Data-Driven Sliding Pressure Optimization for Combined Heat and Power Units

In the field of thermal power generation, steam-extraction combined heat and power (CHP) units face challenges in adapting to variations in heat supply demand and environmental parameters, as their conventional sliding pressure curves fail to meet operational requirements under dynamic conditions. To enhance the economic efficiency and stability of such units, timely adjustments to the sliding pressure curve are critical. This study proposes a multi-input relational model that correlates main steam pressure with power load, heating requirement, and back pressure to accommodate diverse operating conditions. Initially, historical operational data is utilized to establish a sliding pressure optimization model, integrating the Random Sample Consensus (RANSAC) fitting algorithm, to minimize heat consumption rate. Furthermore, a novel equivalent heat load calculation method is introduced, and a computational model for low-pressure cylinder exhaust enthalpy is developed using the Least Squares Support Vector Machine (LSSVM) algorithm, enabling real-time equivalent heat load estimation to support sliding pressure optimization. The proposed model was implemented on a 660 MW thermal power unit, with optimized results integrated into a closed-loop control system. A comparative analysis of operational data before and after optimization reveals that the sliding pressure optimization model significantly reduces energy consumption and enables adaptive adjustment of main steam pressure in response to multi-parameter variations, including load, back pressure, and heating energy demand. This approach provides a practical framework for enhancing the flexibility and efficiency of CHP systems under variable operating conditions.