Parts with varied curvature features play increasingly critical roles in engineering, and are often machined under high-speed continuous-path running mode to ensure the machining efficiency. However, the continuous-pa...Parts with varied curvature features play increasingly critical roles in engineering, and are often machined under high-speed continuous-path running mode to ensure the machining efficiency. However, the continuous-path running trajectory error is significant during high-feed-speed machining, which seriously restricts the machining precision for such parts with varied curvature features. In order to reduce the continuous-path running trajectory error without sacrificing the machining efficiency, a pre-compensation method for the trajectory error is proposed. Based on the formation mechanism of the continuous-path running trajectory error analyzed, this error is estimated in advance by approximating the desired toolpath with spline curves. Then, an iterative error pre-compensation method is presented. By machining with the regenerated toolpath after pre-compensation instead of the uncompensated toolpath, the continuous-path running trajectory error can be effectively decreased without the reduction of the feed speed. To demonstrate the feasibility of the proposed pre-compensation method, a heart curve toolpath that possesses varied curvature features is employed. Experimental results indicate that compared with the uncompensated processing trajectory, the maximum and average machining errors for the pre-compensated processing trajectory are reduced by 67.19% and 82.30%, respectively. An easy to implement solution for high efficiency and high precision machining of the parts with varied curvature features is provided.展开更多
A new approach on cutter path generation for plane milling is proposed. The cutter feed status at the position of each grid mesh can be determined by using a specific algorithm consisting of data pro- cessing and some...A new approach on cutter path generation for plane milling is proposed. The cutter feed status at the position of each grid mesh can be determined by using a specific algorithm consisting of data pro- cessing and some heuristic rules. From the cutter feed status and the coordinates of the grid meshes, the cutter path for milling plane can be generated.展开更多
针对循环水养殖饵料投喂劳动强度大、人力成本高等问题,本文设计了一种自动投喂机器人,并提出了基于物理增强神经网络的模型预测控制(Physics-informed neural network based model predictive control, PINN-MPC)方法,以解决变负载、...针对循环水养殖饵料投喂劳动强度大、人力成本高等问题,本文设计了一种自动投喂机器人,并提出了基于物理增强神经网络的模型预测控制(Physics-informed neural network based model predictive control, PINN-MPC)方法,以解决变负载、湿滑路面下的自主路径跟踪问题。首先,设计了机器人总体架构与路径规划控制方案。其次,构建了变负载、复杂环境下的机器人模型。然后,在传统MPC架构基础上,将关键物理参数视为时变因子,引入多层前馈神经网络对其进行在线预测,提升控制精度。最后,通过仿真和现场试验验证了算法有效性。在单缸投喂试验中,PINN-MPC在2个关键观测点的平均误差分别为0.12、0.18 m,较传统MPC降低50%;纵向速度波动幅度为MPC的50%,横向偏移标准差降低58.3%。在多缸投喂试验中,PINN-MPC将9个目标点间的平均路径误差控制在0.050~0.055 m,轮胎横向受力波动减少58.9%。展开更多
基金Supported by National Natural Science Foundation of China(Grant Nos.51575087,51205041)Science Fund for Creative Research Groups(Grant No.51321004)+1 种基金Basic Research Foundation of Key Laboratory of Liaoning Educational Committee,China(Grant No.LZ2014003)Research Project of Ministry of Education of China(Grant No.113018A)
文摘Parts with varied curvature features play increasingly critical roles in engineering, and are often machined under high-speed continuous-path running mode to ensure the machining efficiency. However, the continuous-path running trajectory error is significant during high-feed-speed machining, which seriously restricts the machining precision for such parts with varied curvature features. In order to reduce the continuous-path running trajectory error without sacrificing the machining efficiency, a pre-compensation method for the trajectory error is proposed. Based on the formation mechanism of the continuous-path running trajectory error analyzed, this error is estimated in advance by approximating the desired toolpath with spline curves. Then, an iterative error pre-compensation method is presented. By machining with the regenerated toolpath after pre-compensation instead of the uncompensated toolpath, the continuous-path running trajectory error can be effectively decreased without the reduction of the feed speed. To demonstrate the feasibility of the proposed pre-compensation method, a heart curve toolpath that possesses varied curvature features is employed. Experimental results indicate that compared with the uncompensated processing trajectory, the maximum and average machining errors for the pre-compensated processing trajectory are reduced by 67.19% and 82.30%, respectively. An easy to implement solution for high efficiency and high precision machining of the parts with varied curvature features is provided.
文摘A new approach on cutter path generation for plane milling is proposed. The cutter feed status at the position of each grid mesh can be determined by using a specific algorithm consisting of data pro- cessing and some heuristic rules. From the cutter feed status and the coordinates of the grid meshes, the cutter path for milling plane can be generated.
文摘针对循环水养殖饵料投喂劳动强度大、人力成本高等问题,本文设计了一种自动投喂机器人,并提出了基于物理增强神经网络的模型预测控制(Physics-informed neural network based model predictive control, PINN-MPC)方法,以解决变负载、湿滑路面下的自主路径跟踪问题。首先,设计了机器人总体架构与路径规划控制方案。其次,构建了变负载、复杂环境下的机器人模型。然后,在传统MPC架构基础上,将关键物理参数视为时变因子,引入多层前馈神经网络对其进行在线预测,提升控制精度。最后,通过仿真和现场试验验证了算法有效性。在单缸投喂试验中,PINN-MPC在2个关键观测点的平均误差分别为0.12、0.18 m,较传统MPC降低50%;纵向速度波动幅度为MPC的50%,横向偏移标准差降低58.3%。在多缸投喂试验中,PINN-MPC将9个目标点间的平均路径误差控制在0.050~0.055 m,轮胎横向受力波动减少58.9%。
