Blast furnace molten iron quality predicting system based on ensemble learning and method

Abstract

The invention provides a blast furnace molten iron quality predicting system based on ensemble learning and a method. The system comprises the components of a blast furnace, a hot air furnace, a firstflowmeter, a second flowmeter, a third flowmeter, a thermometer, a pressure gauge, a humidity gauge, a furnace bosh gas metering analyzer, an oxygen enrichment measuring analyzer, a data acquisitiondevice and a computer. Preprocessing is performed on acquired real-time to-be-tested blast furnace data. Input and output parameters are acquired. A multivariate molten iron quality online predictingmodel of an established root mean square error probability weighted integrated random weight neural network is used for obtaining an online predicting result. The blast furnace molten iron quality predicting system and the method have advantages of preventing uncertainty caused by offline testing hysteresis and manual operation, realizing real-time online accurate soft measurement, supplying key indexes for in-time accurate determining of the internal operation state of the blast furnace by a field operator, updating a soft measurement model parameter by means of latest process data accordingto the working condition change, preventing restriction of a time-invariant model, and realizing high practical value.

Description

technical field [0001] The invention belongs to the technical field of automatic control of blast furnace smelting, and in particular relates to a blast furnace molten iron quality prediction system and method based on integrated learning. Background technique [0002] The blast furnace is a large-scale convective reactor and heat exchanger in the ironmaking process, and blast furnace ironmaking is also an important link in social development. However, the internal smelting environment of the blast furnace is extremely harsh. The temperature in the most violently reactive area is as high as 2,000 degrees, and the pressure is as high as about 4 times the standard atmospheric pressure. Along with the multi-phase coexistence of solid, liquid, and gas, it is difficult to monitor the internal state of the blast furnace in real time. Thus it is difficult to optimize the control of the blast furnace. At present, the indicators that are widely used to indirectly reflect the interna...

AI Technical Summary

Problems solved by technology

[0006] The methods reported in the above-mentioned patents and other relevant literatures are all soft-measurement methods for a single molten iron quality parameter, such as molten iron temperature, Si content, S content, etc., and a single molten iron quality parameter cannot fully reflect the internal complexity of the blast furnace. state, it is impossible to provide comprehensive guidance for on-site operators, and the practical application value is low

Moreover, the models established by the existing molten iron quality soft measurement methods are all time-invariant models, which are not suitable for the slow time-varying characteristics of the working conditions of the blast furnace ironmaking process. Therefore, when the actual working conditions of these models change, the prediction results will be inaccurate.

In addition, most of the model structures in the existing soft-sensing methods are relatively complex, and the calculation time of the soft-sensing model is relatively long, and the industrial site needs a simple and fast soft-sensing method, so the efficiency of the above methods is low in practical applications

To sum up, at present, there is no multivariate dynamic rapid online soft-sensing method for the quality parameters Si content, P content, S content and molten iron temperature in the blast furnace smelting process at home and abroad.

Images