压缩体素模型在五坐标数控仿真中的应用

针对五坐标数控加工仿真中刀具扫描体和零件毛坯之间布尔运算量大,传统基于空间分解的体素建模方法需要大量计算机内存空间,仿真时间很长,仿真精度低等问题,结合Dexel模型提出了一种压缩体素模型,极大地简化了仿真过程中的布尔运算,同时有效地压缩了仿真模型的数据量,仿真的精度也有所提高.并且根据压缩模型特点,设计了基于Marching Cubes的优化表面三角网格提取算法,加快了仿真结果的图形显示.整个加工过程的仿真速度比传统体素模型提高了8～9倍,在实际应用中取得了良好的效果.     

Precise algebraic-based swept volumes for arbitrary free-form shaped tools towards multi-axis CNC machining verification

This paper presents an algebraic based approach and a computational framework for the simulation of multi-axis CNC machining of general freeform tools. The boundary of the swept volume of the tool is precisely modeled by a system of algebraic constraints, using B-spline basis functions. Subdivision-based solvers are then employed to solve these equations, resulting in a topologically guaranteed construction of the swept volume. The presented algebraic-based method readily generalizes to accept tools of arbitrary free-form shape as input, and at the same time, delivers high degree of precision.Being a common representation in CNC simulations, the computed swept volume can be reduced to a dexels’ representation. Several multi-axis test cases are exhibited using an implementation of our algorithm, demonstrating the robustness and efficacy of our approach.     

基于Dexel模型的NC加工仿真和验证研究

对基于Ｄｅｘｅｌ模型的ＮＣ加工仿真和验证进行了研究。在物空间中模拟ＮＣ加工的材料去除过程及验证加工结果，使用独立的坐标系构造物体的Ｄｅｘｅｌ模型，以支持观察定义的动态改变，并通过“移动实例”来近似描述刀具扫描体，从而提高五轴ＮＣ加工仿真计算效率。     

圆柱齿轮滚齿加工三向Dexel模型仿真与验证

针对基于实体建模技术的滚齿加工仿真问题,开发一种圆柱齿轮滚齿加工三向Dexel模型仿真方法。研究圆柱齿轮滚齿加工的运动学模型。根据滚刀和齿轮毛坯的几何参数创建三角形网格模型,并转换为引擎内的Dexel模型。同时,借助Delaunay三角剖分和Alpha形状重构进行高效的CWE计算和未变形切屑几何仿真。采用斜角切削模型来计算每个时间步上所有啮合节点的力分布。使用利勃海尔LC500滚齿机对所提仿真方法进行了验证,并结合旋转测力计和卡尔曼滤波器来补偿结构动力学,从而进行准确的切削力测量。结果表明：所提方法能够准确预测出沿滚刀刃口离散节点的三维力分布,预测误差(均方根和标准差)在4%～12%之间,且当切削条件变化时,仍然可以准确预测轴向和横向的切削力,有助于进一步提高滚齿加工仿真的效率与精度。     

五轴数控铣削加工图形仿真技术

本文提出了五轴铣削加工仿真的实现方法，在此基础上了五轴切削加工仿真，较好的再现了加工中毛的成型过程。     

A Visibility Sphere Marching algorithm of constructing polyhedral models for haptic sculpting and product prototyping

This paper presents a Visibility Sphere Marching algorithm of constructing polyhedral models from Dexel volume models for haptic virtual sculpting. Dexel volume models are used as the in-process models representation during interactive modification in a haptic virtual sculpting system. The stock material represented in a Dexel volume model is sculpted into a designed model using a developed haptic sculpting system. The sculpted Dexel volume models are converted to polyhedral surface models in STL format by the proposed visibility sphere marching algorithm. The conversion turns out to be an interesting and challenging problem. The proposed visibility sphere marching algorithm consists of three sub-algorithms: (i) roof and floor covering, (ii) wall-building, and (iii) hole-filling algorithms. The polyhedral surface models converted from the Dexel volume models can then be input to and processed by available computer-aided manufacturing (CAM) or rapid prototyping systems. The presented technique can be used in virtual sculpting, CAD/CAM, numerically controlled machining verification and rapid prototyping.     

