Monitoring Near Burner Slag Deposition with a Hybrid Neural Network System

This paper is concerned with the development of a system to detect and monitor slag growth in the near burner region in a pulverized-fuel (pf) fired combustion rig. These slag deposits are commonly known as 'eyebrows' and can markedly affect the stability of the burner. The study thus involved a series of experiments with two different coals over a range of burner conditions using a 150 kW pf burner fitted with simulated eyebrows. These simulated eyebrows consisted of annular refractory inserts mounted immediately in front of the original burner quarl. Data obtained by monitoring the infra-red radiation and sound emitted by the flame were processed to yield time and frequency-domain features, which were then used to train and test a hybrid neural network. This hybrid 'intelligent' system was based on self organizing map and radial-basis-function neural networks. This system was able to classify different sized eyebrows with a success rate of at least 99.5%. Consequently, it is possible not only to detect the presence of an eyebrow by monitoring the flame, but also the network can provide an estimate of the size of the deposit, over a reasonably large range of conditions.