Control study based on fuzzy pattern recognition of blast furnace surface

The blast furnace smelting has a very close relationship with the shape of the blast surface and the control strategy of the shape of the blast surface can get stable high yield of the blast furnace, lower coke rate, reduce erosion of the furnace wall, and extend the life of a blast furnace. In this paper, Bayesian data fusion method and cubic Hermite interpolation, combined with the volume constraints are used to fit large series of the six radar scanning surface data, forming the shape of the blast surface sample library .By using fuzzy c-means clustering classification algorithm on the original surfaces in the shape of the sample library, the shape of the blast surface category space is established. For different shape expectations, calculate the distribution matrix needed to achieve the different shape expectations from different categories of the category space. For unknown categorical variable measured by radar in real time, use fuzzy recognition method to match its shape to the category of the categories space, thus using the appropriate distribution matrix to distribute. Due to the blast furnace radar real-time measurement, real-time correction of the shape of the blast surface is achieved to correct the shape of the blast surface until it reaches the expected shape. This control strategy was implemented on a 2500m3 blast furnace of a steel plant, and achieved good control effect and energy saving requirements.