Prediction of the Natural Gas Calorific Value by Using Artificial Neural Network

The current study introduces the use of a multilayer perceptron neural network (MLP-NN) for prediction of the calorific value (CV) of natural gas using the process parameters pressure, temperature and the speed of sound (SOS) determined by an ultrasonic gas flow meter (USM). The MLPNN were trained and tested with data from collected measurements for five consecutive months. As a result, the highest quality MLP 3-15-1 neural network for prediction of the calorific value was experimentally selected.