Fault Degradation State Prediction under Closed-loop Control for 1000MW Ultra Supercritical Unit

Due to the complicated structure and the harsh operating conditions of 1000MW ultra supercritical unit, it inevitably suffers from fault degradation. Once a fault seriously degrades, it may lead to unplanned shutdown, causing serious economic losses and even casualties. In order to ensure the safe and reliable operation of a 1000MW ultra supercritical unit, it is necessary to predict the fault degradation state to determine the reasonable maintenance time, thus eliminating safety hazards and reducing the risk of unplanned shutdown of the unit. Besides, closed-loop control system is commonly used in 1000MW ultra supercritical units to regulate process disturbance or track set points. Therefore, process dynamics caused by closed-loop control system, including serial correlation and varying speed of process variation, should be considered while forecasting fault degradation state. In this work, a combined Canonical Variate Analysis and Slow Feature Analysis (CVA-SFA) is applied to extract features that can fully reflect process dynamics under closed-loop control. Then, Continuous Hidden Markov Models (CHMMs) are built to predict fault degradation state by these features. Finally, the proposed method is applied in a real industrial process of 1000MW ultra supercritical unit to verify its feasibility and efficacy.