Prediction of Molten Steel Temperature during Outside Refining Process Using Neural Network

摘要

A vessel connected to the ladle through an uprising pipe, a gas control valve provided in a gas inlet connected to the uprising pipe, and a gas sensor and a probe for detecting the temperature of molten steel in the ladle, A vacuum PLC, a reflux control PLC for exchanging data between a thermistor and a vacuum sensor, a ferro-alloy PLC having a CRT for controlling the ferro-alloy hopper, The molten steel temperature predicting method using an out-of-scope refining system including a process computer activates a neural network in accordance with a process start command of a process computer, and receives operational performance data detected from an on-site facility sensor during operation, And transmitting a control amount command corresponding to the control amount command to the PLC at a predetermined period / (several tens of seconds) The method of claim 1, further comprising the steps of: calculating the temperature change per minute by reading the molten steel-related data of the current operating charge stored in the online performance file; and informing the operator of the temperature predicted by the neural network to the CRT, Setting a control amount such as an amount of iron alloy for the temperature hit in the present predicted temperature and performance arrival temperature data to a process computer and adjusting the temperature at a constant cycle; and a step of estimating the temperature of the neural network input data and the temperature prediction result Storing the data in the network performance file; and re-learning the neural network when all the data stored in the database is re-edited with new data after completion of the learning, adjusting the weights corresponding to the equipment conditions.