Mert NakipCuneyt GuzelisGuzelis, CuneytNakip, Mert2025-10-06201910.23919/eleco47770.2019.8990400http://dx.doi.org/10.23919/eleco47770.2019.8990400https://gcris.yasar.edu.tr/handle/123456789/6349https://doi.org/10.23919/eleco47770.2019.8990400In this paper we propose a Trend Predictive Neural Network (TPNN) model which uses the sensor data and the trend of that data in order to classify the fire situation. We implemented TPNN for data of multi-sensor fire detector with 6 sensors to detect 7 inputs. We test the performance of the TPNN model by using the multi-sensor dataset which is collected within this study. Our results show that the TPNN model is a fast and accurate model whose execution time is 0.0132 seconds. Furthermore TPNN decreases both the false positive and false negative alarm rates to half of the results of the multi-layer perceptron model.Englishinfo:eu-repo/semantics/closedAccessConference Object