ugg stivali Magnesium Alloys AZ91D ingot material

Magnesium Alloys AZ91D ingot material size changes on the impact of Isothermal Heat Treatment

. 【4】 SekiharaK, OhnishiS, KamadoS, eto1. Semi-solidformingofstrain-inducedAZ91Dmagnesiumalloy [J]. JournalofJapanInstituteofLi 【ghtMetals, l995,[link widoczny dla zalogowanych], 45 (10) :560-565. [5】 Liyuan Dong, Hao Yuan,[link widoczny dla zalogowanych], Yan Fengyun. SIMA AZ91D magnesium alloy prepared by non-dendritic ingot material [J】. Gansu University of Technology, 2002,28 (4) :34-38. [6】 Liyuan Dong, Hao Yuan, Yan Fengyun, et al. AZ91D magnesium alloy in semi-solid isothermal heat treatment of microstructure evolution 『J]. Chinese Journal of Nonferrous Metals, 2001,11 (4): 57l a 575. [7】 Liyuan Dong, Hao Yuan, Chen Ti-jun, et al. Isothermal heat treatment on AZ91D magnesium alloy semi-solid microstructure evolution and formability of .... Chinese Journal of Nonferrous Metals, 2002.12 (6) :1143-l148. [8] KimJN, KimKT, JungWJ. Effectsofisothermalhe ~ ingprocedureandstrontiumadditiononsemisolidformingofAZ91magnesiumalloy [J]. MaterialsScienceandTechnology, 2002, (1 Cool

:698-701. [9】 Yang Guangyu, Haoqi Tang, Jie Wanqi. And so on. Mg-5A1-1.5Ca a 0.4Zn isothermal base alloy microstructure and properties of semi-solid die casting [J】. China Nonferrous Metals,[link widoczny dla zalogowanych], 2005,15 (4): 614.620. [1O] Liyuan Dong, Hao Yuan, Jin flowers, and so on. Semi-solid isothermal heat treatment on AZ91D Magnesium Alloy [J]. T industry in Gansu University of Technology, 2001,27 (1) :27-30. Tin (Continued from page 110) the increase in the number of nodes decreases and then increases, when the hidden nodes for the l4, the minimum prediction error (0.017). Figure 6 (b) can be observed, the forecast error increases with the hidden layers first and then decrease and then increase, when the hidden nodes is 16:00, or. Minimum prediction error (0.112). Figure 6 (c1 can be seen,[link widoczny dla zalogowanych], the prediction error increases with the increase in number of nodes in the hidden layer first increases and then decreases, when the hidden nodes is 6, the prediction error minimum (0.026). From Figure 6 ( a) also shows that the prediction error increases with the increase of number of nodes in the hidden layer decreases and then increases, when the hidden nodes is 16, the predicted and measured values are equal. 3 Conclusions (1) the material the chemical composition as the network input, the mechanical properties as the network's output, you can create reflects the inherent law of the experimental data a mathematical model. (21 errors with the hidden nodes and training times increases; error with the learning rate increase decreases and then increases; Momentum speed up the convergence and prevent convergence. (3) using back propagation (BP) algorithm, based on a relationship between chemical composition of three-layer network model was established 1lx17x4j. come to the training error is 0.726 get a relatively good learning precision. with the network prediction of non-quenched and tempered steel mechanical properties and measured values are close. References: [1] Luo Yi, Wu light wind,[link widoczny dla zalogowanych], Li people. CO welding parameters optimization of artificial neural network design [J]. thermal processing, 2008,37 (5) :93-95. f21 Liu Xiaoling. BP algorithm for neural network prediction using tungsten tensile strength [J]. Materials Science and Technology, 2006,14 (1): 63-65. [3】 Yinhai Lian, Hu Zili. based on BP neural network performance prediction of composite materials [J】. Nanjing University of Aeronautics and Astronautics, 2006,38 (2): 234.238. [4】 MehmetSO.ArtificialneuralnetworkapproachtOpredicttheelectricalconductivityanddensityofAg-Nibinaryalloys [J]. JournalofMaterialsProcessingTechnology ,2008,208:470-476. Tin \
相关的主题文章：

[link widoczny dla zalogowanych]

[link widoczny dla zalogowanych]

[link widoczny dla zalogowanych]

The post has been approved 0 times