Abstract:When the discharge characteristics of turbine valves go to worsen, the power oscillation will be induced, and then the low-frequency oscillation in power grid may be induced . In this paper, a 660MW supercritical unit where power oscillations occurred is taken as an example. Through data mining and Routh-Criterion, the valves discharge characteristics are got by mining the historical data, a comprehensive model of active power regulation system is set up, and the influence of turbine valves discharge characteristics on the system stability is qualitatively and quantitatively researched. Then a critical value (k) of 3.3438 of curvature of valves discharge characteristics curve is got. The system will be unstable over this value. Through data mining analysis, turbine valves discharge characteristics of this unit are found to have worsened among 92.7%～96.6% of total valve position instruction, where the maximum k value is 7.771, far greater than the critical value, which is the reason why several power oscillations occured in this range of total valve position instruction. That can be used for reference in the analysis on similar power oscillation events in the future.