Study On Process Object Modeling Methods

Along with the development of industrial control system and a large number of dataaccumulated in various fields, how to discover valuable information and rules from thembecome the hotspots and difficulty in theory and practice. Flow industry taking the leadingposition in national economy is a pillar industry for economic development. In the productionof flow industry, there have been a mass of historical data and currently generated real-timeand dynamic data. In connection with the complexity of production in flow industry, ascharacteristics of the data magnanimity, high-dimensions, strong coupling, uncertainty,imperfectness, inconsistency and data types for diversity and multi-mode, we implyintelligent computing into data mining in flow object which have significant meaning intheory and practice for discovering information and rules from the accumulated data of flowobject. In addition, modeling, optimization and controlling in practical industry can helpdecision-makers bring forward, discover, analyze and resolve problems.Facing present problems existing in characteristics of fluid industry and modelingmethods in fluid object, this paper which bases on the cement production process builds arelationship network model by Flexible Neural Tree，and in order to obtain fairing parametricmodel. The structure and parameters of flexible neural tree are optimized by Multi-ExpressionProgramming and simulation annealing respectively. The major research contents are asfollows:1. Do research on the theoretical knowledge associated with FNT modeling，multi-expression and simulation annealing algorithm. The research includes the basic idea ofthe multi-expression algorithm and simulation annealing algorithm, and also includes theconnections and differences between the two. That is the basis for the following optimizationstudy of the parametric structure and parameter in FNT modeling.2. Do research on the FNT modeling methods for the single production process of fluidindustry. Because the fluid industry production process is complicated, every productionprocess including many procedure, we make the study of a single procedure is the premisecondition of important to the whole process of modeling. Based on the variety of modelingmethods, we make the research and analysis. This paper takes the production process of decomposition furnace as the search object. In order to obtain the best model ofdecomposition furnace, we apply the combination of the multi-expression algorithm andsimulated annealing algorithm into FNT modeling. On account of the built FNT model is atree structure, so that we adopt multi-expression to optimize the structure ensure that thesuperiority of the tree structure, and simulated annealing to optimize the parameter.3. Do research on the FNT modeling methods for the overall production process of fluidindustry. At present, research on the production process of process industry mainly focus on aproduction process，but neglects the integrity of the process industry. The cement quality isnot determined by a production process，but depended on the three whole running conditionof the process. So we make a whole research on the process of decomposition furnace, rotarykiln and grate cooler.The experiment results show that the method for combination of the multi-expressionalgorithm and simulated annealing algorithm can obtain more precise model and approximatecontours of objects.