镁合金拉伸矫直过程中材料参数在线辨识

发布时间：2018-04-16 20:04

本文选题：拉伸矫直 + 参数辨识　；参考：《河南工业大学》2017年硕士论文

【摘要】：镁合金作为目前已知最轻的合金材料,被广泛的应用于军工机械、汽车制造、医疗器械等领域中。目前针对镁合金研究的存在很多欠缺,已研发成果大多都是初步的和分散的,并没有将其进行统一和标准化。以镁合金矫直为例,工业生产中对镁合金平直度的矫直虽已取得一些重要的基础性研究成果,但有关镁合金矫直理论的研究仍远落后与工程实际需求,这进一步又限制了镁合金的发展。本文以基于弹塑性理论构建的镁合金拉伸矫直理论模型为基础,采用不同算法对拉伸矫直过程中材料性能参数进行在线辨识,以提高材料性能参数的精确度,进而提高矫直位移精度,主要工作如下:1)对拉伸矫直模型进行介绍分析,建立起相应的求解模型和算法,利用matlab软件对该算法实施仿真实验,并通过实例验证该方法的有效性。2)基于遗传算法优化BP网络的原理;设计针对矫直模型的网络结构;通过样本训练,获得要辨识的材料参数值;并将其与BP神经网络算法结果进行比较,确认优化后的网络预测精度更高。3)进一步结合拉伸矫直理论模型,分别描述基于弹性加载第一、第二阶段和弹塑性加载阶段的递推最小二乘辨识方法;将其预测结果与试验结果进行对比分析,该方法的辨识结果满足工业精度。4)研究基于Kalman滤波与扩展Kalman滤波的材料参数联合辨识,分别建立基于弹性阶段的Kalman滤波(KF)模型和基于弹塑性阶段的扩展Kalman滤波(EKF)模型,将预测结果进行分析。最后对本文工作进行总结,并对镁合金矫直过程中材料参数辨识方法研究作了进一步的展望和分析。
[Abstract]:Magnesium alloys as the lightest known alloy materials are widely used in military machinery, automobile manufacturing, medical devices and other fields.At present, there are many deficiencies in the research of magnesium alloys, most of the research and development results are preliminary and scattered, and it has not been unified and standardized.Taking magnesium alloy straightening as an example, although some important basic research results have been obtained in the industrial production of magnesium alloy leveling, the research on the theory of magnesium alloy straightening still lags far behind and the practical engineering needs are still far behind.This further limits the development of magnesium alloys.Based on the theoretical model of tensile straightening of magnesium alloy based on elastic-plastic theory, different algorithms are used to identify the material property parameters in order to improve the accuracy of the parameters.And then improve the accuracy of straightening displacement. The main work is as follows: 1) introduce and analyze the stretching straightening model, set up the corresponding solution model and algorithm, and use matlab software to carry out the simulation experiment of the algorithm.The effectiveness of the method is verified by an example. 2) the principle of optimizing BP neural network based on genetic algorithm, the network structure for straightening model, the material parameter value to be identified are obtained by sample training.Compared with the BP neural network algorithm, it is confirmed that the optimized neural network has higher prediction accuracy. 3) further combining with the stretch straightening theory model, the first one based on elastic loading is described, respectively.The recursive least square identification method for the second stage and the elastic-plastic loading stage is used, and the predicted results are compared with the experimental results.The identification results of this method satisfy the industrial precision .4) the joint identification of material parameters based on Kalman filter and extended Kalman filter is studied. The Kalman filter model based on elastic stage and the extended Kalman filter model based on elastic-plastic stage are established, respectively.The forecast results are analyzed.Finally, the work of this paper is summarized, and the research of material parameter identification in the straightening process of magnesium alloy is further prospected and analyzed.
【学位授予单位】：河南工业大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TG339;TP18

【参考文献】