Method for predicting content of silicon in molten iron on basis of Elman-Adaboost strong predictor

Abstract

The invention discloses a method for predicting content of silicon in molten iron on the basis of an Elman-Adaboost strong predictor, and belongs to the fields of monitoring, modeling and simulation of industrial process. The method comprises the following steps of: firstly selecting proper input variables, normalizing the input variables to serve as inputs of K weak predictors; secondly, determining an initial weight value of a training sample; thirdly, respectively training the K weak predictors, updating the weight value of the sample according to the training results, and carrying out repeated training; and finally, calculating the weights of the weak predictors, and fusing the plurality of weak predictors to obtain a prediction result of the strong predictor. Through the dynamics of a blast furnace ironmaking process, the precision of predicting the content of the silicon in the molten iron is not high; aiming at the problem, the method selects Elman neural networks with relatively good dynamics as weak predictors, and fuses the plurality of weak predictors according to an Adaboost algorithm to obtain the Elman-Adaboost strong predictor; and the researches of the method applied to predicting the content of the silicon still belong to a blank space. Compared with the conventional methods, the method disclosed by the invention has higher precision of predicting the content of the silicon in the molten iron.

Description

technical field [0001] The invention belongs to the field of industrial process monitoring, modeling and simulation, in particular to a method for predicting silicon content in molten iron based on an Elman-Adaboost strong predictor. Background technique [0002] The production level of the iron and steel industry is one of the standards to measure the degree of industrial automation of a country, and blast furnace ironmaking is the core unit operation in the iron and steel industry. Blast furnace ironmaking is the process of reducing iron ore to iron, which is time-varying, nonlinear, multi-scale and dynamic. The silicon content in molten iron is an important variable that reflects the quality of pig iron and the thermal state of the production process. A high silicon content indicates residual coke in the furnace, and a low silicon content indicates that the energy reserve in the furnace is exhausted. Lower silicon content can not only stabilize the quality of pig iron an...

AI Technical Summary

Problems solved by technology

In recent years, many scholars have applied the BP-Adaboost strong predictor model to regression problems. Most of these studies have selected the BP neural network as the weak predictor, and the BP neural network is a static network that cannot effectively reflect the dynamics between input and output data. sex

[0006] Although the Elman neural network has been applied to the prediction of the silicon content of blast furnace hot metal, the application of the Elman-Adaboost strong predictor model to the prediction of the silicon content of the blast furnace hot metal is still blank; and the Elman-Adaboost strong predictor model is used in the estimation of target threats. Some applications, but the two research fields are quite different

Images