Jurnal Teknik Informatika dan Sistem Informasi

Elevator Control Simulation Using Fuzzy Logic Management

2024-08-07T03:42:00+00:00

The use of a permanent magnet synchronous motor as an actuator in a ropeless elevator presents a number of difficulties that must be overcome for the system to be secure and stable. Detent force, one issue with stabilization systems, will be examined in terms of how well it functions under a fuzzy logic controller using a nonlinear test like changes in load and distance to obtain a policy suitable for application in the industrial sector or other human endeavors. The elevator technologies are designed to provide the necessary passenger floors while taking into account the highest standards of elevator performance and passenger pleasure. This work addresses the problem by developing an elevator group controller using a fuzzy algorithm. This project is designed to handle the necessary passenger traffic density while maintaining acceptable passenger waiting times by integrating a fuzzy controller into an elevator system. Within a set of fuzzy rules, three important linguistic variables are added to improve the performance of the elevator group. These consist of load capacity, priority, distance, and average waiting time (AWT). The necessity of floor priority is lessened when there is a great volume of passenger traffic; instead, the expected arrival time should be decreased. While the real elevator prototype is being programmed using a PIC microcontroller acting as a controller, the simulation was completed to visually verify the fuzzy system's priority. Thus, a set of ambiguous guidelines was developed based on real-world issues, primarily the reduction of waiting times and energy usage. The elevator controller will select which elevator will service which incoming hall request when a few are registered. In order to maximize efficiency for financial reasons, high-rise buildings and the ensuing large number of elevators they require provide a significant logistical challenge in terms of time and space conservation. In order to run the elevators properly, complex elevator group control systems are built.

Android Based Travel Document Application

2024-08-29T17:17:55+00:00

In the digital era with high internet penetration and rapid e-commerce growth, the need for digital solutions to increase business efficiency is increasingly urgent. The Android-based travel letter application is here as the right solution to help the sales team manage tasks, track and create reports easily and efficiently. This application offers various benefits, such as easier mobility, efficiency with automation features, increased productivity, and closer consumer relationships. Features such as a camera to upload proof of travel documents, search to make it easier to find travel documents, and travel history to view travel history of travel documents, further increase the value of this application. This application development uses the prototype method and has been tested with user acceptance testing (UAT) and user satisfaction surveys. The results show that 95.3% of the 95 UAT respondents and 99.6% of the 35 survey respondents were satisfied with the application performance. This shows that this Android-based travel document application is suitable for use and can help improve company performance.

Comparative Study on Performance Evaluation of Eager versus Lazy Learning Methods

2024-08-07T03:46:51+00:00

The major revenue in banking sector is generated long term deposits from customers. Many marketing strategies are implemented to target potential customers by examining their impacted characteristics for decision making. Therefore, machine learning as a scientiﬁc computing has drawn many interest in finding best potential customers especially in predicting whether a long term deposit is subscribed or not. In this research, lazy and eager learning of K-Nearest Neighbours (KNN) and Random Forest (RF) is compared. The computation procedure of the prediction makes a sharp distinction between them and accordingly, RF is proven to be more superior than KNN in the term of Accuracy as much as 96%, Precision 93% and F1 score 0.97. Therefore, the ultimate performance of RF relies on the ability to handle non-linearities and its resistance to overfitting makes RF a suitable choice for many predictive applications.

Keywords— Classification; Easy learning; Lazy Learning, Term Deposit

Designing a Referral Model using the Design Science Research Methodology Approach

2024-08-30T00:12:09+00:00

This research aims to create an effective referral model by applying the Design Science Research Methodology (DSRM) in the context of case X. This methodology focuses on integrating theory and practice to produce innovative solutions. By using this approach, the research proposes a model based on a theoretical framework that has proven effective. Data collection was conducted through surveys, interviews, and direct observation. The main focus of this research is the development of a model that not only enhances efficiency in the referral process but also improves service quality for stakeholders. The final results of the research include supporting artifacts for the solution development process, such as PRDs and flowcharts. It is hoped that the results of this study will provide practical guidance for companies or organizations facing challenges in managing complex referral processes. Therefore, this research is expected to make a significant contribution to the development of efficient and effective referral systems.

Dilated-Convolutional Recurent Neural Network for Music Genre Classification

2024-08-29T17:33:34+00:00

In the digital era, utilizing technology to automatically classify music genres has become very important, especially for applications such as music recommendation, music trend analysis, and digital music library management. This research evaluates the use of Dilated-Convolutional Recurrent Neural Network (D-CRNN) in classifying music genres. This method combines the advantages of Dilated-CNN in capturing longer temporal context with the temporal sequence recognition capability of CRNN. The data used is the GTZAN dataset consisting of 1,000 30-second audio recordings, categorized into 10 music genres. Data preprocessing involved converting the audio recordings into Mel-Frequency Cepstral Coefficients (MFCC) images. The model was tested using data without augmentation and with augmentation, resulting in a total of 15,991 images for training. The results show that the use of D-CRNN can improve the accuracy of music genre classification compared to the conventional CRNN method.

Application of Sentence BERT for Job Competency and Course Similarity

2024-07-15T07:40:50+00:00

This research focuses on the application of the Sentence BERT (S-BERT) model, a specialization of the BERT model and an adaptation of the Transformer architecture specifically designed for the Indonesian language, in exploring the concept of course credit transfer consolidation in accordance with the Merdeka Belajar – Kampus Merdeka program. The aim of this exploration is to develop an Indonesian-language S-BERT model and apply it to search and analyze the similarity between activity sequence descriptions and course syllabus (RPS) descriptions. The results of this similarity analysis are the identification of relevant courses based on the given query. The developed model has shown effective capabilities in searching and determining the similarity between activity sequence descriptions and course syllabus descriptions. Courses identified as relevant to the query demonstrate high similarity and compatibility, indicating that the S-BERT model can be relied upon in the process of course credit transfer consolidation within the context of Merdeka Belajar – Kampus Merdeka.

Acne Severity Detection and Classification: Comparing You Only Look Once Methods

2024-07-16T04:13:07+00:00

Acne (Acne vulgaris) is one of the most common skin diseases, especially on the face. Accurate diagnosis and proper treatment are important for optimal care results and improving the accuracy of detection and classification of acne severity. YOLO (You Only Look Once) is a deep learning method used for object detection in images. This study compares the results and performance of YOLOv5 and YOLOv8 in detecting acne on the face. Several experiments were also conducted with data pre-processing, model size, and the use of different basic hyperparameters on both models to understand the impact and differences between YOLOv5 and YOLOv8. The results show that YOLOv5 overall has higher performance in detecting acne compared to YOLOv8, which requires larger hyperparameter values and model sizes to achieve the most optimal results. Conservative hyperparameters (with relatively smaller values or sizes) on YOLOv5 contribute to better performance.