https://repository.seku.ac.ke/handle/123456789/55 2026-03-22T20:14:19Z 2026-03-22T20:14:19Z Mbandi, Aderiana M. https://repository.seku.ac.ke/handle/123456789/8157 2025-09-10T07:44:56Z 2009-01-01T00:00:00Z

Title: Using linear regression and ann techniques in determining variable importance Authors: Mbandi, Aderiana M. Abstract: The use of Neural Networks in chemical engineering is well documented. There has also been an increase in research concerned with the explanatory capacity of Neural Networks although this has been hindered by the regard of Artificial Neural Networks (ANN’s) as a black box technology. Determining variable importance in complex systems that have many variables as found in the fields of ecology, water treatment, petrochemical production, and metallurgy, would reduce the variables to be used in optimisation exercises, easing complexity of the model and ultimately saving money. In the process engineering field, the use of data to optimise processes is limited if some degree of process understanding is not present. The project objective is to develop a methodology that uses Artificial Neural Network (ANN) technology and Multiple Linear Regression (MLR) to identify explanatory variables in a dataset and their importance on process outputs. The methodology is tested by using data that exhibits defined and well known numeric relationships. The numeric relationships are presented using four equations. The research project assesses the relative importance of the independent variables by using the “dropping method” on a regression model and ANN’s. Regression used traditionally to determine variable contribution could be unsuccessful if a highly nonlinear relationship exists. ANN’s could be the answer for this shortcoming. For differentiation, the explanatory variables that do not contribute significantly towards the output will be named “suspect variables”. Ultimately the suspect variables identified in the regression model and ANN should be the same, assuming a good regression model and network. The dummy variables introduced to the four equations are successfully identified as suspect variables. Furthermore, the degree of variable importance was determined using linear regression and ANN models. As the equations complexity increased, the linear regression models accuracy decreased, thus suspect variables are not correctly identified. The complexity of the equations does not affect the accuracy of the ANN model, and the suspect variables are correctly identified. The use of R2 and average error in establishing a criterion for identifying suspect variables is explored. It is established that the cumulative variable importance percentage (additive percentage), has to be below 5% for the explanatory variable to be considered a suspect variable. Combining linear regression and ANN provides insight into the importance of explanatory variables and indeed suspect variables and their contribution can be determined. Suspect variables can be eliminated from the model once identified simplifying the model, and increasing accuracy of the model.

2009-01-01T00:00:00Z Amolo, Wyclife O. https://repository.seku.ac.ke/handle/123456789/7509 2024-01-26T08:05:19Z 2021-01-01T00:00:00Z

Title: Non-technical power loss reduction and fault management using optimal recloser placement Authors: Amolo, Wyclife O. Abstract: Nowadays, it is rare for a power distribution system to run without a unique protective device to handle transients produced by energy theft, lightning, falling trees, and animals such as monkeys, among other things. Researchers employing reclosers to regulate transients have previously examined reliability needs. Non-technical power loss and cost reduction, on the other hand, have not been adequately addressed in order to improve high-quality power supply. As a result, customers have always had to pay extra for system losses. This thesis discusses optimal reclosing, cost of energy not served, and the firefly algorithm strategy to combat this threat. In the event of temporary faults, reclosers are employed to temporarily or permanently lock out the distribution system, preventing damage to system apparatus. The distribution system successfully functions on computerized intelligent settings, based on predefined transient faults in high-risk locations, with appropriate reclosing. Recloser's accurate reactions in diverse situations are intelligently determined. This thesis built an intelligent system that uses the firefly algorithm to install reclosers at specific points along distribution lines, as well as manage and monitor transient faults. As a result, utilizing the optimal reclosing technique, energy not served (ENS) and associated costs are minimized. The results and analysis of the used method show a cost reduction of sixty-one (61%) on energy not served (ENS) during transient. This saving is made feasible by the recloser's optimal placement and reaction time. Other than the Firefly algorithm, the radial distribution system used to assess this can be replaced with a closed network and another new optimization method. Description: Master of Science (Electrical and Electronics Engineering), 2021

2021-01-01T00:00:00Z Oluchiri, Timothy O. https://repository.seku.ac.ke/handle/123456789/7503 2024-01-25T09:05:38Z 2022-01-01T00:00:00Z

Title: Effects of partial replacement of sand with sawdust and fish scales on the properties of concrete blocks Authors: Oluchiri, Timothy O. Abstract: This study investigated the potential use of sawdust and fish scales as aggregates in partial replacement of sand in the manufacture of masonry blocks. Both sawdust and fish scales are by-products of industrial processes and are considered to be organic waste materials. The methodology used in this study comprised of analyzing the physical properties and compressive strength of the samples. The blocks were manufactured by replacing sand by sawdust and crushed fish scales combined in the following ratios of 5%, 10%, 15% and 20% . Both sawdust and crushed fish scales were subjected to a pretreatment process that involved washing and sun drying them for 24 hrs to remove all impurities and moisture content in them. In the case of fish scales grinding had to be done. Then they were mixed with lime to allow for compatibility with the cement matrix at 5% proportion. Tests for the compressive strength for the masonry blocks were done on the 7th, 14th, 21st and 28th days. The compressive strength of the blended masonry blocks was found to be 15.7N/mm2 at the age of 28 days which was found to be the optimum replacement level after replacement of 5%. The production of masonry with a replacement of up to 5% fine aggregates for the sawdust blend was found to be viable. This research therefore aims to assist the construction industry to achieve low cost housing by use of cost effective and environmentally friendly materials. Description: Master of Science in Civil Engineering, 2022

2022-01-01T00:00:00Z Shitanda, Douglas https://repository.seku.ac.ke/handle/123456789/7478 2024-01-16T13:27:18Z 1994-01-01T00:00:00Z

Title: Physical properties of crude oil from indigenous oilseeds in kenya compared tothose of petroleum and conventional Vegetable oils Authors: Shitanda, Douglas Abstract: Physical properties of crude oil from five indigenous oilseeds were determined together with their formulations based on their mixture with diesel. The effects of temperature and methods of extraction on some of the properties were analysed and the results compared to those of petroleum and conventional vegetable oils. The density of the oils decreased linearly with increase in temperature and varied from a minimum of 9124 kg/rrr' to a maximum of 9474 kg/rn' at 293 "K. The viscosity of the oils decreased with increase in temperature varying from a minimum of 52.2 mm2/s to a maximum of 204.9 mmvs at 293°K. The addition of diesel into the vegetable oils resulted in the decrease of their viscosity. The calorific values of the oils ranged from 38 MJ/kg to 42 MJ/kg increasing with the addition of diesel. Specific heat capacity of the oils was less than 2.4 Kl/kg tK whereas thermal conductivity was greater than 1.0 Kl/hrK m. Ash content of the oils ranged from 0.0048 to 0.42 % and their refractive indices ranged from 1.468 to 1.475. The chemically extracted oils had lower values of flash and fire point compared to the mechanically extracted oils. The values were lower by over 37%. Most of the oils were slightly acidic with a pH range of 5.1 to 6.6. The oils showed a high potential for use as lubricants, hydraulic fluids and as fuel in diesel engines. Description: Master of Science in Agricultural Engineering, 1994

1994-01-01T00:00:00Z