Research On Soft Sensing Technology And Burning State Recognition Of Cement Rotary Kiln

As the key equipment for calcining cement clinker,the calcining process has a direct impact on the final yield and quality of the clinker.Among them,the free calcium oxide(f-CaO)is an important performance index to measure the calcination process of the clinker in the rotary kiln and can directly characterize the calcination of the material in the rotary kiln.Therefore,the production control of cement clinker by detecting the f-CaO content of clinker is of great importance to ensure the quality requirements of cement.At present,the acquisition of f-CaO content is still based on the off-line sampling method.The test results have a serious lag in judging the calcining of the kiln clinker,which leads to the difficult application of the traditional control mode of rotary kiln based on the clinker quality.In addition,the burning condition of the rotary kiln is closely related to the f-CaO.It is an important parameter reflecting the calcination status of the clinker in the rotary kiln.It directly determines the final calcination of the clinker and the energy consumption of the cement production process.Because the content of f-CaO,which is closely related to the burning state,can't be detected online,the operators are still relying on the way of "operators observing fire " and the parameters characteristic of other on-line process parameters to judge the burning state in the rotary kiln.This mode of operation will result in a decrease in the accuracy of the identification of the burning state and instability of the clinker quality index,which will have an adverse effect on the final burning of the clinker.Therefore,the on-line detection of the f-CaO content and the accurate recognition of the rotary kiln burning state are important issues to be solved urgently.In view of the above problems,this project takes 5000t/d clinker production line of a cement company in Shandong Province as the research background,a soft-sensing technology was used to establish a soft-sensing model of f-CaO based on LS-SVM,and achieved the online detection of f-CaO content.Using intelligent recognition method based on expert system to realize the recognition of the burning state and various working conditions in the rotary kiln.The specific research content is as follows:(1)For the problem of outliers and random errors in process data,three different data preprocessing methods are used to preprocess the process data.Among them,the burning zone temperature is a key parameter related to the research of this topic.In order to solve the problem of large data fluctuations,this paper proposes a data smoothing method based on local weighted regression scatter smoothing(LOWESS)to smooth it.(2)Aiming at the problem that the f-CaO content of clinker quality index is difficult to detect online,the LS-SVM algorithm is used to perform soft-sensing research.Based on the clinker calcining mechanism analysis and operation experience,the model auxiliary variables are determined,and the time matching scheme between the auxiliary variables and the model output variables is established.Grid search and cross validation method were used to determine the best parameters of the model.A soft sensor model of f-CaO based on LS-SVM was established and verified.(3)Aiming at the difficulty of accurately identifying the burning state of rotary kiln,a method based on expert system for identifying burning state is proposed in this paper.Based on the analysis of cement technology and the burning state of rotary kiln,three basic burning states and related working conditions were summarized,and the related process parameters were determined.Using fuzzy inference rules and expert knowledge to accurately and comprehensively identify the state of burning in the kiln and various working conditions.(4)Software platform development.A software application platform that can be used for f-CaO online prediction and burning state recognition is developed by using C# programming language.The software platform mainly includes three functional modules: data acquisition,f-CaO online prediction and rotary kiln burning state recognition.Among them,the f-CaO online prediction and rotary kiln burning state recognition modules both achieve data exchange between the cement production site and the database through the data acquisition subsystem.Finally,apply the software to the production site.The results show that the software can realize on-line prediction of f-CaO and accurate recognition of the burning state of rotary kiln.The results of this research are of great practical significance for improving cement clinker quality,reducing production energy consumption and guiding cement production.