Case-reasoning-based molten steel temperature prediction method

Abstract

The invention discloses a case-reasoning-based molten steel temperature prediction method. The method comprises the following steps of: 1, establishing a case library in which the values of influence factors of molten steel temperatures and the molten steel temperatures are stored; 2, defining a present problem and a problem solution; 3, sorting case sets in the case library according to key state vectors; 4, roughly selecting the sorted case sets according to the discrete values of all the state vectors; 5, finely selecting the roughly-selected case sets according to the gray scale similarity of the present problem to candidate cases; and 6, preferentially calculating matched cases of which the gray scale similarity is greater than a preset similarity threshold value in a weighted mode to acquire the problem solution of the present problem. Based on a multistep retrieving policy case reasoning method, the defects that the traditional neural network method has long training time and is not suitable for on-line application and the like are overcome; and due to a weighted value-contained gray correlation degree method, the problems of incompleteness and interference existence of production data are solved to a certain extent.

Description

technical field [0001] The invention belongs to the technical field of steelmaking and relates to a method for predicting molten steel temperature based on case reasoning. Background technique [0002] With the improvement of production refinement in steelmaking plants, the requirements for temperature control of molten steel are getting higher and higher, and the requirements for obtaining the temperature of molten steel in each process in a timely and accurate manner are more urgent. The traditional thermocouple measurement method has the disadvantages of high labor intensity, low measurement accuracy and stability, inability to perform continuous measurement, and lagging data feedback, which cannot be adapted to compact and efficient modern production. Due to the inability to obtain the temperature of molten steel in a timely and accurate manner, many enterprises have greatly increased the tapping temperature of the converter, wasting a lot of energy. Due to the complex ...

AI Technical Summary

Problems solved by technology

However, there are deficiencies in the existing case-based reasoning technology applied to prediction: the nearest neighbor strategy or inductive reasoning strategy is used in the case retrieval method, and accurate results are usually not obtained when the case information is incomplete; when the case base is large time, the retrieval time is longer

Images