Browsing by Author "Alemu, Getachew(PhD)"

Now showing 1 - 4 of 4
  • No Thumbnail Available
    Item
    Brain Tumor Detection and Segmentation Using Hybrid Intelligent Algorithms: Design and Implementation
    (Addis Ababa University, 2012-11) Megersa, Yehualashet; Alemu, Getachew(PhD)
    In brain tumor diagnosis, clinicians integrate their medical knowledge and brain magnetic resonance imaging (MRI) scans to obtain the nature and pathological characteristics of brain tumors and to decide on treatment options. However, manually detecting and segmenting brain tumors in today’s brain MRI, where a large number of MRI scans taken for each patient, is tedious and subjected to inter and intra observer detection and segmentation variability. Therefore, there is a need for computer aided brain tumor detection and segmentation from brain MR images to overcome the tedium and observer variability involved in the manual segmentation. As result a number of methods have been proposed in recent years to fill this gap, but still there is no commonly accepted automated technique by clinicians to be used in clinical floor due to accuracy and robustness issues. In this thesis, an automatic brain tumor detection and segmentation framework that consists of techniques from skull stripping to detection and segmentation of brain tumors is proposed with fuzzy Hopfield neural network as its final tumor segmentation technique. Through preprocessing, image fusion and initial tumorous slice classification, the final hybrid intelligent fuzzy Hopfield neural network algorithm based tumor segmentation, and tumor region detection and extraction is achieved. The performance of the proposed framework is evaluated on various MR images including simulated and real, normal and tumorous. Quantitatively the method is validated against available ground truth (manual detection and segmentation) using commonly used validation metrics. The final segmentation mean and standard deviation result in Jaccard similarity index, Dice similarity score, sensitivity and specificity are 0.8569 0.0896, 0.9186 0.0638, 0.9480 0.0402 and 0.9917 0.0387 respectively. Quantitative and qualitative segmentation result indicates the potential of the proposed framework. Key words: Brain tumor, Fuzzy Hopfield Neural Network, Segmentation, detection, MRI
  • No Thumbnail Available
    Item
    Decentralized Motion Coordination Method Design using CO-FIELD Approach of SWARM AI metaheuristics for Improving the Reliability of Bus Transit System
    (Addis Ababa University, 2014-11) Mitiku, Sisay; Alemu, Getachew(PhD)
    Bus transit system plays a major role in combating both air pollution and road congestion and is one of the most important modes of transportation. In spite all this however, it is not considered to be reliable mode of transportation by its customers. The complex nature of the transportation system in general and bus transit operation in particular makes it difficult for the application of traditional mathematical model. In this thesis, the problem of regulating and monitoring the reliability of a bus transit system using a SWARM artificial intelligence solution is addressed. The increasing availability of near-real time data from intelligent bus transit system makes the applicability of such solution more attractive. As the bus transit system is distributed and stochastically dynamic because of uncertain inter-stop trip time and uneven passenger distribution, the application of interaction based and emergent self-organized solution such as swarm ai solution is highly recommended. The problem is formulated as a distributed motion coordination problem. A gradient field (co-field) coordination model of swarm artificial intelligence which is inspired by the nature of naturally found fields such as electro statistic and electromagnetic fields is used to solve the proposed model. Multi-agent simulation model is used both to model the bus transit system and to iteratively design the SWARM artificial intelligence metaheuristics. The simulation is implemented with NetLogo integrated development environment so that the desired emergent phenomena is designed and evaluated. Line 31 of Ambessa Awtobis organization, Addis Ababa, Ethiopia, is taken as a case study to improve the reliability of the developed multi-agent simulation. Different simulation experiment is carried out and different measure of effectiveness of the system is collected. The result from the multi-agent simulation experiment shows that the proposed method is adaptive to wider passenger density scenarios. From the result, we can conclude that decentralized metaheuristics of control methods without any sort of formal mathematical model can be a viable solution for improving the bus transit system reliability problem. More over this method also helps to solve the problem of how effectively to utilize the increasingly available huge near-time data from intelligent transit system. Our recommendation is that a research on design support system of swarm Artificial intelligence solution, such as reducing a programming overhead for rapid prototyping of emergent phenomena is worth doing in the future. Key words: Bus transit reliability problem, Bus holding, Computational field, Multi-agent simulation, Case study
  • No Thumbnail Available
    Item
    Leakage Aware Hardware Architecture and Dynamic Power Scheduling
    (Addis Ababa University, 2011-07) Dilbie, Daniel; Alemu, Getachew(PhD)
