The Intelligent Recognition Of Working Condition For Cement Rotary Kiln Base On Information Fusion

The rotary kiln is the controlled object which is multi-factors, slow time-variable, non-linear, big time lag, close coupling and establishes the precise mathematical model hardly , this article utilizes the classification and recognition algorithm to carry on the fault detection and the tendency type recognition to main examination parameters. Through synthesizing the real-time status of many main parameters in judging the status of every parameter and combing the expert knowledge experience, makes the judgment to the current running status of rotary kiln, realizes intelligent recognition of the working condition. After distinguishing the working condition, then make the corresponding control according to the practical working condition.Through summarizing the related control mode and the control strategy about cement rotary kiln and the analysis of NSP cement technical process, we are clear about the inner and outside factors which affect rotary kiln movement, determine main examination parameters and Control method which influence rotary kiln system movement. Designs the classification and recognition algorithm which is constituted by fault signal detection method based on the fractal and the ART-2 neural network by studing the fractal theory and the multi-sensor information fusion technology,. The classification and recognition algorithm mainly carries on craft fault detection and the tendency type division to the examination parameters. Judges the current running status of examination parameters and gives an alarm if it is abnormally in fault signal detection method based on the fractal. After fault detection, metrical data enters the ART-2 network to carry on the recognition of tendency; The original ART-2 network has three insufficiencies in the practical application: the question about pattern drifting; lack of amplitude model comparison; can not treat negative real data. This article makes the improvement to the algorithm in view of these insufficiencies. The improved ART-2 network enhances cluster accuracy effectively and makes accurate division for the examination parameter tendency type, and verifies the classified recognition algorithm with the actual data; Through summarizing the change of rotary kiln status caused by main craft parameters changing and combing the scene experience, sums up the scene working condition for rotary kiln. The expert knowledge library is composed by these working conditions. After the data of craft parameters is deal with the classification and recognition algorithm in parallel, ascertains the practical working condition and gives an advice for the rotary kiln control through combing the result of recognition and the knowledge library of working condition.Finally makes different module for every parameter, ADO and OPC with VC++ programming base on the design plan of the recognition to rotary kiln working condition, stores the historical data in SQL database. Carries on the classification and recognition to historical data of cement craft parameter in the SQL database through ADO, contrasts its recognition result to the scene actual working condition to complete the knowledge library of working condition. After that, the working condition recognition algorithm may carry on the data communication with DCS through OPC and the classification and recognition to the scene real-time data, implements the on-line identification.The practical application indicates that this system can recognize the working condition of rotary kiln effectively, and makes the foundation for optimal control to rotary kiln. The software adopts the structure of modularization and has the portability, it has quite good usability.