THE 3 rd INTERNATIONAL SCIENTIFIC CONFERENCES OF STUDENTS AND YOUNG RESEARCHERS dedicated to the 99
th
anniversary of the National Leader of Azerbaijan Heydar Aliyev
107
DETECTION OF EMERGENCY CASES AT POWER PLANT EQUIPMENT BASED ON MACHINE LEARNING Bakshiyev Mardan Baku Higher Oil School Baku, Azerbaijan Mardan.bakshiyev.std@bhos.edu.az Supervisor: Ph.D Associate Professor Leyla Muradkhanli Keywords : predictive analysis, regression model, learning algorithm, data analysis
During the period of actively using process equipment events or
accidents that may adversely affect it or cause its failure inevitably occur. A
model able to predict the future emergency would make it possible to timely
take measures for eliminating it, thus helping to achieve more efficient use
of process equipment. Development and investigation of such models is the
subject of predictive analytics. The modern trends in predictive analytics
combine the methods of statistical and intellectual analysis of data with the
use of learned algorithms. They are inherent in all of the presently available
predictive analytics methods applied in thermal power engineering and imply
preliminary “teaching” of the model on the basis of available input data. Such
input data include “historical” values of measured parameters characterizing
the operation of particular process equipment; these data are taken from the
archives stored in the power unit process control system’s set of
computerized automation tools for a long time of power unit operation
(usually 1–3 years). In addition, data about the defects (malfunctions)
revealed for this period of time, which can lead to an accident unless having
been removed, are also used.
