Research And Application For Optimization Settings Of Controlled Variable In Shaft Furnace Roasting Based On Intelligent Decisions

The roasting process of shaft furnace is one of the most important processes inthe Minerals Processing Factory, the primary task of it is to provide the roasting ore ofhematite with higher magnetism by high temperature deoxidization, in order to fitrequest of the integrated production indexes (MTRR-Magnetic Tube Recovery Rate),hour’s yield and consumed gas. At the moment，the controlled variables in shaftfurnace roasting are set by stability control that is the controlled variables are set tofixed values.Due to the existing characteristics of the interference, work conditionscomplexity, multiple variables and strong coupling process, nonlinear and large delayexist in the roasting process of shaft furnace.It is difficult to establish the mechanismmodel for the optimal setting, therefore the single and traditional parameteroptimization method for setting is not available.Data mining is a technical analysis tool to obtain valuable information from largeamounts of data, and it has been widely used ever since it is proposed. Associationrule mining is an important technology in data mining which is able to mine therelationships between the data attributes, and the research and application of datamining is a relatively active and deep branch.In the paper, the association rulealgorithm is adopted to solve the problem of optimal setting for furnace roastingcontrolled variables which can analyze the relationship between the furnace roastingparameters and provide new ideas to shorten the set time of controlled variables,ensure the correct setting of controlled variables and the safety in production. Majorworks in the paper as follows:1. The shaft furnace roasting process is introduced and the collected data of shaftfurnace roasting is analyzed, then the paper adpots association rule algorithm to dealwith the data.2. The k-means algorithm is improved which can get better clustering effect,improve the efficiency of the clustering and make clustering results more accurate forspecific problems.3. According to the characteristics of the shaft furnace roasting data, combinedwith the classical FP-Growth algorithm and Apriori algorithm, T-Apriori algorithmwhich can greatly improve the efficiency of mining such database is put forward.4. Combined with the application requirements,fristly, the paper uses the improved k-means algorithm to do the data pretreatment,then applies the improvedApriori association rule algorithm to build the data mining model of shaft furnaceroasting and mine the the relationship between the detection variables and thecontrolled variables which makes the rule table. The rule table is imported to thecontrol system in the field. In the process of production, the correct values can beacquired through the association rule table， which provides new idea for theoptimization settings of shaft furnace roasting controlled variables.