In a recent and very erudite article, Peter Natchwey of Delta Computers sets forth an analysis of PID position control of hydraulic actuators. Hydraulic actuators have their own complexities which we don’t need to consider, people have done text books on the subject. Hydraulic systems epitomize many aspects of mechatronics, pumping hydraulic fluid to create both pressure and flow, valve position to control flow, and in more complex systems, linear position of the hydraulic cylinder as part of the machine control.
Electrohydraulic systems have been around for many years, but recent advancements in electronics are bringing about changes to the field of hydraulics in much the same way microprocessors are transforming many fields. Low cost computers are able to manage multi-axis motion control applications with relative ease. The solution required in a hydraulic positioning application involves several additional loops to produce stable, accurate motion, but regardless of the mathematical complexity, today’s processor platforms are well able to handle the tasking.
The question at hand is how well does PID theory address positioning of a mechanical system. Proportional-Integral-Derivative control, PID for short, is a poor choice for controlling position of mechanical systems.
Error detection of a load that is in position is no problem. Nothing is moving and any disturbance is an easily detected value that can be corrected. The major question in managing a stable position is what gain should be applied? This becomes a complex problem to solve by itself, even before we consider the dynamics of the motion. The biggest factor is the load mass. Load mass predicts the time allowed for acceleration. A large mass like a 1000 pound roll of printing paper can only be made to accelerate at a certain rate, and there is not much you can do to change the situation. If load has a time constant, then the proportional gain required to stabilize the load while “in position” would have to be based on the time constant of the mass. In a step and direction system, position can be maintained with no PID control.
Regulation of motion generally follows some simple rules. When the load is moving at a constant speed, there is no acceleration or deceleration. So velocity is a rate of position change. Speed of the load can only be disturbed by changes that would show up as following error, the difference between commanded instantaneous position versus actual. Again, the gain setting required for proper regulation would be related to the mass and the time allowed for system corrections. In many systems, an allowable following error value is programmed in and no PID is required.
The simplified approach of looking at the individual states of motion within a given profile is not an end in itself, but a possibility to suggest that other rule bases could be derived and tested as alternatives to the traditional PID approach. At the end of the day, regulation of a mechanical load can be simplified based on the state of the motion, stationary, constant speed, or accelerating and decelerating. In my opinion, the best way to move the motion control industry forward is to begin re-writing the rules.
Tell Us What You Think!