Use this URL to cite or link to this record in EThOS: http://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.517749
Title: Improving the efficiency of photovoltaic power plants with soft-computing model-based controllers
Author: Varnham, Abdulhadi Adrian
Awarding Body: University of Portsmouth
Current Institution: University of Portsmouth
Date of Award: 2005
Availability of Full Text:
Access through EThOS:
Abstract:
This thesis investigates ways of increasing the energy efficiency of photovoltaic power (pv) generating plants. It does so by improving models of PV plants connected to the electricity grid via space-vector-modulated three-phase inverters. Synergies of softcomputing techniques are applied to modelling the current-voltage characteristics of solar cells and to model-based control of the inverters. A novel extension of a radial-basis-function-network model is reported. The model is unique in that it incorporates a new grid-interpolation data pre-processor, which for the first time has allowed the network to be trained with real solar-cell data. Furthermore, the model provides greater accuracy compared to the industry standard model. Coordinate translation of solar-cell characteristics has been incorporated into a neurofuzzy model of solar cells. Significantly, it enables neural-network models of plants to be trained with far fewer data and with greater resilience to model imperfections than has hitherto been possible. Important applications include the modelling of new plants. Soft-computing control strategies were developed that removed the need for expert knowledge in parameter tuning: (i) a genetic algorithm for optimising the gains of the conventional PI controller; (ii) a genetic algorithm for optimising the parameters of a fuzzy logic controller; and (iii) a neural network for optimising the parameters of a fuzzy logic (ANFIs) controller. The ANFIS controller provided the best transient and steady-state behaviour. The models and control strategies were combined together to form model-based controllers that were more accurate and resilient than existing solutions. Increased power production was demonstrated of 1.5% for a plant well characterised by the conventional model, and 8.6% for a plant poorly characterised by the conventional model, with no requirement for expert knowledge. The results are believed to be important for application in developing countries because of the improved efficiencies and the ability to design and install systems without needing expensive resources.
Supervisor: Not available Sponsor: Not available
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral
EThOS ID: uk.bl.ethos.517749  DOI: Not available
Share: