ULTRA-FAST APPROXIMATION OF LOAD FLOW VARIABLES USING NEURAL NETWORKS The new Artificial Intelligence module in PowerFactory offers an innovative method of training neural networks to estimate the results of a large number of different load flow scenarios, so speeding up load flow calculations considerably. Dataset generation, neural network training and use of the neural network are all realised within PowerFactory Generation of training dataset using PowerFactory's inbuilt Probabilistic Analysis Use of distribution and parameter characteristics for the dataset generation, allowing historic data and other information about the electrical network to be taken into account in a comprehensive dataset Explicit selection of training output variables, to focus on specific load-flow quantities Application of pre-trained neural networks in Quasi-Dynamic Simulations, for ultra-fast estimation of network variables Training dataset and neural network are stored externally and so can easily be shared with other users For more information visit www.digsilent.de/pf-ai Serving more than 160 countries. POWER SYSTEM SOLUTIONS MADE IN GERMANYhttp://www.digsilent.de/pf-ai