Interactive Regression Analysis Of Thermal Power Control Parameters Supported By Visualization

As the main force of power production in China,thermal power generation is obviously important to power supply system of China.Operational safety in power production has always been the focus of attention of power enterprises.Therefore,how to improve control capacity of thermal power units and stable operation by optimizing the operation control of thermal power units is an important issues for enterprises.Taking the Digital Electric-Hydraulic Control System(DEH)as an example,after long-term operation or maintenance of the steam turbine,the actual flow characteristic curve of the valve of the DEH system will deviate from the original theoretical curve and finally affect the safe and stable operation of the unit.Therefore,it is necessary to test and calibrate the valve flow characteristic curve of the unit in order to ensure that the indicators of the unit tend to be optimal.Using data fitting method to build turbine flow characteristic model is the basis of regulating valve flow analysis and correction after getting actual flow characteristic data.However,due to the randomness of the historical data generated by the unit,the large number of parameters and the complex mechanism of unit,it is difficult for the traditional modeling methods and tools to effectively model and analyze it.In this thesis,regression analysis and visualization techniques are combined to design an interactive regression analysis tool.The proposed tool provides interactive visualization support in the phases of data characterization analysis,exploring parameter correlations,training models,hyperparameter adjustment,model performance evaluation and model iterative optimization for the analytical characteristics of thermal power control data analysis process.The visual analysis system supports switching between different models for modeling time-series data,observing data distribution characteristics through probabilistic quality analysis,designing multiple visualization techniques and interactive linkage views to support multi-level model screening process,supporting parameter analysis of established models,and supporting users to evaluate model quality from different aspects.Finally,a case study is illustrated based on the real historical data of thermal power units to verify the effectiveness and practicality of the system.The research work in this thesis provides an efficient and powerful support for the fitting,analysis and exploration of steam turbine control valve flow characteristic curves in thermal power control units,and lays a foundation for the correction of steam turbine flow characteristic curves,which can be extended to the regression analysis process of other control parameters.