    In the last couple of decades, battery powered mobile devices such as smart phones have become one of the most prolific electronic devices in history. With this has come an exploding demand for performance and features that cover almost every aspect of our digital multimedia interconnected lives including 3-D gaming, still and video cameras, WAN, Bluetooth, highspeed data connections, and so on. As ever increasing features continue to be integrated into these products, there is an ongoing need to develop innovative ways to reduce power consumption and extend battery life. A core requirement of effective and efficient management of energy is a good understanding of where and how the energy is used: how much of the system’s energy is consumed by which parts of the system and under what circumstances. In this work, a Smartphone is developed, hereafter referred to as the XLP, from the ground up with modular architecture where each module is supplied through an active switch matrix which is memory mapped and updated periodically by the main applications processor in the system. The basic notion of this architecture is achieving zero-leakage power for modules which are not being used. This significantly reduces the idle power consumption which accounts for more than 60% of the average power consumed in smart phones. In addition to this novel approach on the hardware architecture, a stochastic dynamic power scheduling and on-demand power and clock gating policies are developed. A number of possible policies are presented and, under given conditions, one of them is proved to be optimal using the energy response time product (ERP) metrics. The XLP is compared with three commercial smart phones, Openmoko Freerunner, HTC Dream and Google’s Nexus One on similar tests and usage scenarios. The XLP and all these three devices use the ARM microprocessor and run the Linux kernel. The comparison is on performance and power consumption. The XLP is proved to have the lowest power consumption on competitive performance levels.
  • No Thumbnail Available
    Item
    Neural Network Based Data-Driven Clinker Quality Prediction: Case Study on Mugher Cement Factory
    (Addis Ababa University, 2012-10) Negash, Mihreteab; Alemu, Getachew(PhD)
    Soft sensors are key solutions in predicting importance process variables. In process industries, important parameters which are difficult or cost a lot to measure online can be predicted using soft sensors. In this thesis a data driven soft sensor is developed using neural network to predict important clinker quality parameters. The developed predictor is significant and can be categorized to the class of neural network based soft sensors. The significance of the thesis is that it avoids measurement delay incurred while analyzing clinker samples. As a result, quick control actions can be taken and clinker quality can be further improved. This is one of the solutions provided by soft sensors. Many soft sensors have been developed in different application areas and cement factory is the one. Some papers report neural network based predictors that are developed on rotary cement kiln. These works are related to the thesis. However, the thesis has its own new contribution. The first new feature is that it has developed data synthesis strategy. Besides, multiple and advanced neural network architectures are used to get improved result. Moreover, it is of the first kind for the selected case, which is the third line of Mugher cement factory. The thesis is developed stage wise and a desired result is obtained. First, cement production specific to the case is studied. Then, data of all the recorded variables in the factories database is collected. This data collection is accompanied by variable selection and data encoding. The data is processed prior to using it for training the neural networks. This data preprocessing treated missing and outlier values. Based on the cleaned data, new data is synthesized to have enough dataset to work on. Finally, neural network models are developed and trained on this dataset. As a result, neural network models are obtained that can predict LSF, SM, AM and C3S values of clinker with mean square error values of 4.3482, 0.0027, 0.0011 and 10.8759 respectively. In conclusion, in this thesis a neural network based data driven clinker quality predictor is developed. While developing the predictor, Mugher cement factory is used as a case study. The developed predictor estimates LSF, SM, AM and C3S values Key words: Soft sensor, neural network, clinker quality prediction. Soft sensors are key solutions in predicting importance process variables. In process industries, important parameters which are difficult or cost a lot to measure online can be predicted using soft sensors. In this thesis a data driven soft sensor is developed using neural network to predict important clinker quality parameters. The developed predictor is significant and can be categorized to the class of neural network based soft sensors. The significance of the thesis is that it avoids measurement delay incurred while analyzing clinker samples. As a result, quick control actions can be taken and clinker quality can be further improved. This is one of the solutions provided by soft sensors. Many soft sensors have been developed in different application areas and cement factory is the one. Some papers report neural network based predictors that are developed on rotary cement kiln. These works are related to the thesis. However, the thesis has its own new contribution. The first new feature is that it has developed data synthesis strategy. Besides, multiple and advanced neural network architectures are used to get improved result. Moreover, it is of the first kind for the selected case, which is the third line of Mugher cement factory. The thesis is developed stage wise and a desired