A hybrid soft sensor for measuring hot-rolled strip temperature in the laminar cooling process

To overcome the difficulties associated with the frequently varying operating conditions of the laminar cooling process and measuring the strip temperature in the cooling process online, a soft sensor model of the hot-rolled strip is proposed which combines mathematical and hybrid intelligent methods. The proposed approach is based on computational intelligence techniques, where RBF neural networks, CBR and fuzzy logic reasoning are employed to estimate process parameters for predicting the coiling temperature of the strips. A number of simulation tests using industrial data are conducted where the desired numerical results are obtained. It has been shown that the proposed soft sensor has a high potential for being used to effectively measure the strip temperature in the laminar cooling process.