Journal of Engineering and Digital Technology (JEDT)

A Model to Predict the Failure of Centralized Air Conditioners with Variable Refrigerant Flow Using Machine Learning

2024-04-11T08:25:27+07:00

The objective of this research was to develop a predictive model for failure events of variable refrigerant flow air conditioning systems using machine learning techniques. The air conditioning failure events were attributed to the amount of refrigerant. The researcher utilized the RapidMiner Studio software to construct models from four algorithm types: Decision Tree, Naïve Bayes, Support Vector Machine, and Neural Networks. The dataset used consisted of 402 records of variable refrigerant flow air conditioning systems Better Cool Co., Ltd., containing features such as refrigerant pressure, subcooled value, superheated value, number of units turned on/off, and operational status (normal or faulty). To evaluate model performance, the accuracy, precision, recall, and F-measure metrics were calculated and compared across the algorithms. The Neural Networks algorithm demonstrated the highest performance, achieving 95.04% accuracy, 94.17% precision, 100% recall, and 97.00% F-measure, enabling accurate prediction of normal or faulty operational status.

Moreover, the researcher enhanced the model using the Stacking Hybrid Ensemble Method technique. This involved combining the top three performing algorithms: Neural Networks, Decision Tree, and Naïve Bayes, for classification. Consequently, the performance metrics improved, with an accuracy of 95.52%, precision of 97.21%, recall of 97.21%, and an F-measure of 97.21%.

A Study of Setting Time of Cement Paste, Workability and Compressive Strength of Mortar Mixed with Concentrated Latex

2024-08-06T14:16:27+07:00

The objective of this research is to study the physical and mechanical properties of cement paste and mortar mixed with the Low Ammonia Concentrate Latex (LA-TZ). For physical properties, the normal consistency and the initial setting time of the cement paste samples are determined as well as the workability of the mortar samples. For the mechanical properties, the compressive strength of mortar mixed with concentrated latex are studied at the ages of 3, 7, 14, and 28 days. The replacement ratios of concentrated latex were 0.5, 1.0, 1.5, and 2.0 percent by weight of water. The results of the experiment showed that concentrated latex causes the cement paste to require more water content to obtain the normal consistency condition as well as the increasing of the initial setting time. It also causes the flow value of the mortar to be increased as well. It continues to increase water requirement to maintain the flow value of 110±5 period. It is also found that replacing concentrated latex at a ratio of 1.0 percent by weight of water is the most appropriate replacement rate because it provides the highest compressive strength after the control mixture. The compressive strength of the mortar at the age of 28 days to 29.7 MPa, which is sufficient for general use. Moreover, the initial setting time has been increased to 30 minutes from the control mixture, which makes the operation even more comfortable.

An Investigation of CFRP Laminates Cured in a Commercial Toaster Oven: Quality and Economic Assessment

2024-10-25T11:08:10+07:00

This study demonstrated the ability to produce high-quality carbon fiber-reinforced plastic (CFRP) laminates with low void contents using a single cure cycle setting up in commercial toaster oven via Vacuum-Bag-Only (VBO) process. The results were discussed in quality and economic views. X-ray micro-CT 3D images of the resulting laminates revealed the 0.28% void content; therefore, it can be classified as very good quality by Purslow chart. Tensile (ASTM D3039) and flexural (ASTM D790) tests were conducted and the results were compared to void-free laminate considered in finite element analysis. The high tensile and flexural strength of laminates were 451.02 MPa and 767.61 MPa, respectively, which were only 12.12% less than simulation results. The cost analysis was conducted to investigate the breakeven point of commercial toaster oven, autoclave and industrial thermal oven for curing CFRP. The results disclosed that the autoclave contributed the highest oven cost and electricity cost. The breakeven point of commercial toaster oven was 54,000 pieces of similar size of CFRP laminate curing by industrial thermal oven. This suggested that commercial toaster oven was economical choice for moderate production capacity and small working place.

Building Information Modeling for Construction Site Layout Planning

2024-07-11T14:49:40+07:00

The planning of construction site layouts is a critical challenge in construction projects, requiring the allocation of suitable positions for temporary facilities and material storage areas within project boundaries and at appropriate time intervals. Failure to do so can result in significant losses and repercussions. This study aims to develop a construction site layout model by identifying optimal position for tower cranes and construction material storage area that are both feasible and efficient, presented in a format that mirrors reality. The methodology extends beyond merely considering the distance of material movement during construction activities. It aligns time intervals with project scheduling to reduce costs related to crane rental, transportation, storage facilities, material shortages, and opportunity costs. The application of a genetic algorithm is employed to discover the best solutions for positioning tower cranes and storage areas. This approach offers the advantage of approximating the best possible solution and revealing trends in improved solutions. The methodology has been applied to a case study project, illustrating that the development of construction site layouts by identifying optimal positions for tower cranes and construction material storage areas can significantly reduce material movement distances, resulting in reduced time and costs.

