Research On Fault Diagnosis And Reconfiguration Strategy Of PV Arrays In A PV Power Generation System

In recent years,solar energy has attracted more and more attention due to its inexhaustible,geographical free and non-polluting features.In particular,solar photovoltaic power generation has been regarded as one of the most popular means of green power generation.Since photovoltaic arrays in a photovoltaic power generation system are generally installed in the outdoor regions of high altitude and large temperature differences,the application of the photovoltaic arrays is constrained by conversion efficiency,various faults as well as photovoltaic panel lifetime.This thesis aims at enhancing the conversion efficiency of photovoltaic arrays,prolonging the service time of photovoltaic panels,and reducing maintenance costs under complex outdoor environments.For this purpose,fault diagnosis and reconfiguration strategy are investigated for photovoltaic arrays as detailed below.(1)Maximum power point tracking of photovoltaic arrays.Firstly,in order to accurately establish photovoltaic array models under complicated outdoor environments,a photovoltaic array consists of photovoltaic strings connected in parallel and each string has multiple photovoltaic modules connected in series,where each module comprises photovoltaic cells and an anti-parallel bypass diode.On this basis,a modified flower pollination algorithm is developed for maximum power point tracking implementation.In this method,the conversion of dual-mode optimization is controlled by both switch probability and populati on fitness values,which can overcome the disadvantages of the flower pollination algorithm capturing easily local extremums and converging slowly to the corresponding extremums during maximum power point tracking.The performance of this proposed method for photovoltaic arrays is verified by comparing it with the perturb & observe method and flower pollination algorithm under extensive scenarios of different irradiation levels,different temperature conditions,as well as complicated partial shadings.(2)Fault detection of photovoltaic arrays.The latent faults and fault types of photovoltaic arrays based on maximum power point tracking operation need to be accurately identified for the conversion efficiency and operational security of a photovoltaic system.This thesis conducts a fault detection technique for a PV array based on voltage and current index evaluations.Firstly,the output characteristics of the photovoltaic array under common faults are analyzed to extract the fault characteristic quantities,such as the output voltage and current of the photovoltaic array at the maximum power point.Then the corresponding fault detection thresholds are determined according to the formulated voltage and current indexes by fault characteristic quantities.By comparing these determined detecting thresholds with the output voltage and current indexes of the photovoltaic array evaluated in real-time,the potential faults and fault types can be differentiated.The feasibility of the proposed fault detection technology is validated on a 3 ?3 photovoltaic array for various fault scenarios,such as simple string open circuit,single or multiple modules short circuit,simple or multiple degradation modules,and partial shading or variable shading faults.(3)Faulty module locating of photovoltaic arrays.After identifying the latent faults and fault types of a photovoltaic array,the positions of the faulty modules need to be located further.For this purpose,a faulty module locating approach of a photovoltaic array based on an improved voltage sensor deployment is introduced.The array is divided into sub-arrays according to the parity of the number of photovoltaic strings and some voltage sensors are deployed between the photovoltaic strings in each sub-array.Meanwhile,universal faulty module locating rules are formulated under the possible faults,such as open circuit,short circuit,degradation,or partial shading faults.From the readings of the placed v oltage sensors in real-time and the formulated fault locating rules,the location of each faulty module can be accurately identified.The performance of the proposed fault-locating approach is first verified by comparing it with existing fault locating methods for a small-size photovoltaic array.Then a large-scale photovoltaic array of a real 1.2 k Wp grid-connected photovoltaic plant at an experimental base of the China electric power research institute is investigated to validate further the proposed strategy.(4)Faulty area locating of photovoltaic arrays.In practical photovoltaic plants,the position of each faulty module in small-scale photovoltaic arrays needs to be accurately determined.However,the same strategy if applied to large-scale photovoltaic arrays would imply high investment costs.To reduce further the number of sensors employed and save fault locating investment,it is sufficient to identify the areas where faults occur.Based on this observation,a module-block fault locating method is proposed for a large-size photovoltaic array in this thesis,where a module block,as the minimum fault-locating area,consists of one or multiple modules connected in series.In this method,the similar sensors placement scheme above is recalled to obtain the terminal voltages of the module blocks between photovoltaic strings,and the corresponding faulty module-block locating criteria of the photovoltaic array under common faults are formulated.The proposed faulty module-block locating method is tested in a 5 ?12 photovoltaic array at a 3.6 k Wp grid-connected photovoltaic plant providing power supply for greenhouses in western China and a 200 ?100 photovoltaic array of a 1.2 MWp system.(5)Reconfiguration of photovoltaic arrays.Temporary shading faults with slow changes is one of the most frequent and severe faults under complicated outdoor surroundings.To alleviate this issue,a block-level reconfiguration technology for a photovoltaic array is described under complicated shading conditions,where the divided sub-arrays at the block-level are connected to form Total Cross Tied topologies by placing electrical switches between strings in each sub-array.On this basis,a one-time physical reconfiguration strategy based on an intelligent column index is used on the divided sub-arrays.When partial shading emerges in the photovoltaic array,the deployed switches are triggered and the array is connected in Series Parallel-Total Cross Tied configuration while the received shading is dispersed over the whole panels to reduce the influence of shading on its output power.Comparing with the output powers of existing configurations for a 9 ?27 PV array at a real-world 14.5 k Wp grid-connected photovoltaic plant,the effectiveness of the proposed reconfiguration strategy is demonstrated under dynamic regular and irregular shadings.