This week, the IFRF Handbook ventures for the first time into the upcoming area of neural network applications to combustion. Thanks to my colleagues at University of Glamorgan, CK Tan (author) Steve Wilcox (referee) and John Ward (sub editor).
Artificial Neural Networks (ANN for short) offer a powerful alternative to conventional curve fitting techniques when attempting to derive complex relationships between dependent parameters based on an existing set of data. They also offer a highly flexible alternative to mathematical models based on the underlying physics and chemistry of a reacting system such as a combustion process. In particular, when compared to conventional mathematical models, ANNs are fast enough to allow real time simulation and control of heating processes. Today we publish two Combustion files to introduce this topic.
CF46: What is an Artifical Neural Network (ANN)? – Open Domain
CF237: How do I use a neural network to calculate calorific values of solid fuels? – PowerFlam2
CF46 is a general introduction on the topic, and has been placed in the open domain for wide dissemination of the ANN concept.
CF237 is an example of a specific application of the technique to develop a model relating the calorific value of solid fuels to their ultimate analysis. This Combustion File has been prepared as an early phase of work on ANN modelling of pulverized coal substitute fuel co-firing undertaken by the University of Glamorgan within the EC funded PowerFlam2 consortium, coordinated by Cardiff University. Access to the full text of this Combustion File is currently restricted to IFRF Individual Members who are also authorised members of the PowerFlam2 Consortium.
To read or download these and other recently published files, please go to www.handbook.ifrf.net, and select ‘New Combustion Files’ after accepting the disclaimer screen. Other Combustion Files on the same theme may be found by using the links built into CF46 or by searching the Handbook with the Keywords ‘Modelling’ or ‘Calorific’.