Research On Splash Measurement And Control Method Based On AOD Furnace Flame Image Analysis

Argon-oxygen refining of low-carbon ferrochrome is a new process for the production of ferroalloys.The smelting temperature during the smelting process is more than 100 °C higher than that of ordinary smelting.The smelting furnace has a small smelting capacity and high sulfur and other impurities,which makes the smelting process easier.A splash accident occurred.Therefore,the prediction and control of the smelting process in the smelting process of the AOD furnace are the key issues that must be faced and solved during the industrialization of the argon-oxygen refining ferroalloy process.This article takes the smelting process of ferrochromium alloy in AOD furnace as the specific research object,and studies how to predict and suppress the occurrence of splashing through the characteristics of the smelting process.The splash mechanism of the AOD furnace is analyzed first,and a splash characteristic signal is used to control the variables that occur in the splash,so that the characteristic value is controlled within the range of normal smelting to achieve the purpose of suppressing the splash.Based on the above objectives,the main research contents completed in this article are as follows:1.Firstly,analyze the process,process and mechanism of spattering in the process of smelting ferroalloy in the AOD furnace,summarize the spatter-related characteristics,obtain data based on expert experience and actual spattering,and image pre-process the flame image of the furnace mouth.A more suitable method for furnace flame pretreatment was presented.2.The feature extraction of AOD furnace flame was studied.First,the obtained smelting process images were divided into normal work and pre-spatter images.Several edge image extraction and denoising methods were studied and compared.According to the gray histogram of the image,five statistical feature quantities were calculated,and their respective classification capabilities were comprehensively evaluated using the distance between classes.The experiments found that the five feature quantities have the largest discrimination ability for normal work and splashing.Gray value,energy value,average value,average gray value,flame abundance.The decision-level information fusion is used to assign weights and fuse the obtained 5 feature quantities to identify the working status of the AOD furnace as a feature quantity.3.Investigate the process design of the gas supply system,oxygen lance liftingsystem and feeding system,and analyze the furnace during normal smeltingIn the state,the change curve of oxygen flow rate,nitrogen flow rate,oxygen lance height and feeding speed with time.Because it is difficult to establish accurate parameter models in the production process of AOD furnaces,a multi-rule fuzzy control idea is used to divide the image characteristic values into different sections,and the different sections are used to control the oxygen flow rate,nitrogen flow rate,and oxygen gun,respectively.Height and amount of foam inhibitor added.Simulations show that due to the different effects of output on the reaction speed in the AOD furnace,although the system reaches a different stabilization time,the system stabilization time will be within 60 seconds,and the image feature value will be stable within the normal range.4.Finally,the DCS control system is designed to suppress the splash process of the AOD furnace.The bottom layer selects Siemens S7-300 series PLC,and the top-level monitoring center selects Siemens MP377-15 touch screen.The real-time network for communication between the two layers is done via PROFIBUS-DP and MPI bus.The top gun and bottom gun gas supply subsystems and ingredient control subsystems are mainly designed,and the corresponding control interfaces are given.