KMITL

Permanent URI for this communityhttps://dspace.kmitl.ac.th/handle/123456789/1

Browse

Search Results

Now showing 1 - 10 of 20

Some of the metrics are blocked by your
consent settings
The impact of a metallic partial-flow particulate filter on diesel engine combustion and emission characteristics using palm oil biodiesel blends
(2024-02-01)
Thin, Myat Hsu
;
Liu, Hai
;
Thaeviriyakul, Poonnut
;
Wai, Phyo
;
Oh, Ban Seok
Reducing particulate emissions from diesel engines pose a significant challenge in developing countries due to increasingly stringent regulations for both new and older vehicles. While employing wall-flow filters have proven effective for new vehicles, older vehicles necessitate alternative approaches to particulate filtration without complex modifications. Partial flow filters exhibit advantages in such scenarios, characterized by their lack of external components, simplified design, minimal maintenance demands, and resilience to misfuelling. The present study mainly focuses on the evaluation of a partial flow diesel particulate filter (P-DPF) installed on a diesel direct injection compression ignition engine, operating on commercial biodiesel blends, specifically B10 and B20. According to the combustion analyses, the combustion pressure, temperature, and the heat release rate increased with the kinetic energy inside the residual gas molecules due to installation of the P-DPF system. This also resulted in higher indicated power as well as indicated thermal efficiency. However, brake-specific fuel consumption and brake-specific energy consumption, exhibited only a marginal increase, while brake thermal efficiency experienced a slight decrease of 0.65% in the case of B10 and 0.74% for B20 after the installation of the P-DPF system due to the friction loss by the filter backpressure. Furthermore, an incremental increase in exhaust backpressure was observed, ranging from 0.2 kPa at 1000 rpm and 56 Nm to 2.25 kPa at 2000 rpm and 140 Nm. An analysis of emissions limits showed a notable 65% reduction in soot emissions. Comparative analyses were conducted to assess the impact of P-DPF installation on a diesel engine without any manual changes. Ultimately, the partial flow filter (P-DPF) emerges as an effective initial measure in mitigating particulate matter emissions, particularly when employed in a retrofit exhaust after-treatment system.
Some of the metrics are blocked by your
consent settings
Assessing the Implementation and Performance of Automated Trading Software with Non-Biased Human Decisions in the Derivatives Market: Evidence from Thailand
(2024-01-01)
Muayteng, Kittinu
;
Budsaratragoon, Pornanong
;
Nuthong, Chaiwat
In recent years, most financial firms have utilized Artificial Intelligence (AI) technology in Algorithmic Trading (AT) software instead of human traders due to their susceptibility to numerous behavioral anomalies. Therefore, novel automated trading software's profitability and robustness have become an important research area. These software systems leverage historical market data containing a closing price, volume, and technical indicators to produce bi-directional trading signals for long and short positions. However, the AT-based strategies cannot be formulated as a formal mathematical model due to their non-linear nature. Thus, applying an AI-based fuzzy logic approach can make a qualitative trading system feasible for development. Many state-of-the-art AI-based strategies are ineffective regarding risk reduction, robustness, and profit growth. Therefore, we proposed a novel trading algorithm called the Robustness Trading Model in this research study. The recommended model integrates random generation methodology to generate AT strategies for derivative markets. For a case study, we particularly leverage a vital trading market, SET50 Index, and evaluate the approach's effectiveness by comparing results with the Buy & Hold and Mean Reversal strategies. The results show that our approach outperforms these two strategies by minimizing risk and improving profits. More specifically, we have demonstrated that the proposed AI-based approach can predict the future performance of SET50 index futures with up to 57.28% accuracy. Furthermore, this approach can reduce risk (maximum Drawdown) and enhance the total profit per maximum Drawdown (return/risk) when compared with the Buy & Hold and the Mean Reversal strategies. Moreover, we found that the profit-sharing business model has the best commercial software for innovative evaluation systems using financial feasibility compared to the commission-based and subscription business models. Based on these outcomes, we recommend that Thailand's market regulators and policymakers can utilize our model to maximize profits and eliminate any human biases. Finally, we also suggest that the proposed model can be substantially improved by including an extensive set of parameters during the training phase.
Some of the metrics are blocked by your
consent settings
Physicochemical characteristics of ashes deposited on a wall flow diesel particulate filter of compression ignition engine
(2023-06-01)
Rodvanna, Sattatad
;
Srilomsak, Mek
;
Nuthong, Chaiwat
;
Charoenphanich, Chinda
;
Masomtob, Manop
The investigation of physicochemical characteristics of metal oxide ashes, deposited on a diesel particulate filter, was conducted using electron microscopy, energy dispersive spectroscopy, X-Ray fluorescence, and X-Ray diffractometry techniques. Iron is the main component of deposited ash on the DPF, and other ash components consisted of elements such as silicon, calcium, copper, sulfur, phosphorus, zinc, aluminum and minor chromium. It was clarified that this metal oxide ash has catalytic effect that contributed to an enhancement of soot oxidation. The apparent activation energies of soot oxidation on SiC powder and metal oxide ashes powder are approximately 169 and 135 kJ/mol, respectively.
