Self-optimization in compact controllers
Most compact controllers for the industrial sector include automatic adjustment of the control parameters to the connected system. This function is self-tuning called (Engl. Autotune). Two different types of self-optimization are often available in the controller.
Self-optimization according to the vibration method
During self-optimization, the actual value curve describes an oscillation , but - apart from the name - has nothing in common with the Ziegler-Nichols oscillation method. Before starting self-optimization, operating conditions should prevail in the controlled system and the later setpoint must be set. Self-optimization can be started when the system is cold or warm. For reasons of time, the former is particularly recommended for slow stretches, as the system is driven noticeably below the setpoint before the first output level is applied by self-optimization.
procedure
The controller first gives an output level jump of 100% to the system and uses the actual value curve to calculate a switching line below the setpoint. This switching line represents the switch-on and switch-off times during self-optimization. The switching line is calculated in such a way that the actual value does not overshoot, if possible. i.e. does not come above its setpoint. If the switching line is reached, the output level goes back to 0% and the actual value drops. If the actual value falls on the switching line, there is another jump in output level to 100%, which is set back to 0% when the switching line is reached.
As a rule, the calculation of the parameters is completed after two oscillations. This type of self-optimization determines suitable control parameters for most applications. The time required to complete it depends on the route. With an industrial furnace, this can sometimes take several hours. The oscillation method does not automatically guarantee that the actual value does not overshoot. If a critical actual value must not be exceeded under any circumstances, either a lower setpoint should be selected or, if in doubt, this method should be abandoned and the step response method selected instead.
Self-optimization according to the step response method
Before performing this method, the switching line for the subsequent jump height must be determined manually. This has the advantage that the user specifies the actual value curve himself during self-optimization and thus ensures that an unintentional overshooting of a critical actual value is avoided.
preparation
The controller is operated in manual mode and a manual output level setting is used to try out the output power at which the actual value is close to the later setpoint. The jump in output level of self-optimization should result in an actual value that starts noticeably below the setpoint and then runs counter to it. If exceeding the setpoint is not critical, a jump by the setpoint can be made. z. B. w = 500 ° C -> jump from 480 ° C to 520 °.
example
The system setpoint of 500 ° C must not be exceeded. By manually specifying the output level, 500 ° C is achieved through 75% output power. An output of 65% was determined for the degree of rest, which results in an actual value of 460 ° C. The difference is 10%, which is defined as the jump height.
procedure
Self-optimization can be started in any system status. The resting degree is given to the output and the controller waits until a stable actual value is established. As soon as this no longer changes, the jump height is added to the resting degree and the actual value curve is analyzed by the controller. As soon as this has reached its highest gradient, self-optimization is ended and the calculated control parameters are adopted.
This optimization method may take time due to the manual determination of the output levels, but has the advantage over the vibration method that the system does not have to be made to vibrate. The defined actual value increase is sufficient for implementation.
Individual evidence
- ↑ a b c Manfred Schleicher: Control engineering - Basics and tips for the practitioner . Jumo, Fulda 2014, ISBN 978-3-935742-00-9 ( free full text [PDF; 3.2 MB ]).