result is obtained. First, cement production specific to the case is studied. Then, data of all the recorded variables in the factories database is collected. This data collection is accompanied by variable selection and data encoding. The data is processed prior to using it for training the neural networks. This data preprocessing treated missing and outlier values. Based on the cleaned data, new data is synthesized to have enough dataset to work on. Finally, neural network models are developed and trained on this dataset. As a result, neural network models are obtained that can predict LSF, SM, AM and C3S values of clinker with mean square error values of 4.3482, 0.0027, 0.0011 and 10.8759 respectively. In conclusion, in this thesis a neural network based data driven clinker quality predictor is developed. While developing the predictor, Mugher cement factory is used as a case study. The developed predictor estimates LSF, SM, AM and C3S values Key words: Soft sensor, neural network, clinker quality prediction. Soft sensors are key solutions in predicting importance process variables. In process industries, important parameters which are difficult or cost a lot to measure online can be predicted using soft sensors. In this thesis a data driven soft sensor is developed using neural network to predict important clinker quality parameters. The developed predictor is significant and can be categorized to the class of neural network based soft sensors. The significance of the thesis is that it avoids measurement delay incurred while analyzing clinker samples. As a result, quick control actions can be taken and clinker quality can be further improved. This is one of the solutions provided by soft sensors. Many soft sensors have been developed in different application areas and cement factory is the one. Some papers report neural network based predictors that are developed on rotary cement kiln. These works are related to the thesis. However, the thesis has its own new contribution. The first new feature is that it has developed data synthesis strategy. Besides, multiple and advanced neural network architectures are used to get improved result. Moreover, it is of the first kind for the selected case, which is the third line of Mugher cement factory. The thesis is developed stage wise and a desired result is obtained. First, cement production specific to the case is studied. Then, data of all the recorded variables in the factories database is collected. This data collection is accompanied by variable selection and data encoding. The data is processed prior to using it for training the neural networks. This data preprocessing treated missing and outlier values. Based on the cleaned data, new data is synthesized to have enough dataset to work on. Finally, neural network models are developed and trained on this dataset. As a result, neural network models are obtained that can predict LSF, SM, AM and C3S values of clinker with mean square error values of 4.3482, 0.0027, 0.0011 and 10.8759 respectively. In conclusion, in this thesis a neural network based data driven clinker quality predictor is developed. While developing the predictor, Mugher cement factory is used as a case study. The developed predictor estimates LSF, SM, AM and C3S values Key words: Soft sensor, neural network, clinker quality prediction. Soft sensors are key solutions in predicting importance process variables. In process industries, important parameters which are difficult or cost a lot to measure online can be predicted using soft sensors. In this thesis a data driven soft sensor is developed using neural network to predict important clinker quality parameters. The developed predictor is significant and can be categorized to the class of neural network based soft sensors. The significance of the thesis is that it avoids measurement delay incurred while analyzing clinker samples. As a result, quick control actions can be taken and clinker quality can be further improved. This is one of the solutions provided by soft sensors. Many soft sensors have been developed in different application areas and cement factory is the one. Some papers report neural network based predictors that are developed on rotary cement kiln. These works are related to the thesis. However, the thesis has its own new contribution. The first new feature is that it has developed data synthesis strategy. Besides, multiple and advanced neural network architectures are used to get improved result. Moreover, it is of the first kind for the selected case, which is the third line of Mugher cement factory. The thesis is developed stage wise and a desired result is obtained. First, cement production specific to the case is studied. Then, data of all the recorded variables in the factories database is collected. This data collection is accompanied by variable selection and data encoding. The data is processed prior to using it for training the neural networks. This data preprocessing treated missing and outlier values. Based on the cleaned data, new data is synthesized to have enough dataset to work on. Finally, neural network models are developed and trained on this dataset. As a result, neural network models are obtained that can predict LSF, SM, AM and C3S values of clinker with mean square error values of 4.3482, 0.0027, 0.0011 and 10.8759 respectively. In conclusion, in this thesis a neural network based data driven clinker quality predictor is developed. While developing the predictor, Mugher cement factory is used as a case study. The developed predictor estimates LSF, SM, AM and C3S values Key words: Soft sensor, neural network, clinker quality prediction.