Computational approach to enhance performance of photovoltaic system inverters interfaced to utility grids

Abstract

Switch-model power electronic inverters are heavily deployed as the main technology in enabling flow of variable DC power into the AC grid. Various distributed energy system (DES) architectures have been designed depending on many attributes including size and application of the installed system. The reliability of the power electronic interfaces (PEI), i.e. inverters is critical in all these architectures. Recent studies demonstrate enhanced operation of a PEI can be reached by optimum adjustment of its controller parameters. While conventional tuning methods are mostly based on trial and error, their optimum performance can be achieved for primarily first-order systems. For systems with increasing number of PEIs or more complexity, application of these tuning methods becomes challenging and expensive. Thus, considering the operational performance of DES, a controller self-tuning methodology has been presented for PEIs with particle swarm optimisation capability. The optimal parameters for both kW-scale and multi-megawatt PV systems' inverters are determined via a time-domain performance objective function. Typical PV system performances are presented for step changes and dynamic weather conditions. Effectiveness of the controller self-tuning methodology has been demonstrated via the reduction of transient energy, when the system is subjected to dynamic changes and disturbances.