Journal of Engineering and Digital Technology (JEDT) 2022-06-27T16:33:09+07:00 Assoc.Prof.Dr.Ruttikorn Varakulsiripunth Open Journal Systems <p><strong>Journal of Engineering and Digital Technology (JEDT)<br>ISSN 2774-0617 (Online)</strong></p> <p>The policy of Thai-Nichi Institute of Technology (TNI) is to support the dissemination of research article to be useful in the development of knowledge base for society, especially in business and industry sectors. Therefore, the academic journal, namely the "Journal of Engineering and Digital Technology (JEDT)" (formerly known as: TNI Journal of Engineering and Technology, ISSN 2672-9989) has been created and published.</p> <p>Scope and Content<br>Engineering Technology, Industrial Technology, Multimedia Technology, Information Technology, Applied Sciences, Physical Sciences, Biological Sciences, Computer Sciences, Chemical Sciences, and related areas.</p> <p>Journal published for 6 months (Semiannual)<br>- Issue 1&nbsp; January - June<br>- Issue 2&nbsp; July - December</p> <p>From January, 2020 or Vol.8 No.1 (2020) TNI Journal of Engineering and Technology will be published in E-Journal only.</p> A PM2.5 Prediction Model Using LSTM Neural Network in Bangkok Area 2022-02-28T08:55:59+07:00 Sriruk Srithongchai <p>PM2.5 has become a serious concern in Thailand, particularly in Bangkok and its vicinity. It has significant impacts on human health, economy and society. A prediction method based on mathematical models is an alternative approach to obtain estimates of PM2.5 concentrations. This research proposes long short-term memory (LSTM) neural networks to develop models to forecast the levels of fine particulate matter in Bangkok, ambient and roadside areas. Correlation analysis was used to select the key variables, and the RMSE and MAPE criteria were employed as forecasting performance measures. Eight different models were constructed using air quality and meteorological data. The results demonstrated that for ambient area, model 2 (which includes the variables PM10, NO2, CO, O3, SO2 and LAGPM2.5) was the best model with an average RMSE of 8.05 and an average MAPE of 27.22. In roadside area, model 8 (which contains the variables PM10, NO2, CO, Temp, Hum, Press, WindSp, WindDir and LAGPM2.5) showed the best performance with an average RMSE of 4.83 and an average MAPE of 22.57. Additionally, the prediction models at roadside site (model 5-8) were more accurate than the others (model 1-4). Estimates based upon short-term past data, 1 day, tended to have smaller forecast errors.</p> 2022-06-27T00:00:00+07:00 Copyright (c) 2022 Journal of Engineering and Digital Technology (JEDT) An Improved Linear Combination of Two Estimators for Reducing the Mean Squared Error in a Sample Survey under Simple Random Sampling 2022-02-09T00:17:55+07:00 Napattchan Dansawad <p>The objective of this paper is to improve the efficiency of a linear combination of two estimators for estimating the population mean using auxiliary information in a sample survey. We also study some properties of the new estimator by using the concept of large-sample approximations and comparing them with some existing estimators through the numerical study. To achieve this, three data sets are used to support the performance of the new estimator. It has been shown that the new estimator is equivalent in terms of efficiency as compared to usual linear regression and it is better than other existing estimators under consideration in the terms of Mean Squared Error (MSE) and Percent Relative Efficiencies (PREs).</p> 2022-06-27T00:00:00+07:00 Copyright (c) 2022 Journal of Engineering and Digital Technology (JEDT) An Innovative Virtual Kitchen Partnership-as-a-Service to Improve Efficiency of Healthy and Hygienic Meal Delivery Service Management in Culinary Industry 2022-02-28T08:56:22+07:00 Metta Ongkasuwan Chanasit Thanathawee Charoen Russametummachot Akechai Judkrue <p>During COVID-19 pandemic, many food service providers and online delivery providers have changed and partnered into new business model known as virtual kitchen partnership-as-a service providers (VKPaaS) on cloud computing with focus on demand for healthy and hygienic meals with fastest delivery service to customers in health-related isolation environment. However, due to increasing problems of inaccurate delivery services, the VKPaaS attempts to improve its search engine with AI-based technologies in cloud computing and algorithms to determine the proper healthy and hygienic meals with fastest, accurate and economical delivery service from the nearest location to customers. The objective of this study is to study factors affecting new COVID generation customers decision to choose and purchase healthy and hygienic meals that have potential impact on management on efficiency of VKPaaS delivery service in health-related isolation environment. The quantitative survey and qualitative in-depth interview methods were used to collect and analyze data from 554 subjects and 18 food service providers in Thailand, China and USA during 3 months in 2020. The findings suggested a new VKPaaS efficient delivery management model for improvement and advancement of VKPaaS delivery service to new COVID generation customers with five vital variables of performance expectancy, effort expectancy, social influence, perceived trust and price that had significant impact on customers decision to choose and purchase healthy and hygienic meals from online VKPaaS providers. Recommendations for further research include state-of-the art areas of crowdsourcing, artistically design for contactless meals, robotic-mobile vehicles for live visualization service, and compliances in new emerging health-related isolation economy.