The use of solar photovoltaic systems has increased in the past years in an effort to move towards cleaner energy sources. Solar panels are however affected by negative factors such as dust deposition which hinder their performance. The negative effects that dust deposition has on solar panels depend on how much dust gets deposited on solar panels and how it spreads on the top surface. The spread of dust on solar panels can be uniform where all the solar panels in a entire solar photovoltaic array have the same amount of dust deposition. This is an ideal case and can be defined as uniform dust deposition. However, in real life operation, the spread of dust deposition can vary with one solar panel having a different quantity of dust deposition from another. This is defined as non-uniform dust deposition. Non-uniform dust deposition negatively affects the performance of solar panels by reducing the irradiance that reaches the solar cells thereby reducing the performance of the solar panels. The negative effects of non-uniform dust deposition are more significant over time and when there is no intervention to remove the dust. In practice, the negative effects of non-uniform dust deposition on photovoltaic modules has been addressed by periodically cleaning their top surfaces. Periodic cleaning can however increase the operational costs in terms of the cleaning frequency, time taken, cost of cleaning resources and effectiveness. In this study, we propose an optimal cleaning strategy for the solar power plants that are prone to the non-uniform dust deposition. To develop the optimal cleaning strategy, we first investigate the dust deposition process and develop a model to describe the relationship between the solar power generation and non-uniform dust deposition patterns. Then we formulate an optimization model to identify the most cost-effective solar panel cleaning plan. In the optimisation, the additional revenue due to cleaning the solar panels is formulated as the objective function. The decision variables are the number of photovoltaic strings cleaned at each cleaning interval. To highlight the effectiveness of the proposed solar panels cleaning strategy, the developed cleaning strategy is applied to a case study where analysis of the performances of other solar panel cleaning strategies, namely “full cleaning”, “no cleaning” and “random cleaning” is done. The results from the study show that the optimal cleaning strategy outperforms all the other cleaning strategies showing its effectiveness. The optimal cleaning strategy developed is useful to solar photovoltaic plants owners whose plants are located in dusty or polluted areas. It first provides them with an understanding of non-uniform dust deposition. It also provides a way of reducing the effects of non-uniform dust deposition through optimized cleaning which is cost effective and that allows the photovoltaic array to continuously give the desired output. / Dissertation (MEng)--University of Pretoria, 2020. / Electrical, Electronic and Computer Engineering / MEng / Unrestricted