A New Approach for Compressor and Turbine Performance Map Modeling by Using ANFIS Structure

Abstract

Aviation is the one of the continually developing sector in transportation area. People prefer to travel with more comfortable, more economic, time saver and safety vehicles during their travel. According to this selection, the number of passengers in civil aviation is increasing day by day. Due to this acceleration, air-lines expand their fleet and increase the number of their daily flights. Eventually, manufacturing velocity accelerates as well. Today, aircraft manufacturers focus on trying to design safer, more efficient, more environment-friendly, more economic aircraft. No doubt, it is not an easy process. Aircraft has a complex structure, and its design duration takes a very long time.

Aircraft is composed of two main parts: structural part and system part. Both of them have a critical design process. After the design level, manufacturing and test levels are followed respectively. These levels are valid for all parts in aircraft. It means that every component has a critical mission in aircraft. One of the critical sections of the aircraft is the engine part. In aircraft, different kinds of gas turbine engines are used. A typical example of a gas turbine engine is formed in five parts. These are, respectively, inlet, compressor, combustion chamber, turbine, exhaust parts. Basic operation principle of the gas turbine engine is to convert chemical energy into mechanical energy. In compressor part, the air that is taken from inlet is compressed and pressurized. In the combustion chamber, the pressurized air is burnt under the high temperature. Then the gas form is taken to the turbine part and it expands to the ambient pressure crossing through the turbine. Finally it is exhausted in the exhaust part.

In the design process of gas turbines, all parts have different design issues. Especially, compressor and turbine design models have critical points that must be considered due to their geometrical structure. When preparing models for these parts, component maps are used. Especially, for the transient conditions, maps represent the component performance very well. The maps are difficult to directly use in simulations so, different techniques are used to read data from maps. In our study, a technique based on the “Adaptive Neuro-Fuzzy Inference System (ANFIS)” is tested and proposed on MATLAB/Simulink.

Nomenclature

${\dot{\mathrm{m}}}_{\mathrm{corr}}$ :

Corrected mass flow rate, dimensionless

π:

Pressure ratio, dimensionless

N_corr :

Corrected revolution, dimensionless

N:

Revolution

η:

Isentropic efficiency

$\dot{\mathrm{m}}$ :

Mass flow rate

δ:

Compressor input pressure under sea level condition, dimensionless

θ:

Compressor input temperature under sea level condition, dimensionless

O:

Output function

p, q, r:

Premise parameters

f:

Fuzzy rule

μ(x):

Membership function

w_i :

Firing strength of the rule

${\overline{\mathrm{w}}}_{\mathrm{i}}$ :

Normalized firing strength value