The Governor Predictive Controlled Based on LSTM for Optimizing Cofiring Power Generator Operation

Addien Wahyu Wiranata (Institut Teknologi Sepuluh Nopember)
Dimas Anton Asfani (Institut Teknologi Sepuluh Nopember)
Daniar Fahmi (Institut Teknologi Sepuluh Nopember)

JAREE (Journal on Advanced Research in Electrical Engineering)
Vol. 1 No. 10 (2026): January

The renewable energy with cofiring technology has a significant impact on the use of Biomass. The use of biomass with different qualities greatly affects the performance of a plant. Deep Learning Time Series Forecasting is designed for predicting two control parameters cofiring powerplant operation consist of governor control and output generator. Long Short-Term Memory (LSTM) combined with Multilayer Perceptron, Convolutional, and Adaptive Moment Estimation (ADAM) optimizer algorithms are utilized to optimize the process governor control and predict generating power output. Correlation analysis is used to determine the input variables and resulting input parameters of governor control prediction consist of Temperature Steam, Pressure Steam, Output Generator, Coal Flow, Flow Steam. Moreover, the input variable for prediction generation power output are steam flow, steam temperature, coal flow, and steam pressure. The combination of Deep Learning Forecasting is successfully to predict both operation parameter percentage errors of 5.33%.