Research And Implementation Of Visual Time-series Data Analysis For Thermal Power Control System Identification

Nowadays,thermal power generation still plays an important role in China's power production.Power generation enterprises pay more and more attention to the issue of energy saving,how to improve the accuracy of unit control and energy efficiency through optimal control of power generation is an urgent problem for power enterprises.The analysis and modeling of thermal power control historical data can effectively help users understand the operation rules of units and achieve better control of unit operation.Distributed Control System(DCS)equipped with thermal power unit records the historical data of unit operation,which provides data support for the rule analysis and identification of unit control system.Due to the high complexity of control data,the traditional system identification is very complex,and it is even difficult to obtain effective results.In this thesis,visual analysis technology is introduced into system identification and integrated with automatic modeling algorithms,a visual analysis system for system identification is designed which called im DCS,it provides visualization support at various stages include time series feature analysis,model establishment,selection and evaluation,and model iteration optimization.The proposed system supports the iteration prediction of model time series parameters by constructing interactive enhanced multi-resolution curves,and supports multi-level model filtering by designing various visualization techniques and interactive linkage views,the high dimensional and multivariate model structure is displayed in grouped and stacked charts,and uesrs can evaluate model performance from different perspectives through precision evaluation views.Domain experts are invited to do case study based on the real world control data and the requirements of a power plant,and verified the effectiveness and usability of the system.This thesis provides a new and potential support for the analysis,prediction andoptimization of complex industrial control processes.