Improving the performance of Off Gas booster systems using artificial neural networks

In this article we have tried to use the capabilities of neural networks to improve the performance of Off Gas booster systems. Pressure boosting systems are used to collect and return the exhaust gases from the stabilizer unit. In the process of boosting gas pressure by Off Gas compressors, pressure and temperature parameters are very important and their measured value is crucial to controlling the process. We use neural networks in three parts: control, quantification and pneumatic diagnostics. In the control part, the inlet gas pressure is evaluated using a neural network-based PID controller and compared with conventional methods for determining proportional, integral, and derivative coefficients. In estimating pressure and temperature quantities, we use neural networks as software sensors and their performance is compared with hardware sensors. Finally, the ability of neural networks to detect pneumatic defects in the control valves used in the inlet and outlet of the Off Gas compressor is investigated.