Determining the Optimal Parameters of Oil Checking Processing Using Engineering Experimental Design Techniques

2024-08-08T02:05:17+07:00

Focuses on reducing waste in the plastic injection molding process by using the Design of Experiments (DOE) technique. Preliminary studies found that the oil level checking devices did not meet production specifications due to the plastic solidifying before completely filling the mold. This issue arose from unstable machine conditions, including injection pressure (P) in kilograms per square centimeter (kg/cm²), injection time (T) in seconds, and injection speed (S) in millimeters per second (mm/s). To identify the optimal parameters, the central composite design method was employed. The experimental parameters used were: injection pressure (P) at 966, 1000, 1050, 1100, and 1134 kg/cm²; injection time (T) at 17, 20, 25, 30, and 33 seconds; and injection speed (S) at 13, 15, 18, 20, and 22 mm/s. The optimal parameters determined were an injection pressure (P) of 1022 kg/cm², injection time (T) of 32 seconds, and injection speed (S) of 17 mm/s. Applying these optimal parameters in actual production resulted in a reduction of waste from 2.27% to 1.68%, and the cost per unit decreased from 5 baht per piece to 4.27 baht per piece.

Development of a Community Data Platform for Analytics and Management Planning

2024-04-17T08:30:11+07:00

Data from each community comprises various categories, influenced by factors such as community condition, climate, lifestyle, culture, local cuisine, plant types, and businesses. Most of this data has not been stored digitally. Analyzing data from individual communities, sub-districts, provinces, and on a national scale can be delayed and incomplete. Some of the data that was previously stored digitally are in separate, scattered, and unconnected storage systems. As a result, managing this data collectively is challenging. This study proposes the design and development of a community data platform for analytics and management planning. The platform supports storing diverse data types such as text, numbers, locations, dates, times, images, and documents. It can link data from other existing systems through defined APIs and stores raw data on Google Firebase Firestore, files on Google Cloud Storage, and data for quick retrieval on Elasticsearch. Furthermore, the platform includes the design and development of applications to record various data categories for each community. It supports the automatic creation of data structures for each data category to make data collection by community members more convenient. The design and development of this multi-data management system allow for flexible gathering of different types of data from each community. As a result, it can be used to address problems and develop cities as desired.

Development of Data Warehouse Using Data Integration Techniques to Reduced Data Redundancy for Workload Evaluation System Based on Academic Staff

2024-09-03T13:52:37+07:00

In the era of big data, data integration to create data warehouses is vital to the development of organizations with a variety of data. There are a lot of redundancies. This redundancy affects the duration of decision-making and the overall efficiency of an information-driven organization. The objectives of the research were to: 1) develop a data warehouse using data integration techniques to reduce data redundancy; and 2) analyze data about redundancy reduction. This paper proposes the data warehouse development, which has five steps. Firstly, the process of data structure study, which uses the SQLyog program as a tool to understand the data structures of KIMs and WE Systems. Secondly, data integration is performed based on the ETL process via the SSIS platform. Thirdly, the data warehouse development process uses the SSMS program to create a new database that consolidates the integrated data. Fourthly, the process of testing on the data warehouse is done using SQL and PHP as tools to create the report systems for decision-making. Finally, the data redundancy analysis process is conducted to determine the rate of data redundancy reduction. This research uses raw data from the WE and KIMs databases, compressing 8,505 entries. The data sets are classified into two groups, which are the research group (5,490 entries) and the academic service group (3,015 entries). According to experimental results, the reduction in data redundancy for the research group between the two systems is 1,381 entries, which is equal to 74.84%. Similarly, the reduction in data redundancy for the academic services group between the two systems is 468 entries, which is equal to 84.47%. The research demonstrates that the development of data warehouses using data integration based on the ETL process can effectively reduce data redundancy within organizations. As a result, the performance of the data warehouse development process can improve the efficiency of decision-making and enhance data-driven operations within the organization.

