Identification and Application of the Flow Characteristics of Steam Turbine Valve Based on LSSVM

Abstract

The accurate identification of the valve flow characteristic of steam turbine is of great importance for the turbine control. In this paper, a method based on least-squares support vector machine (LSSVM) is proposed to identify the valve flow characteristics of steam turbine. The historical operation data of the unit was screened to obtain the data under stable condition. Then LSSVM was applied to identify the relationship between the main parameter vector, which was composed of the integrated valve position command, main steam pressure, regulating stage pressure, etc., and the calculated actual steam flow of turbine. Finally, on the basis of the established LSSVM model, the working condition of steam turbine valve flow characteristic testing was simulated by changing the values of main parameter vector, so as to identify the steam turbine valve flow characteristics. This method can be implemented without doing any steam turbine valve flow characteristic test, which greatly reduces the workload and avoids the adverse effect of the test method on the operation security and stability of the unit.

Keywords

• least squares support vector machine
• data screening
• flow characteristics
• identification