Research On Identification And Optimization Method To Steam Turbine Flow Characteristics Based On Unit History Data Mining

With the rapid development of China power industry, high-parameter and large-capacity units have been put into operation, there have been some new situations as well as problems in the process of safe electric power production. Especially for an unit after running for long time, flow passage retrofit or DEH reconstruction, the flow characteristics has inevitably deviated from the original design, causing unit's load fluctuations when valve control switching, insufficient or excessive ability of primary frequency modulation and coordinated controlling, etc.In order to ensure the safety, high-quality and economic operation of the power prid, each branch has highlighted unit's control quality, and required that steam turbines' flow characteristics parameters of speed control system in plant, should be well matched with Primary Frequency Regulating Function and Automatic Generation Control Scheduling Mode. Meanwhile, the characteristic parameters is the numerical representation of unit flow characteristics. Thus accurately acquiring the flow characteristics is an important premise to realize precise control for steam turbine.In this paper, the significance of control valves' flow characteristics identification and optimization is briefly analyzed. Further more, the data filtering method is utilized to fetch the unit stable data, then by means of reasonable hypothesis, scientific deduction and analytical investigation, corresponding flow characteristics identification methods have been put forward, within its' applicability analysis and limitation conclusion. Identification of multiple instances show that different identificatio n methods for different objects could successfully identify unit actual flow characteristics curve, which laid s a solid foundation to flow characteristics optimization.Based on the induction of ten classical data mining algorithms, rational improvements to K-Medoids algorithm have been made. Taking it as a data base, the existing optimization methods for flow characteristics have been summarized and analyzed, a new optimization method has been worked out. The applications of the method show a good performance, which not only fit the project reality but also effectively solve the emerged issues induced by unreasonable linearity.Applying the proposed identification a nd optimization methods herein, real flow characteristics identification and optimization research have been carried out on an 1000 MW unit, and the error or accuracy of identification methods have been analyzed, as well as the reliability of improved mining algorithms and optimization methods, which verifys the validity and correctness of the identification and optimization methods.Embarking from the practical application, a set of application software have been successfully developed, which idenfys and optimizs the flow characteristics based on historical data, which realizes the automatic for flow characteristics' identification and optimization.