基于Philip入渗模型的土壤水分入渗参数BP预报模型

基于黄土高原区大量大田土壤入渗实测资料,借助BP神经网络模型建立了基于Philip土壤入渗模型参数的预报模型,并分别讨论了BP神经网络土壤水分入渗参数稳渗率预报模型、土壤水分入渗参数稳渗率预报模型和Philip模型的90min累积入渗量预测的单项和综合误差。结果表明:基于常规土壤理化参数土壤含水率、容重、黏粒含量、粉粒含量以及有机质等建立BP神经网络模型对Philip土壤入渗半经验半理论模型参数吸渗率S和稳渗率A进行预测是可行的。预测参数A和S以及Philip模型的90min累积入渗量的预测值与实际值的相对误差分别为2.074 60%、3.079 98%和2.037 56%,都在可接受的范围内;研究结果可为世界范围内大量实用的地面灌溉技术参数优化提供强有力的依据支撑。

Predicting the Rainfall-Triggered Landslides in a Forested Mountain Region Using TRIGRS Model

Landslides are natural disasters which can pose a serious threat to human and property in many areas around the world. The Transient Rainfall Infiltration and Grid-based Regional Slope-stability (TRIGRS) model was used to investigate the rainfall-induced shallow landslides in a forested mountain region, Korea. Various input data for TRIGRS model include time-varying rainfall, topographic characteristics, soil depth, material strength, and hydraulic properties. A series of calculations were conducted in determining the slope stability over the Jangheung region in Korea during the storm occurred on August 6, 1998. The results show that TRIGRS model captured about 64.1% of landslides that were extracted from the IKONOS2 imageries. The model demonstrated how the factor of safety changed with time during a storm considering both the transient and spatial responses of pore water pressure in its slope stability calculation.

树脂砂铸造球铁件表层球化不良的改进

对呋喃树脂砂铸造球墨铸铁砂箱加工面灰斑缺陷进行了金相组织分析,确认灰斑缺陷是由球化衰退导致,为此分析了灰斑部位和正常部位的化学成分,发现灰斑区域的S和C都明显高于正常区域,因此断定球化衰退是由于树脂砂中的S渗入到铸铁件表面,使球状石墨畸变而引起的。为解决灰斑问题,采取了如下防止措施:(1)在铸件渗S部位设置冷铁;(2)在砂型表面用涂料阻隔含S气氛渗入到铸件表层;(3)控制树脂及固化剂加入量,减小灼烧减量,使树脂砂w(S)量降低。生产结果显示,铸件加工表面灰斑问题获得解决。