Research On Intelligent AGC Control Technology For Thermal Power Plant Based On Kdd And Big Data

The electric power industry is an important basic industry of our country.Due to the lack of adjustment of power system regulation and the rigid operation mode of power grid operation,the ultra-supercritical power generation unit is difficult to bring into full play the advantages of energy saving and efficiency.Automatic Generation Control(AGC)is an important part of dispatching automation for generating units.It is the most important task to focus on improving AGC's regulation and operation efficiency.Large data is a hot spot in various fields of research in recent years.Knowledge discovery is a high summary of the process of mining large data value.Therefore,combining the large operation data of ultra supercritical unit operation,it is important to study the intelligent AGC control technology based on knowledge and big data to improve the flexibility and adaptability of AGC control.In this paper,the AGC control system of ultra supercritical unit is taken as the research object.Based on knowledge and large data technology,the problems of intelligent identification,intelligent control of AGC system and AGC control performance analysis under CPS evaluation standard are studied.The research content includes the following aspects:First,the knowledge discovery processing model is introduced,and the large data of electric power infrastructure is analyzed in detail.The application prospect and challenge of the knowledge discovery content such as data mining,association analysis and artificial intelligence algorithm are analyzed in the light of the characteristics of the power industry.Secondly,based on the actual production of power plant,the AGC model structure of unit is set up.According to actual operation historical data,this paper proposes an approach nearest neighbor rule based data to reduce the large data capacity,using Newton iterative method,artificial neural network and Bias neural network three typical identification method for system identification system of oxygen content control of boiler combustion,the error of three models shows that the neural network model of Bias have outstanding advantages in accuracy,rapidity and feasibility.Then the AGC system identification based on knowledge and large data is studied by this method,and the identification results show that the intelligent identification method is also generalization.Furthermore,according to the load control mode and system characteristics of unit AGC system,and combined with predictive knowledge and neural network in knowledge discovery,the design of intelligent predictive controller is carried out.The identified Bias neural network AGC model is used as the prediction part of predictive control,and neural network algorithm is used for feedback correction and rolling optimization.The simulation results show that under the sliding pressure and constant pressure operation mode,the load tracking,the intermediate point temperature and the main steam pressure control effect of the intelligent AGC predictive controller meet the control quality requirements.Finally,the power deviation control effect of ACE and tie line are analyzed based on SCPS standard and CPS standard.According to the SCPS of single unit,the single station unit is independently analyzed.Based on CPS standard,TBC-TBC mode is adopted to simulate the interconnection of two generating units and then change the load instructions of two units respectively.The basic principle that the load disturbance can be handled in the area under the TBC-TBC mode and the CPS standard of the control strategy is verified by this method.