Refractory wear and skull growth on the hearth walls and the bottom of the blast furnace have been researched. A series of thermocouples were installed in the hearth, and the temperature measurements were recorded in a structured query language every minute. A heat transfer model was used to study the temperature evolution and hearth wear profile using a commercial software package (MATLAB version 5.0) based on computational fluid dynamics. The location of the 1150uC isotherm in the hearth lining has been calculated. An online monitoring tool was used to analyse the temperature distribution in the hearth and offers, to the plant operators, periodic information on the refractory state. Electromotive force (EMF) probes were installed in the hearth to estimate the variations in the liquid level in the hearth and to determine the thermal state (TS) evolution. Good correlation is seen between EMF and TS, and the EMF amplitudes in the different tapholes follow and even precede the local TS.
The adaptation of blast furnaces to the new technologies has increased the operation information so that the sensor information can be known at every moment. However this often results in the supply of excessive data volume to the plant operators. This paper describes an industrial application for self-organized maps (SOM) in order to help them make decisions regarding blast furnace control by means of pattern recognition and the matching of temperature profiles supplied by the thermocouples placed on the above burden. The classification of patterns via easy color coding indicates to the operator what the blast furnace operational situation is, thus making the necessary corrections easier.KEY WORDS: ironmaking; blast furnace; neural networks; self-organized maps (SOM); forecasting. perature data. This temperature is measured each hour by means of two probes, each containing twelve thermocouples. The probes are identical and their anchorages are located in the wall, describing a diameter of the blast furnace section, so the expected behaviour would be a symmetrical temperature profile. However irregularities in the feeding processes, in the load composition and some other factors like cooling yield asymmetric profiles. Besides that, a great dispersion of the data, with many outlier samples, is confirmed. This dispersion (above all in the central thermocouples) is due to the feeding rotational movement so the load is not at the same level at every point. The cooling system acts upon the central zone of the blast furnace if the temperature exceeds 450°C, which may cause an M-profile like that of Fig. 2 and contributes to the dispersion of the data. Figure 3 shows the pig iron temperatures measured in the blast furnace with their mean and standard deviation. The pig iron temperature should remain over 1 450°C from the time it leaves the blast furnace until it is processed in the plant to make iron, this being due to the risk of undesirable reactions between compounds that would yield poor quality iron. This is a big problem in the industry so a lot of effort has been put into improving it.
10)The main objective of using a neural network is to supply useful information to the plant operators. This information must be displayed quickly and in an easy to understand format because an excessive amount of information is overwhelming to plant operators, and therefore useless. In neural networks, each neuron learns a different pattern during training, even if most of them are quite similar to others. In order to avoid excessive information due to the use of a high number of neurons it is necessary consult seasoned plant operators, whose knowledge based on their experience serves to classify all these patterns into two or three classes. Three classifications can be compared: one based on the plant operators' knowledge, the neural network classification and a third, based on the criteria detailed in Table 1, which is purely mathematical but real and commonly used. In later sections we will demonstrate that the classification...
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.