Sripatum Review of Science and Technology

Editorial

2023-12-22T11:02:26+07:00

Diurnal Variation of Visibility with its related Meteorological Factors in Thailand

2022-09-20T13:20:33+07:00

Visibility is an important factor in our daily life and visibility is considered a standard meteorological parameter. The objective was to study diurnal variation of visibility in Thailand and its related meteorological factors. A research study of secondary data consisting of visibility data, cloud data, relative humidity data and rainfall data for a weather 3 hours for 10 years (2009 - 2018) of 43 stations (from all regions in Thailand). To study the diurnal variation and analyze the correlation by using the rescaling (min-max normalization) analysis of visibility and meteorological factors. The results revealed that the diurnal variation and diurnal correlation of visibility and meteorological factors were consistent. Relative humidity has the greatest impact on changes in visibility diurnal variation. Relative humidity greater than 90 percent, except for the Southern (West Coast) is 66.65 percent and this results in low visibility because of the 24-hour variations of the relative humidity and light scattering of aerosols depends on ambient relative humidity since hygroscopic particles absorb significant water at high relative humidity. The factor affecting relative humidity and cloud was air temperature, diurnal variation of pressure, physical characteristics of measurement area and measurement time.

Developing a Model Forecasting Extra Fuel for Airbus A320-200 Landing at Suvarnabhumi Airport: a case study of Thai Smile Airways

2022-10-10T11:07:34+07:00

This research aims to develop a model forecasting the amount of extra fuel of Airbus A320-200 landing at Suvarnabhumi Airport to achieve economy purpose within an acceptable level of flight safety. This research was carried out with mixed methods using quantitative leading qualitative technique by gathering aeronautical variables which was summarized into 8 items derived from captain interviews associated with the secondary data from the 427 historical flight data, divided into 307 training set with 120 testing set. The data were analyzed by Artificial Neural Networks technique to create the Extra Fuel Forecasting Model (EFFM). The backward test forecasting was then performed for the efficiency and expense comparison, along with utilization of the EFFM by pilot in flight. The results found that the accuracy of the EFFM was assessed by the sum of square error, equaled to 0.346. By the relative error, the accuracy value of actual flight test equaled to 91.67%. However, if THAI Smile Airways utilized the EFFM from 1 April 2019 to 31 March 2020, the airline would have reduced expenses by 10,929,197.60 baht. In conclusion, the EFFM is function to save the fuel costs and assure the pilots to operate flight safely.

Efficiency Comparison of the Population Standard Deviation Estimation Methods for Data set with Normal Distribution and Containing Outlier

2022-12-06T16:11:16+07:00

The objective of this research was to compare the efficiency of four population standard deviation estimation methods, mean absolute deviation, adjusted range, adjusted standard deviation and sample standard deviation – for a normal distribution when data set containing outlier when the lowest absolute bias and the lowest mean square error were used as criteria. Under 90 simulation scenarios that normal distribution with mean equals 30 and the population standard deviation equals 1, 5, 10, 15 and 20, sample size equals 10, 20, 30, 50, 70 and 100 with the percentage of outlier equals 0%, 10% and 20%. The results found that sample standard deviation was the most efficient when the data set not contain outlier and it was more efficient if the sample size was larger, but the efficiency would decrease when standard deviation was increase and found that adjusted standard deviation has more efficient than adjusted range in all situation which the data set not contain outlier and found that mean absolute deviation was the most efficient when the percentage of outlier equals 10% and 20%.

Application of Repetitive Scheduling Method and Time-Cost Trade-Off for Road Construction Projects

2023-02-02T12:53:14+07:00

This research aims to study and apply the Repetitive Scheduling Method (RSM) integrated with the Time-Cost Trade-Off method to generate a project planning that is suitable for both duration and cost for two road construction projects. The information on the duration and costs of each activity were collected and analyzed using the RSM and Time-Cost
Trade-Off methods by increasing the number of group workers and equipment or machines in the Controlling Sequence. The results showed that the suitability of the first project was at the project duration of 40 days, compared to the CPM method, reducing the project duration by 39.39%, and reducing the total cost of the project by 5.11%. For the second project, the suitability was at the project duration of 82 days, compared to the CPM method, reducing the project duration by 35.94%, and reducing the total cost of the project by 4.74%. Therefore, this research illustrates the application of the Repetitive Scheduling Method (RSM) integrated with the Time-Cost Trade-Off method that can be applied to projects. The proposed method can effectively reduce the duration and cost of road construction projects.

Preliminary Classification of Gemstone Types on Mobile Application using Deep Learning

