Structural Motion Control and Optimal Trajectory Planning for High-Productivity Manufacturing

在加拿大，每年通过制造业创造的价值超过260亿美元，这些行业依赖于计算机数字控制(CNC)加工和精密运动控制。这些领域包括航空航天(70亿美元)、汽车(170亿美元)、模具(5亿美元)、生物医疗设备(17亿美元)以及机器人和自动化(5亿美元)。通过创新新技术在这些部门实现的任何生产率提高或产品质量改进，都将对加拿大的竞争力和财富创造产生重大影响。这就是这一发现研究计划的目标，目标是开发新的运动控制策略和2)最优轨迹规划算法；能够提高在多轴机床和机器人上进行的制造操作的零件质量和生产率。运动控制研究的目标是通过应用同步位置和振动控制来提高生产机器进给驱动(即移动轴)的高速定位精度和动态刚性。这提高了可用于跟踪快速运动指令和抑制加工和摩擦干扰的带宽(即响应频率范围)，从而提高了零件的精度和质量。在高横移率下提高定位精度还可以提高生产率。然而，当机器的运行条件因姿态、部件装载/卸载或部件磨损而改变时，进给驱动器的动态响应也会改变。因此，在存在这种变化的情况下，运动控制算法保持其稳定性(为了安全)和性能(为了质量保证)是至关重要的。拟议的研究将调查新的稳健和自适应控制技术，能够处理这种变化，同时在上述制造部门使用的生产机器上实现可靠和切实的质量改进。这些将建立在申请人的早期工作的基础上，与工业中使用的最先进技术相比，进给驱动器的精度和硬度提高了40%-50%。第二个重点是开发沿3轴和5轴刀具路径的新进给速度(即刀具推进)优化算法，以便在机器和制造过程的物理限制范围内最大限度地减少制造周期时间。这是一个复杂的、非线性的问题。典型的自由形式加工刀具路径可能包含数十万个曲线段，对于这些曲线段，需要在运行时或以有效的离线方式计划时间最优进给速度。文献中复杂的算法产生较短的周期时间，但其计算复杂性阻碍了它们的工业实现。另一方面，工业控制器应用过度简化的假设，产生次优结果和保守的周期时间。申请人最近开发了一种高效有效的进给优化算法，目前正在加拿大制造的数控系统中进行商业化。该发现计划将研究更新的、潜在的更强大的算法，能够以更低的计算成本实现进一步的周期时间缩短，而不是来自文学或行业的最先进的轨迹优化技术。该发现计划将支持2名博士、2名硕士和5名本科生的培训，他们将在解决高科技制造中的两个重要问题方面进行基础研究。这项核心研究的产业化推广和应用还将有助于培养更多的高素质人才，从而在通过这一发现研究计划实现的技术创造和培训方面显示出倍增效应。

In Canada, over $26B of value creation takes place annually through manufacturing in industries that rely on Computer Numerically Controlled (CNC) machining and precision motion controls. These include aerospace ($7B), automotive ($17B), dies and moulds ($0.5B), biomedical devices ($1.7B), and robotics and automation ($0.5B). Any productivity increase or product quality improvement achievable in these sectors, by innovating new technologies, would have significant impact on Canada’s competitiveness and wealth generation. This is the objective of this discovery research program, which targets the development of: i) new motion control strategies, and 2) optimal trajectory planning algorithms; capable of enhancing the part quality and productivity of manufacturing operations carried out on multi-axis machines, such as CNC machine tools and robots.The research on motion controls targets improvement of the high-speed positioning accuracy and dynamic rigidity of feed drives (i.e., moving axes) of production machines, by applying concurrent position and vibration control. This enhances the bandwidth (i.e., responsive frequency range) available for tracking rapid motion commands and rejecting disturbances due to machining and friction, leading to improved part accuracy and quality. Enhancing positioning accuracy at high traverse rates also enables higher productivity. However, as a machine’s operating conditions change due to posture, part loading/unloading, or component wear, the dynamic response of the feed drives also changes. Hence, it is vital that the motion control algorithms retain their stability (for safety) and performance (for quality assurance), in the presence of such variations. The proposed research will investigate new robust and adaptive control techniques capable of dealing with such variability, while achieving reliable and tangible quality improvement on production machines employed in the mentioned manufacturing sectors. These will build upon the applicant’s earlier work, which has achieved 40-50% accuracy and stiffness improvement on feed drives over state-of-the-art techniques used in industry.The second thrust focuses on developing new feedrate (i.e., tool progression) optimization algorithms along 3- and 5-axis toolpaths, in order to minimize the manufacturing cycle time within the physical limits of the machine and manufacturing process. This is a complex and nonlinear problem. Typical freeform machining toolpaths may contain hundreds of thousands of curved segments, for which a time-optimal feedrate needs to be planned on-the-fly, or in an efficient manner off-line. Elaborate algorithms in literature yield short cycle times, but their computational complexity prevents them from industrial implementation. Industrial controllers, on the other hand, apply over-simplifying assumptions that yield sub-optimal results and conservative cycle times. The applicant has recently developed an efficient and effective feed optimization algorithm, which is currently being commercialized inside a Canadian-built CNC. This discovery program will investigate newer and potentially more powerful algorithms, capable of achieving further cycle time reduction at lower computational cost, over state-of-the-art trajectory optimization techniques from literature or industry.This discovery program will support the training of 2 PhD, 2 MASc and 5 undergraduate students, who will conduct fundamental research in tackling two important problems in hi-tech manufacturing. Industrial dissemination and application of this core research will also help train additional highly qualified personnel, thus demonstrating a multiplying effect in terms of technology creation and training achieved through this discovery research program.