摘要：For the issues of the discontinuous measurement of the material level of the mechanical stock rod and the measurement of radar probe with weak anti-disturbance ability, high accuracy fluctuations and poor stability, combining with the semi-fuzzy and clustering characteristics of the radar gauge rod data, a new semi-fuzzy clustering algorithm based on membership degree classification is first proposed to achieve clustering analysis of radar gauge rod data. Then, to obtain the accurate material level information, a sliding window correction model based on mechanical stock rod data is built to correct the radar probe data, which can effectively get the continuous real time information of the material level in the blast furnace. Both the simulation results and industrial validation show that the proposed method can provide real-time and effective material level information, and it has a practical value for industrial production.