基于射线投影的多视场深度像融合

首先确定匹配好深度像的最小轴对齐包围盒.通过盒面,沿着x,y,z 3个轴方向投射3个均匀分布的射线组,计算得到与所有深度像的交点,并通过Dexel结构来存储射线组与交点.加入距离和法向夹角2个判断依据来去掉重叠点,从而将多个视场深度像融合为完整的、无冗余的三维模型.实验结果表明,该方法误差小、速度快,而且简单有效.     

Efficient stochastic simulation of thermal spray processes

Thermal spray processes allow the production of coatings with a wide variety of useful properties, and therefore are used in a lot of different industrial applications. However, in specific applications, it can be a challenging and time consuming task to tune the parameters of a thermal spraying system so that required surface properties, such as wear- , corrosion- or heat resistance, can be achieved. Computer-aided stochastic simulations of thermal spray processes try to overcome these problems by modelling the coating build-up on a per-droplet basis, and by deriving the properties of interest from the generated structure afterwards. Because of the large number of droplets that need to be treated during the simulation to generate a sufficiently large virtual coating, the use of efficient algorithms is essential. On the other hand more complex algorithms are needed to increase the accuracy of the simulation.In this report we present novel extensions to a simulation model by Ghafouri et al. concerning plasma spraying. The extensions include a physically inspired splat spreading approach, a method to generate splat fingers, and a filtering technique that allows for the calculation of pores below overlapping splats. Furthermore, features characterizing coatings beyond the commonly used porosity or surface roughness are presented and analyzed with respect to their appropriateness. Features are considered appropriate if they are depending in a unique way on the parameters of the simulation model. This kind of features might be used in an automatic adaptation of the simulation model to real processes. Finally, several contributions are made in order to increasing the computational efficiency of the simulation. Most important, dexel models are introduced to represent the coating.     

Identification of the specific cutting force for geometrically defined cutting edges and varying cutting conditions

Cutting force modeling is a major discipline in the research of cutting processes. The exact prediction of cutting forces is crucial for process characterization and optimization. Semi-empirical and mechanistic force models have been established, but the identification of the specific cutting force for a pair of tool and workpiece material is still challenging. Existing approaches are depending on geometrical idealizations and on an extensive calibration process, which make practical and industrial application difficult. For nonstandard tools and five axis kinematics there does not exist a reasonable solution for the identification problem.In this paper a co-operative force model for the identification of the specific cutting forces and prediction of integral forces is presented. The model is coupled bidirectionally with a multi-dexel based material removal model that provides geometrical contact zone information. The nonlinear specific forces are modeled as polynomials of uncut chip thickness. The presented force model is not subjected to principal restrictions on tool shape or kinematics, the specific force and phase shift are identified with help of least square minimization. The benefit of this technique is that no special calibration experiments are needed anymore, which qualifies the method to determine the specific forces simultaneously during the machining process. In this paper, experiments with different cutting conditions are analyzed and systematically rated. Finally, the method is validated by experiments using different cutting conditions.     

Dexel-Based Direct Slicing of Multi-Material Assemblies

Slicing is an important procedure in rapid prototyping (RP) pre-processing, and can be grouped into two categories: direct slicing and adaptive slicing. At present, investigations into the use of both direct and adaptive slicing methods are taking place. However, not many direct slicing approaches have been reported in the literature. The methods are also restricted to some solids in CSG or some CAD systems. Also, approaches on adaptive slicing are too complicated. The method proposed in this paper employs dexel encoding for direct slicing multi-material (MM) assemblies in RP. One advantage of using a dexel model is that Boolean operations can be performed simply on 1D line segments. Dexels can also be easily converted to tool paths in RP machines. Compared to the ray representation of CSG trees, dexel models can be extended to represent MM assemblies with material properties. Therefore, the method has high potential for direct slicing. In this paper, traditional dexel models are first extended for rapid manufacturing single solid and MM assemblies. Compared to other adaptive slicing approaches, a much more efficient and simple dexel model, for adaptive refinement in the building direction is then developed. To further improve the surface finish, a layerwise refinement approach is also discussed. Finally, the computational complexity of the proposed method is studied.