Providing QoS for Multi-Class Traffic in IEEE 802.11 DCF with Constant Contention Window

2024-08-13T09:28:59+07:00

This paper presents two Medium Access Control algorithms named Limiting Accessible Slots (LAS) and Splitting Accessible Slots (SAS), which enable the IEEE 802.11 distributed coordination function (DCF) with Constant Contention Window protocol to support traffic with different quality of service requirements. LAS algorithm restricts slot access to group 1 users while group 2 users can access all slots. While SAS algorithm restricts slot access to both groups of users. Each group of users can only access the slots allocated to them. There are no slots that are shared by both groups of users. From the results, we found that each group of users has different average delay time for accessing the channel, so these algorithms can be used to provide different Quality of Service (QoS) for multi-class traffic. In conclusion, appropriately setting the accessible slots for group 1 and group 2 users is crucial for achieving the desired QoS levels, highlighting the importance of tailoring these parameters to the specific user distribution.

Real-time Intrusion Detection System Using Machine Learning for Enhanced and Efficient Cybersecurity in Organizations

2024-05-12T20:50:34+07:00

The purpose of this research is to create a real-time cyber intrusion detection system using machine learning to strengthen and enhance cybersecurity within organizations. RapidMiner Studio was used to analyze intrusion data with 4 algorithms: Decision Tree, Naïve Bayes, Random Forest, and Gradient Boosted Trees. The algorithms were evaluated and compared using the Royal Air Force Cyber Intrusion Detection Dataset (RTAF Dataset) to determine their accuracy based on precision and recall values. The top 2 from 4 performing algorithms were then combined using ensemble machine learning techniques. The resulting algorithm was evaluated and compared based on precision and recall values. The algorithm with the highest performance was then used to create a real-time cyber intrusion detection system using machine learning. The results showed that the real-time cyber intrusion detection system using machine learning for strengthening and enhancing cybersecurity in organizations, developed using a software development process combined with ensemble machine learning-based real-time cyber intrusion detection using the Gradient Boosted Trees algorithm in conjunction with the Naïve Bayes algorithm using the Stacking technique, can provide prediction results with an accuracy of 99.77% and a precision of 88.59%.

The Comparison Between Structural Equation Modeling and Multiple Linear Regression Analysis to Predict the Parents’ Intention of Using Child Car Restraint in Bangkok

2024-10-02T09:45:28+07:00

This research aims to study the causal variables influencing on the intention of parents in Bangkok to use child car restraint. The questionnaire is developed based on the Theory of Planned Behavior, Government Influence, Price Fairness, and Safety Awareness. The data are collected from 316 parents in Bangkok who have children’ age under seven years old. The Structural Equation Modelling (SEM) is used to test the causal variables relationship. The results show that the causal variable that most influence on intention to use CCR is safety awareness (SA), since the value of direct effect coefficient () is 0.526. The next casual variable influencing on intention to use is perceived behavior control with the value of effect coefficient () of 0.273, and the third influent casual variable is attitude with value of 0.179. In addition, by applying the Multiple Linear Regression (MLR) with stepwise regression analysis results in the appropriate predictive equation of the parents’ intention to use child car restraint. The coefficients of the MLR model of factor attitudes, perceived behavioral control, and safety awareness are 0.260, 0.212, and 0.530, respectively. The results from the SEM and MLR are the same direction. With the integration method of SEM and MLR, the predictive equation from MLR is well reliable with corresponding to the results from SEM method.

The Use of Prestressed Rubber Pads for Flexural Strengthening of Reinforced Concrete Bridge Pier

2024-08-06T14:16:30+07:00

Repairing and strengthening damaged bridge piers with the loss of the cross-sectional area of the rebar due to rusting has been studied extensively. Several effective methods have been proposed and are widely used. However, most of the materials used for strengthening are quite expensive. Locally available and low-cost materials sometimes are more appropriate alternatives for structural reinforcement. This research aims to study flexural behavior of reinforced concrete bridge piers strengthened using prestressed rubber pads. Samples replicated existing piers of bridges under the responsibility of the Department of Highways were constructed. The percentage of rebars in the control specimen is 1.2%, equal to the minimum percentage of rebars in the existing bridge piers. Three samples were tested under flexure and axial load. Each test specimen had a different reinforcement quantity and approach consisting of 100% of the minimum steel reinforcement amount, 25% of the minimum steel reinforcement amount, and 25% of the minimum steel reinforcement amount along with prestressed rubber pads. The test results showed that strengthening of bridge piers using prestressed rubber pads can increase flexural strength by 46.5% and the analysis results from Response-2000 software provided the flexural capacities of the specimens relevant to those from the experiment.