An Equivalent Modeling Method for Integrated Hydro-Photovoltaic-Wind Systems Based on Particle Swarm Optimization and Neural Networks

The fault transient characteristics of power grids have undergone a fundamental transformation as a consequence of the advent of hybrid scale access to power grids by new energy sources. Nevertheless, the existing stand-alone equivalent model is unable to accurately characterise the fault transient characteristics of the hybrid light-water-wind electric field/cluster. Furthermore, the simulation computation of the stand-alone equivalent model is considerable in scale and the simulation efficiency is low. This paper proposes a particle swarm neural network-based clustered equivalent modelling method for a light-water-wind hybrid electric field. Firstly, a transient model of the light-water-wind hybrid electric field is established, and the characteristic influence factors of faults are constructed through theoretical analysis. These include the distance between the light-water-wind hybrid electric field and the point of common coupling (PCC), the input of current limiting measures on the DC side, the light, the water speed, the wind speed, and the reactive power at the outlet. Subsequently, the Euclidean distance is calculated for the set of characteristic influencing factors, and the initial centre of clustering of wind turbines is extracted based on the improved maximum-minimum distance method. Ultimately, the clustering equivalent modelling is accomplished through the utilisation of the aforementioned feature influencing factors and the classification initial centre as the training set. The simulation results show that compared with the single-machine equivalent model, The simulation error has been reduced from the order of magnitude of 10-2 p.u. to the order of magnitude of 10-4 p.u., while the simulation efficiency of this method is has been reduced from the order of magnitude of 10-2 s to the order of magnitude of 10-4 s.