2023-07-19T10:13:45+07:00

การตรวจสอบชนิดอัญมณีที่ได้มาตรฐานในปัจจุบันต้องส่งตรวจห้องปฏิบัติการตรวจสอบอัญมณี
ซึ่งวิธีการดังกล่าวมีระยะเวลาดำเนินการและมีค่าใช้จ่ายค่อนข้างสูง จึงมีการนำเทคโนโลยีการเรียนรู้เชิงลึกที่ช่วยให้คอมพิวเตอร์สามารถจำแนกภาพได้อย่างแม่นยำมาประยุกต์ใช้เพื่อแก้ปัญหาดังกล่าว งานวิจัยนี้จึงนำเสนอการพัฒนาแบบจำลองจำแนกชนิดอัญมณีด้วยการเรียนรู้เชิงลึก รวมทั้งพัฒนาโมบายแอปพลิเคชันสำหรับจำแนกชนิดอัญมณีแบบอัตโนมัติ เพื่อให้ผู้ที่ไม่มีองค์ความรู้เกี่ยวกับอัญมณีสามารถจำแนกชนิดอัญมณีได้ การดำเนินงานประกอบด้วยกระบวนการสร้างแบบจำลองจำแนกภาพโดยใช้โครงข่ายประสาทเทียม
คอนโวลูชัน ซึ่งฝึกสอนให้คอมพิวเตอร์สามารถจำแนกภาพอัญมณีที่มีลักษณะทางกายภาพใกล้เคียงกันจำนวน 6 ชนิด ได้แก่ ทับทิม โกเมน ซิทริน บุษราคัม เพริดอตและเขียวส่อง จากการทดลองประเมินประสิทธิภาพของแบบจำลองพบว่า แบบจำลอง MobileNetV1 ให้ค่าประสิทธิภาพดีที่สุด โดยมีค่าความถูกต้อง 95.00%
ค่าความแม่นยำ 95.00% และค่าความระลึก 95.10% มีความเหมาะสมกับการนำไปใช้งาน จากนั้นนำแบบจำลอง
ไปพัฒนาส่วนติดต่อผู้ใช้ในรูปแบบโมบายแอปพลิเคชัน ประเมินประสิทธิภาพแอปพลิเคชันโดยใช้การหาค่าความถูกต้องในการจำแนกกลุ่มบนชุดข้อมูลทดสอบ ผลการวิจัยพบว่า แอปพลิเคชันสามารถจำแนกภาพอัญมณีได้อย่างมีประสิทธิภาพและง่ายต่อการใช้งาน โดยมีค่าความถูกต้องเฉลี่ย 82.50% แสดงให้เห็นว่า
แอปพลิเคชันที่พัฒนาขึ้นสามารถนำไปใช้ประโยชน์ในการตรวจสอบและจำแนกชนิดอัญมณีเบื้องต้นได้

The Development of Monitoring System for Elephant Intrusion Detection in Agricultural Areas to Reduce Human-elephant Conflict with Convolutional Neural network technology

2022-12-20T10:56:48+07:00

Human-elephant conflict occurs due to the migration of elephants from their habitat to human agricultural areas in search of food and water resource. This research has therefore introduced Convolutional Neural Network Technology to reduce Human-elephant conflicts, with a framework called YOLOv5 is utilized for real-time object detection from video footage using embedded neural brain devices to monitor and prevent wild elephant intrusions into agricultural areas. First, researchers collected elephant image datasets from monitoring areas, leveraging the advancements in deep learning frameworks to develop Yolo-based architecture models suitable for embedded devices, ensuring both speed and accuracy. In this work, the researchers adjusted the hyperparameters for the YOLO model variants: YOLOv5N, YOLOv5S and YOLOv5M. The computations were conducted with reduced complexity and the proposed models are well-suited for embedded devices. After testing, it was observed that the YOLOv5S model achieved an average accuracy of 95.68% mAP@.5 with a speed increase of up to 50% compared to the YOLOv5M model, which had a maximum accuracy of 95.46% mAP@.5. Then, before deploying the models on embedded devices, researchers augmented the elephant image dataset from the internet for the YOLOv5S model. This augmentation improved the accuracy to a new value of 98.02% mAP@.5. After deployment, it was found that deep learning could accurately detect instances of wild elephant intrusions into agricultural areas, even though it may sometimes be slightly slower than human observation.

Applying Normalized Difference Vegetation Index from UAV for Fertilizer Cost Reduction in Rice RD33 Cultivation

2023-06-20T16:35:19+07:00

Due to the advanced technology in this era, farmers and government agencies are increasingly using UAVs to assist in cultivation and agriculture management, especially to reduce labor costs while increase agricultural productivity. Thus, we use of remote sensing data obtained from a drone and supervised classification for rice cultivation management to be more efficient. The specific objectives include 1) to analyze the growth of rice RD33 using the Normalized Difference Vegetation Index (NDVI) from an UAV and 2) to classify the cultivated area of rice RD33 from the NDVI image for analyzing the cost of fertilizer. Hua Taphan Model in Amnat Charoen Province is used as the study area in this study. Regarding the results, NDVI of rice RD33 showed the highest values during the reproductive phase (NDVI values between 0.2-0.4), and gradually decreases during the flowering and maturity phase. For data classification, the experimental farm was categorized into 3 classes: bare soil, low-density rice growing and high-density rice growing. Confusion matrix was used for assessing the overall accuracy and the Kappa coefficient which are 80 percent and 0.68 respectively. This research has applied the classification technique to determine the appropriate amount of fertilizer. The results of the research can reduce the amount of fertilizer during rice ripening phase by 6.67 percent.

Modeling Carbon Footprint Measurement Performance in Warehouse Activities

2023-09-07T09:14:13+07:00

The objectives of this research were (1) to study activities that affect on carbon footprint emissions in warehouse, (2) to create a model for measuring the efficiency of carbon footprint emissions in warehouse activities, and (3) to develop guidelines for reduce carbon footprint in warehouse activities. This study was research and development by using qualitative research. In-depth Interview was used to collect data with 17 samples that obtained by purposive sampling, according to the criteria for determining and selecting experts. Descriptive data analysis and triangulation methods were used to analyze data. The results found that (1) warehouse activities include receiving, storing, picking, shipping, and delivering goods affects on carbon footprint emissions. However, effective management of these activities could reduce environmental impacts (2) a warehouse carbon footprint efficiency measurement model was created based on data from a study of warehouse activities and the measurement program of the Thailand Greenhouse Gas Management Organization, the model could assess the carbon footprint of each activity in warehouse by evaluating the total energy use (3) the guidelines for reducing carbon footprints in warehouse activities consist of using determine the source of carbon footprint emissions in warehouse processes and energy using that emits the highest to the lowest, direct and indirect of greenhouse gas emissions, as well as improving and developing transportation methods, using environmentally friendly materials and increasing efficiency in saving energy.

Instructions for preparing the manuscript

2023-12-22T11:07:23+07:00