Asee peer logo

Neural Network Modeling Of A Power Generation Gas Turbine

Download Paper |


1997 Annual Conference


Milwaukee, Wisconsin

Publication Date

June 15, 1997

Start Date

June 15, 1997

End Date

June 18, 1997



Page Count


Page Numbers

2.305.1 - 2.305.11

Permanent URL

Download Count


Request a correction

Paper Authors

author page

William E. Cole

Download Paper |

NOTE: The first page of text has been automatically extracted and included below in lieu of an abstract

Session 2533

Neural Network Modeling of a Power Generation Gas Turbine

William E. Cole Northeastern University


Over the past several years, I have supervised students creating Neural Network computer models of operating processes for their senior project. Processes modeled include a gas turbine power generator, a furnace, and building energy use. Models were created and used for parametric analysis within the scope of a one semester course. This modeling effort brought the actual operating process into the classroom, demonstrated to the students the value of computer modeling, and demonstrated that fundamental principles taught in the classroom apply to actual operating processes. This paper focuses on using neural networks to model processes, what students can learn from developing a neural network model, and one student’s model of a gas turbine power generator.


The complexity of operating processes and the inherent difficulty of modeling real equipment makes modeling of industrial processes extremely difficult. The real equipment does not necessarily perform exactly as characterized by the idealized equations used in the models. Consequently computer models created from first principles are complex and frequently do not fit the operating data very well. Additionally, these models cannot account for the individual nuances of operating equipment and are not able to accommodate changes as the equipment ages. Consequently realistic models cannot be created from first principles within the scope of a one semester project.

An alternative technique to model complex processes is to utilize neural networks. Neural network modeling contrasts with conventional computer modeling in that a detailed understanding of the process is not required. The neural network uses operating data to create the model. Neural networks have been used to model complex processes such as distillation columns,1 nuclear reactors,2 and automotive fuel injection.3 Additionally, realistic models can be created within the scope of a one semester project. Neural networks have also been used in a graduate level course at Tufts University.4

A neural network is composed of processing elements and connections as shown in Figure 1. The processing elements are arranged in three layers. In the first layer, each element represents one of the input parameters and the element in the third layer represents the output parameter. For educational purposes, each network should have only one output element. Additional processing elements are arranged in a second hidden layer. Each input processing element is connected to each hidden element and in turn each hidden element is connected to the output

Cole, W. E. (1997, June), Neural Network Modeling Of A Power Generation Gas Turbine Paper presented at 1997 Annual Conference, Milwaukee, Wisconsin.

ASEE holds the copyright on this document. It may be read by the public free of charge. Authors may archive their work on personal websites or in institutional repositories with the following citation: © 1997 American Society for Engineering Education. Other scholars may excerpt or quote from these materials with the same citation. When excerpting or quoting from Conference Proceedings, authors should, in addition to noting the ASEE copyright, list all the original authors and their institutions and name the host city of the conference. - Last updated April 1, 2015