Some of the metrics are blocked by your
consent settings
Traffic signal timings optimization based on genetic algorithm and gradient descent
(2020-05-01)
Yadav, Alok
;
Nuthong, Chaiwat
Traffic congestions are a recurring problem that results in significant losses both financially and environmentally. optimizing traffic signal timings is one of the most cost-effective ways to mitigate such effects. optimization of traffic signal timings capable of minimizing congestion is, however, computationally expensive. Research needs to be conducted to develop algorithms capable of better optimization using fewer computational resources. This paper presents a novel approach to traffic signal optimization that combines genetic algorithms and a gradient descent like algorithm to obtain optimized traffic signal timings. The genetic algorithm is used to arrive at a starting point for gradient descent; gradient descent is then used to obtain further improvement.
Some of the metrics are blocked by your
consent settings
Vehicle classification with deep learning
(2019-02-01)
Maungmai, Watcharin
;
Nuthong, Chaiwat
Nowadays, there are many traffic surveillance systems which are installed in almost every city to record events and traffic. The surveillance system is used for various objectives, e.g. vehicles searching and real-time traffic monitoring, etc. For the searching purpose, the system can be used by policeman such as outlaw’s vehicle identification in crime. Typically, the officers manually identify the vehicle in recorded video according to its appearances. Although the accuracy of this approach is good, it is time-consuming and inclined to faults due to human fatigue for long duration videos. Moreover, hiring employees is costly. Recently, there are several Machine learning methods which can be applied to classify vehicles, e.g. Fuzzy Logic, Decision Tree, Adaboost, Random Forest, Neural Network, etc. Convolutional Neural Network (CNN) is also one of such methods. CNN is a type of Deep Learning which is in the category of the neural network. The method is very well-known in image recognition field at the present because of its performance. In the proposed vehicle classification, there are two vehicle characteristics, i.e. types and colors. Types consist of four classes while colors consist of seven classes. CNN is then used as to classify vehicle images. The experimental results show that CNN can achieve high performance in real-world applications.
Some of the metrics are blocked by your
consent settings
Front moving vehicle detection and tracking with Kalman filter
(2019-02-01)
Rin, Vannat
;
Nuthong, Chaiwat
Immature behavior, fatigue, and inattentional driving have been mainly identified as major issues in safety. Such human behaviors are considered to be the core problem for the occurrences of traffic accidents. As a result, Intelligent Transportation System including Safety Driver Assistance Systems plays an important role in traffic safety. This system becomes the focused research topics recently. This paper aims to make an investigation of front moving vehicles detection and tracking for the forward collision warning system. In the proposed system, there are two major parts, i.e. vehicle detection and vehicle tracking for position prediction. For the detection procedure, this paper makes use of computer vision techniques with shifting three frame difference method on video sequence and combines with edge detection. For object’s boundary selection, this work applies the blob-detection to extract the center of mass and the bounding box features of the detected object. After having selected those features of the front vehicle the tracking process is carried on by utilizing Kalman filter algorithm in order to improve the accuracy of object detection and position prediction. Three videos obtained on the urban structured road in Thailand are tested. Experimental results show that the detection rate of front moving vehicle performs better when Kalman filter is applied for tracking.
Some of the metrics are blocked by your
consent settings
Application of neural networks for vehicle classifiers: Extreme learning machine approach
(2019-01-18)
Jiaramaneepinit, Boonnithi
;
Nuthong, Chaiwat
Machine learning has been a popular topic in research field for many applications. One of the applications is traffic surveillance system. In many areas, traffic surveillance system is installed in order to gather and estimate important traffic information. Nowadays, there are several systems used for information's extracting, and classifying. One of the well-known approaches is decision tree. It uses a tree-like model that decides consequences outcomes from events. However, in some application, decision tree does not perform well. Another widely used approach is neural network, which has promising performance. It has been developed and become one of the most popular computing systems in research field. The traditional approach in training neural network is backpropagation. However, it has several drawbacks. One of them is the training time. In recent decades, Extreme learning machine (ELM) was proposed for training single hidden layer feed-forward neural network (SLFN) in the extremely fast way. It minimizes training error by utilizing dataset in one-shot calculation. This paper focuses on classifiers in traffic surveillance system. The classification divides into two main tasks. One is vehicle types' classification. Another is vehicle colors' classification. Neural networks trained with ELM are applied to the dataset. The performance are then compared to decision tree based approaches with ensemble methods. The experimental results show that ELM achieves better accuracy than of decision tree based approaches in both tasks.
