A data-driven method for main steam temperature control in waste-to-energy plants: its theory and a 90-day continuous demonstration at a real-world plant

Abstract This paper presents a data-driven control method for waste-to-energy (WtE) plants to address the challenges of maintaining the main steam temperature, a crucial factor for energy efficiency. The proposed method utilizes a learned data model to track the optimal transition route to achieve the control setpoint. Through a 90-day demonstration operation at an active WtE plant, the method was validated in terms of interpretability/explainability, tracking performance, versatility in distinct operation modes, long-term operability and stability, and the working mechanism of the control logic. The results demonstrate that the proposed data-driven control method outperforms the existing PID control method, providing superior control performance and adaptability to various operational conditions. The capability of the method to acquire improved control laws from existing operational data showcases its potential for enhanced control performance. Further testing of the disabled components, which were theoretically presented but not utilized during the demonstration operation, is expected to enhance the effective addressing of the non-Markovian nature of the process. The findings highlight the significance of data-driven control methods in achieving safe and efficient operations in WtE plants.