Multi-source data fusion in detection of blast furnace burden surface

In consideration of the difficulty of directly using the multi-sensor detecting data in detection of the burden surface of a blast furnace (BF), a novel approach is put forward. The method fuses height data and temperature data and makes use of material mechanism to estimate the non-detecting points to obtain the burden surface. First, multi-sourced data obtained by dissimilar sensors are dealt with in both the time dimension and the spatial dimension. Then, a specific means of loop domain registration is proposed to derive the height of burden surface from the temperature of burden surface. Finally, by combing with the physical properties of surface shape and using Bayes fusion for the theoretical shape and multi-sourced data, the image of burden surface shape of BF is acquired. The experiments indicate that the measurement accuracy has improved by 5.4%, and the resolution of BF has improved by 0.43 as compared with that the traditional burden surface shape estimating method. The method provides necessary guidance for energy saving operation of blast furnaces.