System Identification Methodology of a Gas Turbine Based on Artificial Recurrent Neural Networks

Abstract

The application of identification techniques using artificial intelligence to the gas turbine (GT), whose nonlinear dynamic behavior is difficult to describe through differential equations and the laws of physics, has begun to gain importance for a little more than a decade. NARX (Nonlinear autoregressive network with exogenous inputs) is one of the models used to identify GT because it provides good results. However, existing studies need to show a systematic method to generate robust NARX models that can identify a GT with satisfactory accuracy. In this sense, a systematic method is proposed to design NARX models for identifying a GT, which consists of nine precise steps that go from identifying GT variables to obtaining the optimized NARX model. To validate the method, it was applied to a case study of a 215 MW SIEMENS TG, model SGT6-5000F, using a set of 2305 real-time series data records, obtaining a NARX model with an MSE of 1.945 × 10−5, RMSE of 0.4411% and a MAPE of 0.0643.

Monitoring and Diagnosing Faults in Three-Phase Induction Motors Using Artificial Intelligence Techniques

“…In this way, recurrence can be defined as a type of short-term memory, which allows the network process of retrieving information from the recent past. The NARX model is a form of recurrent Artificial Neural Network with feedback mechanisms, especially suitable for non-linear modeling of systems, focusing on time series, where past data is used to predict future values [21]. NARX networks are an efficient alternative to recurrent networks, offering similar computational efficiency.…”

Assessing Forest Quality through Forest Growth Potential, an Index Based on Improved CatBoost Machine Learning

“…Finally, the integrated machine-learning algorithm CatBoost is adopted, with an R 2 coefficient of 0.89. The evaluation performance indicators used are R-squared (R 2 ), the mean square error (MSE), the root mean square error (RMSE), the mean absolute error (MAE), and the symmetric mean absolute percentage error (SMAPE) [40][41][42][43][44].…”