Harasymiv V. Identification and control of parameters of the two-stage centrifugal compressor of natural gas using artificial intelligence methods

The thesis is dedicated to the development of the control method of parameters of the two-stage centrifugal compressor of natural gas using its mathematical model. Implementation of the method in the automatic control system of the compressor station gives the opportunity to watch the operation of the centrifugal compressor and to keep track of pre-damage situations throughout timely interruption of operation of the centrifugal compressor. In addition, it is possible to reduce the number of unrequested stops due to the implementation of new algorithms using methods of artificial intelligence. The mathematical model of the two-stage centrifugal compressor is developed, which shows the dependence of the centrifugal compressor productivity on its technological and geometrical parameters and considers losses in gas flow and friction of disks. The method for the optimal complexity empirical model building is improved based on genetic algorithms and the parallel algorithm for solving linear equation systems by the Gauss, which gives an opportunity to increase the computation efficiency. The parametric identification method for the mathematical model of the two-stage centrifugal compressor is developed which makes it possible to determine its parameters and their deviation from the initial values, which corresponds to conditions where the centrifugal compressor is a new one or after its overhaul. Application software for the control of parameters of the two-stage centrifugal compressor is developed, which gives the opportunity of its integration into existing software SCADA system Citect and allows to reduce gas losses and the cost of repair or preventive maintenance.