Whether robots are assisting in assembly or moving storage and retrieval machines (SRMs) in automated warehouses, all examples require exact positioning and therefore maximum precision from the drive technology. The running speed must be maintained exactly. The best possible adaptation of the control system to the machine is a prerequisite.

Today, many control technology experts follow the theory and use physical models to automatically adapt drives to the characteristics of the machine, i.e. the controlled system. These models simulate the behavior of the real machine in a simplified way by taking masses, springs and dampers into account. If theory and reality matched, the problem would be solved. Unfortunately, deviations of the model from the real machine cannot be avoided: For example, the spring in the model is actually a shaft, and this is not massless, but heavy, or a gearbox has play and therefore a force-dependent elasticity that the model does not reproduce exactly. However, the control optimization can only be as good as the correspondence between the machine and the model. In addition, despite the abstractions, specialists are still required for the adaptation, who also have to invest an enormous amount of time and effort.

In industry, it is common practice to adjust the drive manually: To do this, it is brought to its - audible - stability limit, and the controller setting used is multiplied by a safety factor. According to the publication on the vibration method by Ziegler and Nichols from 1943, the factor is between 0.45 and 0.6 depending on the type of control loop (P/PI/PD/PID) - roughly half. This method is easy to apply and does not contain any model errors. The disadvantage is that the human sensor is not particularly accurate and an expert is required for the adjustment.

The drive as a sensor

Lenze has now developed an auto-tuning function based on the Ziegler-Nichols method. The latest versions of the i950 and i700 servo drives are already equipped with the autotuning function, and devices already in the user's possession can be retrofitted via a firmware update. No additional measuring components are required, instead the drive becomes a sensor: while the controller settings are being changed, the control logic evaluates the data from the position encoder and the current flow to detect when the stability limit is reached. The settings are then calculated and stored in the drive configuration.

For auto-tuning, Lenze relies exclusively on measuring the physical control path. Instead of abstraction and approximate values, there is an exact measurement in the real machine. Errors resulting from the use of a model can thus be avoided, as can the disadvantages of manual adjustment.

Automated controller optimization with the autotuner has several advantages: The controller setting is reproducible, so that the drive runs equally well on all systems. In particular, the optimization of the integral components of the controllers leads to significant improvements that go beyond the Ziegler-Nichols method by considering the tracking errors. In addition, autotuning is very robust, as the controller setting ensures a sufficient distance to the stability limit: it also works for gearboxes with backlash and with variable loads.

Autotuning function for retrofitting

Lenze had selected a number of plant operators whose existing machines had been retrofitted with the autotuning function. Up to 20 % higher cycle rates were achieved in on-site tests by optimizing the inverter parameters. This reduces the time required for future commissioning.

The autotuning function is implemented in the Easy Starter and PLC Designer tools. Both were updated for the Hannover Messe 2021 Digital Edition.

Author

Dr. Johannes Kühn is Head of R&D Motion Control at Lenze.