合金热挤压中基于混杂模型的突变预测与仿真  

Simulation of Catastrophe Prediction in Alloy Extrusion Based on Hybrid Model

作　　者：周理[1,2] 蒋新华[1] 王雷[2] 

机构地区：[1]中南大学信息工程学院,长沙410083 [2]福建工程学院电子信息与电气工程系,福州350108

出　　处：《系统仿真学报》2012年第12期2587-2592,共6页Journal of System Simulation

基　　金：福建省高校产学合作科技重大项目(2010H6002);福建省教育厅科技A类项目(JA10221)

摘　　要：通过研究合金热挤压过程中的突变现象与观测参数之间的关系,提出了一种观测值增量估值模型,在此基础上构建了一种基于增量估值模型的混杂系统,实现对挤压过程连续状态和离散状态的有效辨识;其中,针对连续状态的辨识,为了对参量进行有效检测,将参量设计为增量加速度向量,根据观测量瞬间变化时原有稳定系统观测模型与实际情况不匹配,从而导致新息序列统计特性发生变化的事实,构造了一种增量向量检测器;针对离散状态的辨识,采用隐马尔科夫模型,结合混杂模式转换概率矩阵,构建了一种离散状态辨识器。最后,基于新提出的混杂系统模型,设计并实现了一种有效的突变预测算法,仿真实验结果表明,新算法具有良好的突变预判性。An incremental estimation model was proposed by analyzing the relationship between the catastrophe process and parameters. Then, a hybrid system was built, which could identify continuous state and discrete state simultaneously. For the continuous state recognition, the increments of the parameters were designed as the acceleration vector for effective parameter detect. If the estimation of the system mismatches the actual measurement, the detector will find the statistical properties change of the sequence. Meanwhile, for identification of discrete states, a discrete state identifier was constructed by combining the hidden Markov model and mode switching probability matrix. Finally, a catastrophe prediction algorithm was designed and implemented based on the new proposed hybrid system model. Experimental results show that the new algorithm could predict the catastrophe effectively.

关 键 词：突变预测 仿真 增量估值模型 混杂系统 合金热挤压 

分 类 号：TP274.2[自动化与计算机技术—检测技术与自动化装置]