The research object used in the dissertation was screw pump, which was widely applied in engineering field. The analysis of end face profile formation was completed on its main working components, i.e., rotor and stator. With the purpose of finding optimal parameters to improve the efficiency of screw pump design, the key technologies involved in the parametric modeling of rotor and stator were analyzed. The three-dimensional (3D) design software SolidWorks was used for the secondary development and parametric modeling of rotor and stator. After that the simulation models of different kinds of screw pumps were established based on the cycloid type, variation coefficient, and screw head number. Finally the COSMOSMotion was used to analyze the motion characteristics on the equidistant line of rotor, including velocity and acceleration. This design and modeling method has been used in screw pump enterprises for design and development, laying the foundation for finite element analysis and further optimization of screw pump.
针对室内环境中视觉同时定位与建图(simultaneous localization and mapping,SLAM)精度不高和实用性较差等问题,采用深度相机作为传感器,提出一种基于改进词袋模型的视觉SLAM算法。该算法通过增加节点距离的方式,对传统的词袋模型进行改进,采用octree方法转化点云,生成可用于导航的八叉树图,并进行改进前后词袋模型对比实验、数据集精度实验和实验室实测。结果表明,改进后的词袋模型相似度计算能力和区分度更强,SLAM算法在环境有回环和相机运动较慢的情况下,效果较好,可满足室内同时定位与建图及后续导航需求。
