Characteristic Analysis Of AGC And Load Response Curve In Thermal Power Units

With the gradual deepening of the market economy, as the electricity treated as a commodity, the price should be determined by its quality. The quality of electric generated in power plants could be assessed in both the generation tasks dispatching from the power grid and the power plants’performance. It’s not fair for the thermal units only considered the plants’performance while ignoring the tasks’difficulty.In order to propose a quantitative standard, this article consideres the difficulty level of power generation tasks, while ignoring the voltage quality, as the "Two Rules" has provided the completion of the tasks. Since the automatic generation control (AGC) scheduling command signals could stand for the generation tasks, this paper focuses on researching the characteristics of the AGC signals.This paper analyzes the signals in two aspects such as the general data analysis and the nonlinear complexity. Firstly it analysis deeply in the signals in one1000MW thermal unit’s a year’s SIS platform data, selected the peak, mean, standard deviation, the peak between adjacent difference four common characteristics to show the time characteristics of the command signals.Then the Kolmogorov complexity and sample entropy two nonlinear dynamic parameters have been chosen to analysize the signals as for the signals’fluctuation. This method is used in multi-scales such as months, days and hours and the results illustrate the selected characteristics can quantitatively analysize the AGC signals in a certain extent, meanwhile the data analysis shows that sample entropy has better recognition ability than the Kolmogorov complexity in signal complexity analysis.For overcoming the "over-coarsening" insufficiency that traditional Kolmogorov complexity method may produce., this paper proposes wavelet K complexity (WKC), and it is combined with wavelet transform. By analyzing the characteristics of the load response curve, the results show that improved wavelet K complexity algorithm has better effectivity than two others. Suggest to pick the Com by WKC as the basis of the grading divisiory in generation tasks, then quantitative research in the quality of electric is put forward.