signal processing to decrease background noise, identify probable mosquito wingbeats, and categorize different mosquito species according to the frequency of the wingbeats. Audacity was used to create recordings for reference, along with manual checking. ThingSpeak, an internet platform, received classified data and provided real-time display and analysis. Real-time classifications from the module were shown, and a histogram showed how frequently identified mosquito wingbeat frequencies were distributed. This made it possible for users to keep an eye on and track the mosquito population in the region where the module was placed. The findings showed that UV light attracts mosquitoes more potently than yeast. However, it was mentioned that combining yeast and UV light as luring techniques would be interesting for further study. The monitoring of mosquito populations is made easier by the interface with ThingSpeak, which offers real-time data display.
The breast tumor is the form of tumor that is diagnosed more frequently than any other type of tumor and it is the leading cause of death from breast cancer among females globally. Breast cancer has the greatest incidence and fatality rate among women worldwide. Early on, it was largely elderly women who were affected, but that has since shifted. Younger women have been more affected by breast tumor in recent years. The most effective methods for dealing with this condition right now include screening for it in early stages and diagnosing it. Rather than relying solely on traditional approaches, it encourages the researcher to instead focus on cutting-edge technologies. Objectives: Breast tumor detection using a textile-based (jeans) flexible antenna. Methods: As a substrate, jeans serve as a model for the proposed textile antenna, while copper serves as a patch (tuning fork shape) and ground plane in a simulated version of the antenna. Multiple simulations were run on CST MWS-2021 to determine the return loss ($S_{11}$), VSWR, 3D & 2-D radiation pattern, surface current radiation efficiency (%) with a focus on the specific absorption rate (SAR) all of which are important for optimizing antenna performance and ensuring human safety in the presence of electromagnetic waves. In order to guarantee the identification process for tumor of varying sizes (R=10, 20 & 30 mm), the return loss results are compared across several cases. The tuning fork shape textile antenna operates on ISM band (5.79 GHz). Findings: With the primary goal of identifying breast tumor, the proposed structure was designed with a flexible textile substrate (jeans), low cost and an excellent radiation efficiency %. Application: The proposed structure is a significant improvement over prior studies in terms of low fabrication cost, adaptability and radiation efficiency %, directivity, SAR value. The SAR value was simulated and found 1.37 W/Kg, 0.837 W/Kg for 1 gm and 10 gm tissue. The suggested antenna meets the SAR standards given by the FCC (1 gm) and the ICNIRP (10 gm). The most significant advantage, however, is that the proposed antenna can be used in conjunction with microwave scattering technology to aid in the identification and detection of breast tumor with just a marginal difference between healthy and unhealthy breast.