</p> 2022-06-27T00:00:00+07:00 Copyright (c) 2022 Journal of Engineering and Digital Technology (JEDT) Buildings Classification from Satellite Images by Transfer Learning 2021-12-17T23:07:29+07:00 Piyanate Touncha-em Ekarat Rattagan <p>Satellite imaging technology is essential for various applications such as real estate analysis, disaster monitoring, and more. However, as satellite data is big data, it is time-consuming and challenging for humans to analyze it, even a simple task such as detecting the types of buildings in a large area. In this paper, we develop a satellite imaging-analytics technique by applying the transfer learning algorithm to learn and classify different types of buildings in Thailand. The proposed model is learnt and tested on our created datasets, namely 4CateSAT, including the images of buildings including (1) airports, (2) stadiums (football fields), (3) schools, and (4) temples in Thailand. We also apply well-known algorithms to handle the imbalanced data, and the experimental results show that the accuracy of the best model is 96.88%.</p> 2022-06-27T00:00:00+07:00 Copyright (c) 2022 Journal of Engineering and Digital Technology (JEDT) Design and Prototyping of an Automatic Solar Panel Cleaner Based on Arduino 2022-04-04T18:18:49+07:00 Patcharin Intamas Promphak Boonraksa <p>This research paper presents the design and construction of an automatic solar panel cleaner on Arduino. Since dust is an important factor affecting the efficiency of the cell panel, regular washing of the cell panel is necessary for the photovoltaic system to increase its efficiency of the system. This research was tested using an 80 W monocrystalline panel and an Arduino washing control. It is easy to use and cheap to allow the control of the panel to achieve the desired results. In collecting the experimental results, the relationship between the electric power before and after cleaning the panel was collected. In the first case, the solar panel was cleaned every 1 hour. The results showed that the output power of the washed and unwashed panels is approximately the same. Due to too frequent a wash control timer, the dust accumulation is not enough to see the change in power output. Case 2, the solar panel washing time is controlled every 3 months. The results showed that after 3 months of solar panel cleaning intervals, the amount of dust accumulated was about 2.5 grams per square meter. The power was greatly reduced. When cleaning the panel, results in higher power output. Then calculate the efficiency of the panel system increased by automatic panel washing. It was concluded that the calculated efficiency of the panel system was increased to 12.11%.</p> 2022-06-27T00:00:00+07:00 Copyright (c) 2022 Journal of Engineering and Digital Technology (JEDT) Development of Rapid Prototyping for Ceramic Products Using Layer Manufacturing Technique 2022-01-28T08:37:51+07:00 Nopporn Bukwan Sasitron Khonthon Sukhuman Rianthong <p>This research aims to design and develop a rapid prototyping machine and explore the forming parameters, included the performance evaluation of rapid prototyping machine technology for ceramic products which deploy layer manufacturing technique. Procedural of this research started from 3D ceramic product design in Unigraphics NX version 6.0 program, next created a cross section line in the 3D file base on a height of product at each layer in development a rapid prototyping machine. The machine size is 300×400×500 millimeter which is composed of 3 components, 1) Base structure 2) Injection rate control unit 3) Electronic control system and software. In wiring electronic circuit, RNR Motion board is applied. USB port used as communication port between hardware and Mach3 CNC controller software is used to control the motor moving in 4 axes as X, Y, Z and A. Testing has explored to find the effective parameters to create prototype. The control parameters are diameter size of injector which is fixed at 2 millimeters, and PBB compound clay which contained the moisture range as 40-50%. The dependent parameters are injector rate, velocity and interval of forming level. The testing results shown that the appropriate injection rate is 5 mm/min, the velocity of axes moving in x, y, z is 60 mm/min, the appropriate interval of forming level is 0.80 millimeter and results of circle forming style, rectangle forming style and free style forming production are good performed, after put the forming product into flaming with 1200 Celsius, the forming product is perfect, no rupture and no twisted. The performance of rapid prototyping machine is measured by product size after forming, found that the mean of +/- deviation is 0.2 millimeter.</p> 2022-06-27T00:00:00+07:00 Copyright (c) 2022 Journal of Engineering and Digital Technology (JEDT) Sightseeing Guidance System to Maximize Satisfaction Using Real-Time Spot Information 2022-02-06T15:42:59+07:00 Hirotoshi Honma Yuya Sato Yoko Nakajima <p>This study proposes a personalized sightseeing planning system that optimizes travel routes to maximize tourist satisfaction considering cost and time constraints. The proposed mathematical model considers the places the tourist wants to visit, cost, and time available and recommends the optimal number of places that can be visited and the shortest routes to these places. The proposed system could successfully suggest local tourist spots that can be visited in the given time and budget. We believe that our study makes a significant contribution to the literature because travelers at present have to rely on information available from websites, guidebooks, social networking sites, or from family and friends to gather information of places they plan to visit. Such information may be brief or need not be up-to-date as real-time factors like weather, seasons, temperature, time of day, and crowding or recent attractions added may not be available. Further the model can be easily adopted by tourism industry worldwide, while the tourists receive reliable and accurate travel advise to enhance the travelling experience.