Reliable and efficient operation of the gas-pumping unit to a large extent depends on the technical state of the combustion chamber. The poor technical state of the combustion chamber reduces not only the efficiency of the gas pumping unit, but also increases the harmful emissions to the environment of harmful substances such as nitrogen and carbon oxides. Therefore, the determination of the technical state of the combustion chamber by the results of observations on the operation of the gas turbine installation is an actual scientific task.
As a result of the analysis of literary sources and experimental studies, it is established that the mediated parameters characterizing the technical state of the combustion chamber are the total concentration of nitrogen oxides and the mean square value of the vibration velocity of the combustion chamber body. Over time, the elements of the combustion chamber are degrading, which leads to a change in the signs of the technical state of the combustion chamber. Indicators of a technical state form a plane of signs, which, with the help of the neural network Kohonen, are divided into three areas, each of which corresponds to a certain technical state of the combustion chamber - "good", "acceptable", "permissible". A method for constructing separate curves based on genetic algorithms is developed to distinguish the areas of technical states. The equation of separate curves is taken as a polynomial of a certain degree. The developed method on the basis of the class separation criterion determines not only the parameters of the equation of a separate curve but also the structure of such an equation by minimizing the class separation criterion. With the help of the developed software of the task of the technical condition assessment of the combustion chamber, the plane of the signs is divided into three classes and separate split curves are constructed.
Improving the Efficiency of Gas Pumping Units: Problems and Solutions