PINN vs LSTM: A Comparative Study for Steam Temperature Control in Heat Recovery Steam Generators

TL;DR

The paper tackles robust HRSG steam-temperature control under valve leakage by comparing a purely data-driven LSTM controller with a physics-informed PINN controller that adaptively tunes the PI-plus-feedforward gains. While LSTM improves over fixed-gain PI under normal operation, it struggles with unseen faults; the PINN, incorporating thermodynamic laws in its online learning, delivers superior fault tolerance and performance. The key finding is that physics-informed learning yields a 54% reduction in integral absolute error relative to the LSTM under leakage faults, highlighting the importance of embedding physical constraints for safety-critical industrial control. Practically, this work supports the adoption of PINN-based adaptive control for more reliable, fault-tolerant operation of HRSGs in flexible, real-world power plants.

Abstract

This paper introduces a direct comparative study of Physics-Informed Neural Networks (PINNs) and Long Short-Term Memory (LSTM) networks for adaptive steam temperature control in Heat Recovery Steam Generators (HRSGs), particularly under valve leakage faults. Maintaining precise steam temperature in HRSGs is critical for efficiency and safety, yet traditional control strategies struggle with nonlinear, fault-induced dynamics. Both architectures are designed to adaptively tune the gains of a PI-plus-feedforward control law in real-time. The LSTM controller, a purely data-driven approach, was trained offline on historical operational data, while the PINN controller integrates fundamental thermodynamic laws directly into its online learning process through a physics-based loss function. Their performance was evaluated using a model validated with data from a combined cycle power plant, under normal load changes and a challenging valve leakage fault scenario. Results demonstrate that while the LSTM controller offers significant improvement over conventional methods, its performance degrades under the unseen fault. The PINN controller consistently delivered superior robustness and performance, achieving a 54\% reduction in integral absolute error compared to the LSTM under fault conditions. This study concludes that embedding physical knowledge into data-driven control is essential for developing reliable, fault-tolerant autonomous control systems in complex industrial applications.

Figures (10)

• Figure 1: Block diagram of a Combined Cycle Power Plant, showing the gas turbine, HRSG, and steam turbine. The HRSG serves as the critical interface between the gas and steam cycles.
• Figure 2: The desuperheater (DSH) unit at the Pareh-Sar power plant, where valve leakage has been identified as a critical fault source affecting steam temperature regulation.
• Figure 3: Schematic of the desuperheater process. Spray water is injected into the superheated steam flow, and evaporation/mixing dynamics regulate outlet steam temperature.
• Figure 4: Visual examination of the damaged attemperator: (a) displacement of the spray nozzle tip, (b) surface cracks near the weld joint, (c) multiple internal cracks along the nozzle inner surface, and (d) fractured surface exhibiting an irregular, fibrous texture.mukhopadhyay_failure_2011
• Figure 5: Control loop structure illustrating the LSTM-based gain tuning strategy for HRSG superheater temperature regulation, with the leakage fault modeled in the spray-water valve actuation.