Accurate Line Resistance Estimation in a Multi-source Electrical Power System of the More Electric Aircraft: An Intelligent and Data-Driven Approach

Abstract

Knowledge of the line resistance is very important in the analysis of the electrical power systems (EPS) and their control operations. Some of the existing line resistance estimation methods require the utilization of many devices, involve the injections of disturbances to the systems and are computationally intensive. Hence, the processes take longer times to accomplish, have the possibility of making errors, could affect the power quality and inquire additional costs for the system. Surrogate modelling is an excellent alternative to ease the burden associated with complex computation, save cost and increase the reliability of the system. In this paper, an artificial neural network (ANN)-based surrogate model is proposed for the estimation of line resistance in the DC grid of the more electric aircraft (MEA) electrical power system (EPS). A neural network (NN) model is employed and trained based on a set of data obtained from multiple simulations to serve as a dedicated surrogate model of the detailed MEA EPS simulation model. The surrogate model is trained to establish the relationship between the output current of the converters to the corresponding line resistance within the design space with high accuracy. Thereafter, for every change in the line resistance between the parallel-connected converters and the DC bus in the MEA EPS, the output current of the converter can be provided as input to the surrogate model to predict the corresponding line resistance. The results obtained show that the surrogate model can accurately estimate the line resistance with an error of less than 1% provided the line resistance is within the design space used in training i