</p> 2022-06-27T00:00:00+07:00 Copyright (c) 2022 Journal of Engineering and Digital Technology (JEDT) The Model of Sentiment Analysis for Classifying the Online Shopping Reviews 2021-12-22T15:32:15+07:00 Pisit Bowornlertsutee Worapat Paireekreng <p>Nowadays, users can make a decision to order online goods and services from searching information related to goods and services. These may be based on opinions and reviews from previous purchasers as a guideline for purchasing decisions. Moreover, the current opinions information and reviews are enormous and increased all the time. consumers have to spend time for information analytics. Therefore, the model of sentiment analysis regarding goods and services reviews is needed. This research aims to build a model of Sentiment analysis with 3-level of emotion. they are positive neutral and negative, regarding previous user’s reviews and opinions towards online products and services. The techniques used in this research are Machine Learning including Word Segmentation and Bag of Words which compared four categories of sentiment analysis methods: LSTM, SGD, Logistic Regression and Support Vector Machines. There are 5 steps for model building as following: 1) Data Preparation Phase 2) Word Tokenization Phase 3) Training &amp; Streaming Phase 4) Classification phase and 5) Model Evaluation Phase. The consumers’ opinions were gathered the datasets from open data with number of 12,900 comments. The model can help consumers to make a decision for purchasing of goods and services, and help entrepreneurs gain the information. This is to improve products and services in the future. This proposed method can classify the opinions into 3 scales which are positive, neutral and negative opinions. In summary, the proposed sentiment analysis model can perform LSTM accuracy is at 81.27%, Logistic Regression accuracy is at 69%, SGD accuracy is at 66% and Support Vector Machines accuracy is at 65%. However, the LSTM shows better performance on the classification compared to other techniques with deep learning approach. It also found that the F1-score can be implemented for Thai text appropriately.</p> 2022-06-27T00:00:00+07:00 Copyright (c) 2022 Journal of Engineering and Digital Technology (JEDT) The Study of the Effect of Channel Obstacles on Stream Flooding 2022-01-21T17:07:13+07:00 Pongpan Kanjanakaroon Chuchoke Aryupong Surachai Amnuaypornlert Pornyamol Natenapakorn Wichai Namkaew <p>The purpose of this research is to study the effect of obstacles across channels to determine if they increase flooding of the stream. The HEC-RAS model was used to simulate water flow and levels with and without obstructions across streams. The level of water and the height of the stream bank were analyzed to determine the distance of flooding along the stream in both cases. The results showed that the effect on the water level in front of the obstacles such as a weir or a culvert is different. The water level in front of the culvert is lower than the water level in front of the weir at low flow. However, when the stream flow exceeds the flow capacity of the culvert, the water level in front of the culvert will be higher due limited by the cross-section of the culvert receiving the water flow. The flood distances and water levels along the stream under high flow show that the flood distances decrease to 580.6 meters from 953.4 meters along the channel when no obstacles restrict the flow. The existence of obstructions in the channel causes the water level to rise higher than usual thus causing the flood distances to increase along the upstream side of the obstruction while flooding does not occur downstream of the obstruction. This research shows that culverts are prone to causing greater flooding than weirs during heavy rainfall events.</p> 2022-06-27T00:00:00+07:00 Copyright (c) 2022 Journal of Engineering and Digital Technology (JEDT) Using Deep Learning with Thermal Imaging Camera to Record Employee Attendance System 2021-11-22T13:40:57+07:00 Amonpan Chomklin Nuttareepan Nittayoosakulchot <p>Today every country in the world faces a COVID-19 situation, making life different from daily life or work life. Therefore, organizations or companies adopt a primary diagnostic method for COVID-19 by having a system to scan an individual facial temperature using a thermal imaging camera to check if that person has initial symptoms of COVID-19 or not. This research focuses on the development of an attendance record system with a thermal imaging camera combination with Deep Learning to optimize the collection and processing of data to classify or identify employees precisely and reduce step to record the working hours of employees. The experiment found an average of face recognition the mean accuracy was 81.85%, and the mean processing time was 0.26 seconds. The research was satisfying when compared the research on the development of a time recording system with face detection and recognition using the Haar-Like Feature technique to detect faces and using the Local Binary Patterns Histogram to recognize faces with accurate of facial recognition at 48%. According to the experiment, the result was highly satisfying in terms of accurate data and processing time. Moreover, the developed system produces accurate and precise information with convenience and safety.</p> 2022-06-27T00:00:00+07:00 Copyright (c) 2022 Journal of Engineering and Digital Technology (JEDT)