PhD Theses: Department of Computing and Information Technology

Permanent URI for this collection

http://repository.embuni.ac.ke/handle/embuni/3872

Browse

Recent Submissions

Now showing 1 - 2 of 2
  • No Thumbnail Available
    Item
    Design, fabrication and characterization of an appropriate solar thermal electricity generating system
    (2015) Millien Kawira, Erastus
    The sun provides an abundant and clean source of energy. However the supply of this energy is periodic following yearly and diurnal cycles, intermittent, unpredictable and it is diffused. Its density is low compared to the energy flux densities found in convectional fossil energy sources like coal or oil. There have been attempts to produce solar thermal power using parabolic trough technology as was demonstrated by Luz Company which built a solar electricity generating station with a power output of 354 MW in USA. Also the largest solar power plants in the world using parabolic trough technology are the Andasol 1 to 3 which are established in Spain. Therefore it was necessary to undertake design and fabrication of a solar thermal electricity generating system consisting of a collector, steam storage system, heat exchanger, turbine and generator as a local solution for power production. The design layouts were done using auto cad. The testing of the steam storage system and the heat exchanger were done using TEMA (Tubular Exchangers Manufacturers Association Standard and ASME (American Society of Mechanical Engineers). Locally produced heat transfer fluids (water, saline solutions, vegetable oils and engine oils) were tested for their suitability in solar power production using guidelines obtained from National Renewable Energy Laboratories. Some of the parameters investigated included mass flow rates, power output, efficiency, steam flow rate, heat absorbed, heat emitted, evaporation ratio, proportion of flash steam, number of heat transfer units among others. The efficiencies of the concentrator when using the heat transfer fluids were in the range of 48.8% to 60.1% for closed collector and in the range of 46.7% to 56.6% for the open collector. The length of complete discharge for the steam storage system ranged from 4.4 hrs to 6.9 hrs. The power output for the heat transfer fluids were in the range of 287.9 W to 467 W. The steam storage was found to have an efficiency of 93.5 % and a thermal capacity of 4.54 kJ. The rate of heat transfer was an average of 68.4 Js-1kg-1 to 46.3 Js-1kg-1. The thermal efficiency for the heat transfer fluids was in the range of 0.85 to 0.66. Available power from the impulse type turbine was 478.4 Watts, isentropic efficiency was 83.5 %, cycle power output was 497.8 W, turbine output was 468.2 Wand gear efficiency was 87.9 % and generator efficiency of 86.9 %. The overall efficiency of the system was 34.97%. Coupling of the steam storage system and the heat exchanger increased the capacity of steam storage to 4.15 KJ, at a maximum temperature of 249.5 °C and at a pressure of 7.2 Nm-2. Coupling of the steam storage system and the heat exchanger increased the capacity of steam storage to 4.15 kJ, at maximum temperatures of 249.5 °C and at a pressure of 7.2 Nm-2. A single stage impulse turbine was fabricated which had an average efficiency of 61.6% and a maximum power output of 498 W. The solar thermal characterization and collection was done in solar intensities of the average range 700 Wm-2 to 1100 Wm-2. In Coolidge irrigation facility, a thermocline storage tank has a capacity of 19.8 GJ and operates at a temperature of 288 °C. The thermal capacity of the storage system used in this study was 4.15 kJ and was operating at a temperature of 249.9 °C.
  • No Thumbnail Available
    Item
    A Hybrid Of Capsule and Dense Neural Networks with backward Regression model for Coffee leaf disease identification
    (UoEm, 2023-04) Jepkoech, Jennifer
    Diseases that affect coffee have serious repercussions for policymakers, seed industries, farmers, and consumers. Over the years, farmers have had to travel long distances searching for pathologists to check their coffee leaves and inform them which disease affects them. The pathologist approach is time consuming and costly. Automated technology in the agricultural sector has saved farmers’ time. The use of convolutional neural networks (CNN) to classify plant diseases has been implemented and showed reputable results. However, they still experience difficulty due to loss of information through pooling, difficulty in error propagation, computational complexities, and translation invariance, which leads to failure in pose preservation, and loss of shape and texture information. This work developed a hybrid neural network, which adopted three subnets for feature collection. To promote strong gradient flow and signal reception from all preceding layers, hybrid neural network model adopted DensNet loop connectivity where error signals were easily propagated to all the preceding layers in the subnets more directly. The Hybrid neural network model used channel-wise concatenation features from all preceding layers to collect rich feature patterns. This allows the model to combine complex and simple features, hence high generalization. This work used images from Mutira coffee plantation that were taken using a digital camera and with the help of a plant pathologist. The hybrid neural network model displayed a 99.7 % F1 score. In comparison, the conventional CapsNet model displayed 87 % F1 score accuracy on testing our framework based coffee dataset comprising 5 coffee disease categories with 58,000 images. The developed model showed relatively higher and stable accuracy when sensitivity analysis was performed by varying testing and training dataset percentages. Support Vector Machines (SVM), AlexNet, ResNet, VGGNet, Inception V3, Artificial Neural Networks (ANN), and VGG 16 deep learning approaches scored 84.5\%, 88.6\%, 99.3\%, 97.87\%, 99.14\%, and 98.2\%, respectively when the coffee dataset was used. The work used only five features with the target variable instead of the total ten features. This was done through the rigorous backward regression process. Due to the reduced number of features, model computation also reduced. This was also evident from the results of the test error. The overall test error for the developed hybrid neural network model was 0.16 while the test error for the original CapsNet model was 0.26. These findings indicate that hybrid neural network model may be a decent and, in most cases, superior and less expensive alternative for phrase categorization models founded on convolutional neural networks (CNNs). Therefore, several classifiers could be fused to enhance the accuracy of plant diseases classification.