End point inference control in the process of argon-oxygen refining low carbon ferrochrome based on adaptive fuzzy neural network

In the process of argon-oxygen refining low carbon ferrochrome, the temperature and carbon content control of end point are the core of the whole blowing process, however refining process is very complex, multiphase and high temperature and the boundary conditions are constantly changing in the control process, the carbon content of molten iron can not be detected continuously online, the reaction process is affected by many uncertain factors ,it brings great difficulty to the end point control of refining process, for the above question, Based on mathematical models, this paper presents an adaptive fuzzy neural network inference control method, Control the end point temperature and carbon content of molten iron, select the temperature of molten iron as the secondary output, select carbon content as the main output, the carbon content of molten iron is controlled indirectly by controlling the temperature of molten iron, an inference controller is designed, for when the model is mismatched, the filter parameters of the controller remain always be unchanged, for characteristics that affects control performance of the system, an adaptive fuzzy neural network is proposed to dynamically adjust the filter parameters online, the simulation results showed that this method has good performance of anti-interference, robustness and set value tracking, the method also provides an effective way to solve the control problem of industrial objects whose output and disturbance are unmeasurable.