Parameterization Of Neural Qec - Siemens SINUMERIK 840D sl Function Manual

Recommended commissioning procedure
As mentioned above, the neural network integrated in the control automatically adapts the
optimum compensation data during the learning phase.
The axis involved must perform reversals with acceleration values constant section by
section. Before activation of the learning phase, the parameters of the neural network (QEC
system variables) must be pre-assigned in accordance with the requirements.
In order to simplify commissioning as much as possible, NC programs are provided as
reference examples.
As described in subsection "Commissioning the neural QEC", the commissioning engineer
must first learn the characteristic for the axes using these reference examples and the
recommended QEC parameter values and check the contour accuracy achieved using the
circularity test (see Section "Circularity test"). If the results do not meet the requirements, re-
optimization must be performed changing the parameters appropriately (see Section
"Parameterization of neural QEC", "Learning the neural network" and "Further optimization
and intervention options") (so-called "relearning").
5.6.3.2

Parameterization of neural QEC

Machine data
The basic configuring data for the neural QEC are stored as machine data.
● MD32490 $MA_FRICT_COMP_MODE (friction compensation method)
● MD32500 $MA_FRICT_COMP_ENABLE
● MD32580 $MA_FRICT_COMP_INC_FACTOR
● MD38010 $MA_MM_QEC_MAX_POINTS
With these machine data, the neural QEC is activated as soon as the memory space is
reserved in the non-volatile RAM. The procedure and assignments are described in Section
"Commissioning the neural QEC".
All other data are set using system variables.
Extended Functions
Function Manual, 09/2009, 6FC5397-1BP20-0BA0
(2 = neural QEC)
(friction compensation active)
(weighting factor friction compensation value with short traversing movements)
(maximum number of compensation values for QEC with neural networks)
5.6 Friction compensation (quadrant error compensation)
K3: Compensation
423