Modeling and prediction of boiler drum in a thermal power plant

Neural Networks-based Process Model and its Integration with Conventional Drum Level PID Control in a Steam Boiler Plant

“…Several methods for controlling drum level in power plant drum boilers are proposed in the literature ranging from those employing single-input single-output (SISO) PID and fuzzy adaptive techniques [12], to those based on neural networks (NN) [20], and model predictive control (MPC) [19].…”

“…The authors in [20] also carried out similar work and went on to propose a methodology based on what is called heuristic technique for developing system identification models through application of NARX NN to predict the drum level in a coal fired power plant also in India. The drawbacks in these works were the resolution of the data used for training the neural networks, recorded on hourly basis, which would severely affect the accuracy of the drum level predictions and in turn would also affect the precision in the control of the level.…”

“…Besides [20] and [21] where an attempt was made to explore the possible application of neural networks to drum level prediction and control in coal fired power plants, there has not been much reference in the literature to the practical integration of data-driven modeling techniques with traditional PID control of drum level on existing thermal power plant steam boilers. This paper attempts to fill this gap by proposing opportunities for the selection of machine learning models in controlling the water level in a power plant drum boiler and their integration with the conventional PID controller in cascade control configuration.…”