柔性神经网络及其在开关磁阻电机建模与仿真中的应用

作为一种新型的神经网络,创建了具有高度柔性特性的网络结构,给出了柔性神经网络(FNN)的基本原理,并将其应用于开关磁阻电机(SRM）的建模与仿真,展示了SRM新型建模方法的主要优点。FNN在实现系统功能的同时,需要较少的神经元和迭代循环,大大降低了网络的复杂性,加速了网络的学习与实时计算速度。     

供应链企业业务外包决策模型研究

为适应供应链企业业务外包决策的需要，建立了供应链企业业务外包模型，分析了供需双方独立决策时各自的产量决策和生产能力投资决策．讨论了基于价格合同的外包形式，建立了2个企业进行业务外包决策的两阶段动态博弈模型，并证明了其存在子博弈精练Nash均衡．     

盾构隧道施工对富水软弱地层的扰动分析

为研究盾构隧道施工对富水软弱底层的扰动影响,以大连地铁某标段盾构隧道施工为例,首先基于修正剑桥模型建立土体本构关系,利用Shell结构单元模拟盾构初衬;然后采用流固耦合方法研究土体固结过程对盾构开挖引起软弱地层扰动问题;最后根据仿真结果与现场实测数据,绘制地表沉降对比分析曲线,给出盾构施工引起地表沉降的动态变化趋势.结果表明：孔隙水在盾构开挖完成后仍持续向隧道方向渗透,并引起距隧道较近区域的扰动趋势大于周围较远区域.本研究对提高富水软弱地层条件下盾构施工过程建模的准确性和实效性、指导盾构施工具有指导作用.     

特大型桥梁群桩基础承载特性离心模型试验研究

苏通大桥超大型主桥索塔群桩基础置于第四系土层中,其承载变形特性是设计急待解决的技术问题。通过离心模型试验,研究了5种工况下主桥索塔群桩基础的竖向承载变形特性,分析了承台竣工、裸塔竣工、成桥阶段群桩基础的沉降及桩顶荷载分布规律。结果表明,5种工况下主桥索塔群桩基础的极限承载力大于5 000 MN,成桥阶段的总沉降为148~186 mm,基础整体稳定,桩顶荷载成W形分布,桩底注浆和冲刷形态对群桩基础的承载变形特性有一定影响。     

Large motion mountain biking dynamics

A bond graph model of a mountain bike and rider is created to develop baseline predictions for the performance of mountain bikes during large excursion maneuvers such as drops, jumps, crashes and rough terrain riding. The model assumes planar dynamics, a hard-tail (front suspension only) bicycle and a rider fixed to the bicycle. An algorithm is developed to allow tracking of a virtual tire-ground contact point for events that separate the wheels from the ground. This model would be most applicable to novice mountain bikers who maintain a nearly rigid relationship between their body and the bicycle as opposed to experienced riders who are versed in controlling the bicycle independent of the body. Simulations of a steep drop are performed for various initial conditions to qualitatively validate the predictions of the model. Results from this model are to be compared to experimental data and more complex models in later research, particularly models including a separate rider. The overarching goals of the research are to examine and understand the dynamics and control of interactions between a cyclist and mountain bike. Specific goals are to understand the improvement in performance afforded by an experienced rider, to hypothesize human control algorithms that allow riders to perform manoeuvres well and safely, to predict structural bike and body forces from these maneuvers and to quantify performance differences between hard-tail and full suspension bicycles.     

Food supply mechanisms for cold-water corals along a continental shelf edge

In recent years it has been documented that deep-water coral reefs of the species Lophelia pertusa are a major benthic habitat in Norwegian waters. However, basic information about the biology and ecology of this species is still unknown. Lophelia live and thrive under special environmental conditions of which factors such as temperature, water depth, water movement and food supply are important. The present work explores the hypothesis that Lophelia forms reefs in places where the encounter rate of food particles is sufficiently high and stable over long periods of time for continuous growth. This is done by relating the distribution of reefs with the results of numerical ocean modelling.Numerical simulations have been performed with an idealized bottom topography similar to what is found outside parts of the Norwegian coast. In the simulations the model is first forced with an along slope jet and then with an idealized atmospheric low pressure. The model results show that the encounter rates between the particles and the water layer near the seabed are particularly high close to the shelf break. This may indicate that many Lophelia reefs are located along the shelf edges because the supply of food is particularly good in these areas.A sensitivity study of the particle supply in the area close to the seabed for increasing latitude has also been done. This shows that the Ekman transport in the benthic layer tends to create a steady supply of food for benthic organisms near the shelf edge away from the equator.