As motion control engineers, the
question is, do we keeping tuning
servo motor PIDs as we have done
for decades, or do we stop and ask,
“'Is there a better alternative?” INSIGHT# 2010-09MD H FEBRUARY 25, 2010
Improving Production Machinery through Servo System Performance
By Sal Spada and Himanshu Shah
Keywords
PID, Proportional Integral Derivative Control, Feedback, Stability
Overview
Is it possible that the venerable proportional-integral-derivative (PID) con-troller has seen the end of its useful life? As motion control engineers, the question is, do we keeping tuning servo motor PIDs as we have done for decades, or do we stop and ask, “'Is there a better alternative?” PID control
algorithms touch every facet of automation, but domi-nate in the servo motor control applications for robotics, machine tools, and packaging machinery. Improvements in motion control systems have largely
involved ease-of-use, facilitated by the move from ana-
log to all-digital systems. It is time to look for improved algorithms that can support the higher positioning accuracies, higher velocities, and prod-uct variations today’s manufacturers require. Moreover, modern controls need to move beyond the static mechatronic model of a production ma-chine and consider mechanical wear, environmental variations, and wider ranges of operation.
Change Is Coming to Machinery Automation
Perhaps the most compelling issues facing the automation market are to improve performance, minimize initial investment in design and develop-ment, and ensure that the systems perform according to the original design goals. Designing controllers for production machinery often involves many different elements of closed-loop control, but the servo motor loop largely dictates the performance of the machinery. In packaging and converting machinery, there are from three to sixty independent servo loops. Metal-
working machinery and robotics typically have three to eight loops.
Despite all the advances in advanced
control theory since the mid 1960's
when state space techniques became
prevalent, the most dominant closed
loop compensation scheme in single
output single input systems remains
the PID compensator. Despite all the advances in advanced control theory since the mid 1960s when state space techniques be-came prevalent, the most dominant closed loop compensation scheme in single input/single output systems remains the PID compensator. Certainly, engineers have augmented PID over the years with feedforward algorithms, anti-windup, gain schedul-
ing, notch filters, and nonlinear compensators,
which extended the life of this relatively simple controller. Furthermore, numerous motion control suppliers offer pole placement algorithms as an alternative, but these are much more difficult to tune and are very sensitive to parametric variations in the system.
PID, used in the industry for nearly 100 years, has remained popular pri-marily because the tuning techniques are prescriptive and the solution is relatively forgiving. The Zeigler-Nichols paper, published in 1942, pro-vided a very clear set of rules that engineers could follow to tune these systems effectively. However, the reality is that any application – whether a machine tool or a packaging machine – must periodically be retuned dur-ing its operational life, since the mechanics are subject to change. In some
cases, particularly in robotics, the payload
varies and the systems are generally de-
tuned to ensure the system is stable.
Many suppliers have attempted to auto-
mate the tuning of the PID controller fully
upon initial installation and, in some cases,
continuously during operation. However, the motion control industry has not widely
accepted continuous adaptive tuning,
mainly because of the requirement to interject an excitation with a suffi-ciently rich input into the servo motor system. This is generally necessary to ensure identification of the system parameters. Moreover, computation-al constraints, model rigidity, and switching costs further inhibited adoption.
Mechatronic Performance Demands Adaptive Solutions
The benefits of an adaptive controller begin at the conceptual phase of pro-
duction machine design. An engineer typically simulates the dynamics Traditional Proportional Integral Derivative Controller
The optimal solution that uses adaptive control technology is invariant of the
system parameters, actively rejects
disturbances, monitors in-process
quality, provides a predictive
maintenance indicator, and only requires
minimal tuning. It is impossible not to
find this concept compelling. based on a mathematical model of the electro-mechanical system to avoid having to return to the design board once a physical prototype is built. Al-though based on the selection of various electrical and mechanical components with well-defined characteristics, the exact performance of the design is still difficult to predict. This is because the final integration of components can introduce unpredictable results. An accurate model of be-havior is difficult to simulate when integrating complex components into a complete system. The modeling accuracy therefore breaks down in the real-world design.
Many simulation and analysis tools are available to form a bridge between the system mechanical design, control system design, and the prototype or implementation development phases. Engineers often augment the simula-
tion and modeling phase of the design process
with empirical analysis on the actual system to determine unmodeled dynamics not envisioned or incorporated in the original simulation. But what if the control algorithm could be made system-independent, thus no longer requiring this level of analysis? Adaptive controllers are based
on the assumption that the automation engineer
does not know the system parameters accurately, nonlinearities, or the sto-chastic model of the disturbances. Many variations on adaptive controllers exist. These include model reference, hybrid model reference, and system identification adaptive controllers. The algorithms are relatively complex and generally require some understanding of the underlying system (i.e., state space model) to prevent the system from going into an uncontrollable state. The optimal solution that uses adaptive control technology is inva-riant of the system parameters, actively rejects disturbances, monitors in-process quality, provides a predictive maintenance indicator, and only re-quires minimal tuning. It is impossible not to find this concept compelling. Adaptive Controls a New Generation of Servo Controls
Sustainability increases the demand from industry for control technology to move beyond PID, providing the motivation for developing adaptive con-trollers. Specifically, adaptive controllers that address four areas of weakness in PID: 1) passive response to the tracking error; 2) noise degra-
dation in the derivative control; 3) oversimplification and the loss of per-formance in the control law; and 4) complications of integral control. Conclusion
?System characteristics generally change over time. Users should de-mand automation controllers that ensure the performance of the system adapts to changes and provides in-process quality and system health indicators for better operations and maintenance decisions. ?Sustainability issues, whether energy consumption, improving quality of finished goods, or predicting maintenance, can all benefit from adap-tive controllers.
For further information or to provide feedback on this Insight, please contact your account manager or the author at sspada@https://www.wendangku.net/doc/fc153231.html,. ARC Insights are pub-lished and copyrighted by ARC Advisory Group. The information is proprietary to ARC and no part of it may be reproduced without prior permission from ARC.