Some of the metrics are blocked by your
consent settings
Extended Extreme Learning Machine: A Novel Framework for Neural Network
(2018-07-02)
Jiaramaneepinit, Boonnithi
;
Nuthong, Chaiwat
Machine learning has been a focus research topic of superior tasks in many real-world applications. One of the famous preferred system is neural network. This approach has been invented for decades but becomes popular recently due to its satisfied results in many applications. The success of applying neural network involves model training which conventionally uses backpropagation method. However, it has many drawbacks. In recent decades, Extreme learning machine (ELM) was first proposed for training single-hidden layer feedforward neural network (SLFN). It optimizes training error by utilizing the whole training dataset with a one-shot calculation. However, for the training in datasets with large number of input features or high-dimensional datasets, original ELM encounters many difficulties. One of them is that the original ELM has no learning process from an input layer. This lead to an incomplete representation of data when it is transferred from one layer to another. Another difficulty involves training instability which causes fluctuation in testing accuracy. This is because networks' input weights are randomly generated. To circumvent these difficulties, the imposing architecture, namely Extended Extreme Learning Machine (X-ELM), is proposed. X-ELM uses ELM as an extension part in order to predict the outputs based on ensemble approach. The proposed framework extends the usage of ELM to apply to more complex network structures, such as networks with multiple hidden layers or networks with multiple computing systems. The proposed framework is applied to vehicles characteristic classifications' datasets. The experimental results show that X-ELM achieves better testing accuracy than of ELM in real-world applications.
Some of the metrics are blocked by your
consent settings
Lane detection using Randomized Hough Transform
(2018-02-07)
Mongkonyong, Peerawat
;
Nuthong, Chaiwat
;
Siddhichai, Supakorn
;
Yamakita, Masaki
According to the report of the Royal Thai Police between 2006 and 2015, lane changing without consciousness is one of the most accident causes. To solve this problem, many methods are considered. Lane Departure Warning System (LDWS) is considered to be one of the potential solutions. LDWS is a mechanism designed to warn the driver when the vehicle begins to move out of its current lane. LDWS contains many parts including lane boundary detection, driver warning and lane marker tracking. This article focuses on the lane boundary detection part. The proposed lane boundary detection detects the lines of the image from the input video and selects the lane marker of the road surface from those lines. Standard Hough Transform (SHT) and Randomized Hough Transform (RHT) are considered in this article. They are used to extract lines of an image. SHT extracts the lines from all of the edge pixels. RHT extracts only the lines randomly picked by the point pairs from edge pixels. RHT algorithm reduces the time and memory usage when compared with SHT. The increase of the threshold value in RHT will increase the voted limit of the line that has a high possibility to be the lane marker, but it also consumes the time and memory. By comparison between SHT and RHT with the different threshold values, 500 frames of input video from the front car camera will be processed. The accuracy and the computational time of RHT are similar to those of SHT in the result of the comparison.
Some of the metrics are blocked by your
consent settings
Vehicle classification using neural network
(2017-11-03)
Sotheany, Nou
;
Nuthong, Chaiwat
Traffic congestion is one of the major problems in the big cities. Many systems are proposed to solve this problem, for example, Intelligent Transportation System(ITS). In general, ITS consists of many subsystems including traffic monitoring. The monitoring system can provide the information of vehicles that pass through the monitoring area. This information can suggests about the status of the traffic. For this reason, developing a system that can classify vehicle with high accuracy plays important role to support ITS. The proposed system of vehicle classification using neural network presents in this paper includes other techniques i.e. vehicle detection and vehicle occlusion handling. Furthermore, Back-propagation neural network and Radial basis function network are implemented for vehicle classification. The training dataset of these two neural network classifiers is taken from a traffic video while the testing dataset is taken from another traffic video. The experimental results show that both techniques can classify vehicles only 71% accuracy because the different size of detected vehicles in those two videos. In order to solve this problem, affine transform function is implemented to scale the object of the testing video. After implementing affine transformation to the system, the accuracy of Back-propagation technique increases to 86% while Radial basis function remains the same. The experimental results illustrate that Back-propagation neural network can classify vehicles with higher accuracy and better performance than Radial basis function network.

KMITL

Browse

Filters

Settings

Sort By

Results per page

Search Results