Research On Typical Working Conditions Of Cement Calcining Process

In the NSP cement production, calcining is one of the most critical parts. The calciningsystem undertakes the task of pre-decomposition, calcination, and cooling, ect.Identifying thecurrent working status of calcining system online in the real time, and choosing the propercontrol strategies are key to achieve precise control of cement production process.The level of working condition recognition and control of calcining system is low atpresent, and the identification and adjustment of the process parameters mainly rely onoperator’s experience. This approach has a lot of subjectivity and hysteresis. Therefore,operating advanced technology to identify the working conditions and identify the trends ofthe process parameters and the overall working condition modes is significant.Through the study of commonly used identification methods, combined with thecharacteristics of the cement calcination, this paper advances a two-stage ART-2neuralnetwork method to conduct trend recognition and pattern recognition continuously, andrealize the working condition recognition of calcining system online. The two-stage ART-2neural network can be divided into four parts, key process parameters selection, datapreprocessing, process parameters trend recognition, and pattern recognition. According to thedemand of working condition recognition and the characteristics of cement production, selectthe key process parameters, and so filter the data to make the curve becomes smooth toobserve data fluctuation trend. Datas collected on the site need to be recognized for the trendsafter preprocessing. According to the practical needs, the trend of parameters can be dividedinto ascending trend, steady trend, and fall trend. Neural activation patterns can be got basedon the trend recognition results, and it can be used in pattern classification of the second-classART-2network, the result is working conditions we require. In this paper, we use two-stageART-2neural network method to recognize the working conditions of cement decompositionfurnace, rotary kiln, and grate cooler. The two-stage ART-2neural network can conduct trendrecognition and pattern recognition continuously after processing, and output the recognitionresults directly. This method has the advantages of easy operation. For cyclone preheater,because the link has single variable condition, so we use fuzzy theory directly to constructmembership function, judge the parameter values based on the level of membership, and we can get the working conditions.This paper use VB and MATLAB program to realize the function of condition recognitionsystem. The working condition judgement rules and data pre-processing part adopt MATLABprograms,and the man-machine interface adopt VB programs. VB call M files and identify thecurrent working conditions, display the working condition recognition results and thecorresponding working condition descriptions in the interface...