A Simulated annealing global maximum power point tracking method for photovoltaic systems experiencing non-uniform environmental conditions

摘要

Photovoltaic (PV) systems have immense potential due to the abundance ofudavailable solar energy and the capability of these systems to be implemented inuda distributed manner. This clean, renewable and sustainable energy source canudprovide a solution to concerns about the shortage of fossil fuels, global warming,udgreenhouse gas emissions and pollution in general. Residential PV systemsudenable consumers to take control of generating electricity to satisfy their ownudload requirements and potentially export any excess energy to the distributionudgrid. Investing in a PV system requires significant initial capital and PV cellsudhave a very limited efficiency. Residential customers who invest in a PV systemudexpect to be able to have the best return on their investment by utilising theudpower available in the sunlight to the greatest extent possible. The potentialudpower available from such systems can be dramatically reduced due to shadingudof the modules and ineffective control strategies to overcome the influence ofudshading. PV cells exhibit a non-linear Power-Voltage (P-V) characteristic leadingudto a unique point corresponding to optimal operation. This point is referred to asudthe Maximum Power Point (MPP), and varies depending on the environmentaludconditions. Typical conditions in a residential environment involve obstaclesudsuch as trees, houses and power poles which may cause shading across all orudpart of the PV system throughout the day. Shading from these obstacles leadsudto increased non-linearity in the P-V characteristic as multiple maxima can beudexhibited. Traditionally, Maximum Power Point Tracking (MPPT) techniquesudhave been developed to track a single maximum on the P-V characteristicudbased on simple techniques such as hill climbing. These techniques inherentlyudfail when multiple maxima are exhibited under Partial Shading Conditions (PSC).udThe work documented in this thesis consists of two main parts. In the first part,uda study of modelling PV cells and an extensive shading study for an eight-moduleudPV system is conducted. This analysis has led to the classification of partialudshading phenomena based on the time scale as either constant, static or transientudpartial shading, and exploring the effect that each aspect of partial shadingudhas on the relative location of the Global Maximum Power Point (GMPP).udThe second part comprehensively explores the concept of MPPT and presentsuda review of techniques proposed in the literature with consideration of theirudperformance under non-uniform environmental conditions. A Global MPPTud(GMPPT) method is proposed based on the global optimisation technique ofudSimulated Annealing (SA) and its performance is verified through simulationsudand experimental application.udThe key contributions of this thesis include proposing a shading classificationudbased on the time of influence of the shading and studying how this affects theudrelative location of the GMPP, development and optimisation of a SA basedudGMPPT method, and the merging of these results to develop a comprehensiveudand enhanced GMPPT strategy.udThe main concerns associated with modelling PV cells and modules are introducedudand a model of the BP380 PV module is developed based on theudcommonly used Single Diode Model (SDM) for PV modules. The SDM providesuda good balance between accuracy and simplicity and is shown to model theudexperimentally measured P-V and Current-Voltage (I-V) characteristics of theudBP380 modules with acceptable accuracy. A model is also developed andudexperimentally validated based on combining two series-connected modulesudmodelled using the SDM to explore PSC.udA PV system comprised of eight series-connected PV modules, modelled basedudon the BP380 PV modules, is developed to explore the influence of PSC. Audmethodology for calculating the position of the shadow tip and determiningudwhich cells are shaded by an object is proposed and used to perform five caseudstudies exploring the effects of constant, static and transient partial shading.udConstant partial shading is defined as a mismatch in the potential of the modulesudin a system based on factors such as manufacturing tolerance, cell degradationudand damage over time. This type of shading should remain roughly the sameudfor all time. Static partial shading is considered as shading that moves muchudslower than the movement of clouds across the sky and represents the shadingudthat occurs on the modules due to the presence of obstacles in the environment.udFinally, transient shading is the quickest shading phenomena and is representedudby the changing irradiance due to the movement of clouds across the sky.udAn extensive review of maximum power extraction strategies is presented. Eachudtechnique is assessed against key criteria identified as being essential for auduniversally applicable GMPPT strategy. In particular, the methods are assessedudon whether they can locate a global maximum reliably, the method complexityudand its ease of application to other PV systems. The analysis suggests thatuda global maximum power extraction strategy with moderate complexity andudlimited dependence on system specific parameters is needed.udThe proposed SA based GMPPT method is introduced in the form of simpleudstudies showing the effectiveness of the method in converging to a GMPP basedudon a two module PV system, eight module PV system, basic grid connectedudsystem and through experimental verification on the two series-connected BP380udPV modules. The key parameters of the SA method are explored in more detailudto assess their influence on the effectiveness of the proposed method.udThe main advantages of the proposed methodology for GMPPT is that it is notudsignificantly more complex than the common perturb and observe (P&O) MPPTudtechnique, yet has far superior performance in converging to the GMPP. Additionally,udwhen compared to the Particle Swarm Optimisation (PSO) methodudwhich is commonly used for GMPPT, the SA based method has less complexityudyet similar performance in converging to the GMPP. By incorporating understandingudof the relative location of the GMPP under different PSC, the techniqueudis enhanced to improve accuracy and convergence time. In residential environmentsudone of the key factors is minimising cost and ensuring maximum effciency.udFor this reason, a low cost and low complexity MPPT method that can achieveudGMPP identification is essential, to enable systems implemented in residentialudenvironments to be utilised to the greatest extent possible.