Research On The Prediction Of Molten Steel Temperature In Refining Furnace And Optimization Of Refining Process

Refining with LF refining is an important step in the steelmaking process.It can further remove inclusions to obtain clean steel,accurately adjust the composition and temperature of molten steel,and improve the shape of non-metallic inclusions.The precise control of molten steel temperature is the most important thing in the production process of LF refining furnace.It helps reduce labor costs and operational risks,improve steel quality,and reduce energy consumption.Therefore,it is very important to establish the LF refining furnace steelmolten temperature prediction model and process refining optimization model.The specific research results of this article are as follows:(1)Based on the mass production and material consumption data collected on site,this paper conducts data preprocessing and correlation analysis,combines expert experience and mechanism analysis to determine the influencing factors of molten steel temperature,and establishes a prediction model for the end point temperature of molten steel in LF furnace.First,a temperature prediction model was established based on the BP neural network.The simulation of the training model found that because the weights and thresholds of the BP neural network are random,the results of each training are different and the prediction accuracy rate is 79%,which does not meet the production demand target of the enterprise.Then,in view of the shortcomings of the BP neural network,this paper uses the gray wolf algorithm to optimize the BP neural network and retrains the model simulation.The training results are basically the same,and the model accuracy is also increased to 88%,which basically meets the production requirements of the enterprise,but no matter how to adjust the gray wolf The parameters of the algorithm and the accuracy of the model can no longer be improved.The deep network can handle deeper and more complex problems.In order to further improve the accuracy of the model,the deep belief network is used to establish the LF refining furnace molten steel temperature prediction model,and through parameter optimization,the optimal parameters are selected for simulation training.The simulation results show,The accuracy of the model based on the deep belief network reached 93%,and the accuracy of the model was significantly improved.(2)Collect the images of the temperature measuring gun on the spot,compare and analyze the image preprocessing methods through simulation,and select the most suitable image filtering and image segmentation methods.After image filtering and image segmentation preprocessing,colorimetric temperature measurement is adopted.Method to obtain the molten steel surface temperature at the moment the temperature measuring gun is lowered.The temperature difference between the obtained result and the temperature measured by the thermocouple of the lower temperature measuring gun is very small.(3)In order to reduce energy consumption,combined with expert experience and mechanism analysis,it is determined that the minimum product of electricity consumption per ton of steel and heating time is the optimization goal,the constraints are determined and the optimization algorithm is established to establish an optimization model.The onoptimizati model is on based the principle of the steel plant LF furnace circuit.The original model was established 20 sets of Q235B steel grade production data are used for experimental simulation and compared with historical data production.Experimental results that theshow optimized operating parameters can effectively reduce energy consumption.