This work presents the use of an ant colony optimization (ACO) based neuro-evolution algorithm to optimize the structure of a long short-term memory (LSTM) recurrent neural network (RNN) for the prediction of aircraft turbine engine vibrations. It expands upon previous work using three different LSTM architectures, with the new evolved LSTM cells showing an improvement of 1.35%, reducing prediction error from 5.51% to 4.17% when predicting excessive engine vibrations 10 seconds in the future. These results were gained using MPI on a high performance computing cluster, evolving 1000 different LSTM cell structures using 168 cores over 4 days.
Paper
Full text
Optimizing LSTM RNNs using ACO to predict turbine engine vibration
Semantic Scholar · Computer Science · 2017
Abstract
This work presents the use of an ant colony optimization (ACO) based neuro-evolution algorithm to optimize the structure of a long short-term memory (LSTM) recurrent neural network (RNN) for the prediction of aircraft turbine engine vibrations. It expands upon previous work using three different LSTM architectures, with the new evolved LSTM cells showing an improvement of 1.35%, reducing prediction error from 5.51% to 4.17% when predicting excessive engine vibrations 10 seconds in the future. These results were gained using MPI on a high performance computing cluster, evolving 1000 different LSTM cell structures using 168 cores over 4 days.