MONITORING OF QUALITY OF THE AUTOMATED CONTROL BY TECHNOLOGICAL LINES OF PICKLING OF BAND STEEL

Abstract

In this paper the approach to process control rolling strip steel using diagnosis neural network  was considered. An approach to automated control of metal loop length in pickling bath was proposed. A method for diagnosing the state of pickling line using quality index, which is determined by using the neural network model was designed. The first step of the offered procedure consistsof determination of class of quality of products in accordance with the output dynamics of process. The second step is a construction of simulation model. Optimal quality of products corresponds to minimum metrical distance between the mean value of optimal remain and his actual value.Obtained dependencies allow assessing the condition of the pickling unit without it stopping. The simulation results confirm the effectiveness of the proposed method to obtain the guaranteed accuracy of identification and reduce the deviation of output parameters from preset values.