Cell Mapping-Based Modeling And Control For The Pattern Moving

Complex dynamical processes are frequently encountered in many fields of engineering,such as the sintering process of ironmaking,the process of alumina production by sintering,the cement sintering process in rotary kiln and so on.By examining the dynamics for systems of this kind,we find that they have some common features including various complex physical and chemical changes such as hydrodynamics,thermo-mechanics,heat transfer,phase transition,moving boundary,multiple parameters describing process and product quality frequently existing and so on,which lead to difficultly construct a mechanical model.Meanwhile,the dynamics in nature follow the statistical laws.Consequently,an effective modeling method is lack for such systems dynamics with statistical law.Without models,it is a troublesome issue to analysis system performance.For systems of this kind,in this paper,the problem of the modeling and control based on cell mapping is studied.The detailed results are as follows:1)For the problem of the pattern class variable uncomputable,it is endowed with statistical attributes by the cell measure,then a cell mapping model of the pattern moving is constructed Firstly,a pattern scale "space"is constructed by an improved iterative self organizing data analysis technique algorithm(ISODATA)clustering method.According to clustering result,a cell space is presented by dividing the output space into a number of discrete cells.The pattern class variable is measured via the corresponding cell center Based on this measure,a controlled autoregressive moving average(CARMA)model of the pattern moving is established for this class of complex systems by pattern recognition approaches.Secondly,the model parameters are identified by an improved quantum-behaved particle swarm optimization(QPSO)algorithm.Thirdly,based on the model,the corresponding cell mapping model is derived by the state space representation.Finally,the global analysis of pattern moving system is studied by applying the simple and generalized cell mapping algorithm.2)For the model of the pattern motion,a controller is established using cell mapping theory.Firstly,the system input of bounded closed sets is discretized,all of cell mapping pairs are obtained by exhaustive method through state cell space.Secondly,the measurement of the reference pattern is taken as the target cell,and the cost function as the performance index,the shortest path method is used to obtain the optimal path for all controllable cells to reach the target cell.Finally,the related information on the path is recorded to form a control table.In other words,a cell mapping-based controller is obtained.In view of the uncertainty of the model parameters,the robustness of the controller is studied by investigating the variation number of one-step transfer cells.3)For the problem that cell size is affected by clustering parameters,the influence of clustering parameters on control system performance is analyzed.Firstly,the influence of parameter changes on the global characteristics is analyzed,and the change of the equilibrium cell and the attraction region are obtained.Secondly,the influence of parameter change on regulation performance is studied.Finally,via the actual data collected from Anyang Sintering Plant,a simulation example is given to demonstrate the effectiveness of the proposed approach.