Method and system for controlling environment of glass furnace based on causal reasoning

Abstract

The invention provides a method and a system for controlling the environment of a glass furnace based on causal reasoning, which are used for a total-oxygen glass furnace furnace taking natural gas and oxygen as fuels, and the method comprises the following steps: taking a data set of a glass furnace related parameter time sequence and a glass quality time sequence acquired by a sensor as input; obtaining a sparse causal model through a condition selection algorithm; taking the sparse causal model as input, and obtaining a sparse causal model in which false positive causal association is removed through an instantaneous condition independence test; based on an information theory, adopting a source entropy concept to measure causal strength, and giving a corresponding control decision according to the causal strength. According to the method, for a glass furnace environment with a high-dimensional nonlinear time sequence and a time lag characteristic, the time lag causal relationship can be reasoned; and the causal association strength can be quantified, and decision support suggestions are provided for environment control of the glass furnace.

Description

technical field [0001] The invention relates to the technical field of glass melting furnace environment control, in particular to a method and system for controlling the glass melting furnace environment based on causal reasoning. Background technique [0002] The glass melting furnace is a complex system accompanied by various physical and chemical reactions. During the clarification and homogenization process in the melting pool or forehearth, the molten glass needs to be kept in a specific temperature range for a long time. In order to obtain good glass Quality requires certain control over the environment inside the furnace. At present, the traditional PID control technology is mainly used in related fields in our country. This technology is relatively cheap and easy to design, but for a high-dimensional, nonlinear and time-lag system like a furnace, the control effect is not ideal, and it is difficult to produce high-quality glass. In addition, many researchers have ...

AI Technical Summary

Problems solved by technology

[0004] Although these improved algorithms perform well under experimental simulation conditions, they cannot achieve good performance in practical applications.

One is that the complex furnace environment cannot be perfectly simulated; the other is that the neural network-based machine learning algorithm cannot explain causality and is not convincing for actual production decisions

Images