menu

Different ways of configuring self-learning limits are explained in more detail below. Within tolerance-range-based monitoring, the following procedures are possible:

In addition, for all these settings, there are the additional maximum value, wear limit and missing limit functions. These limits can be displayed as additional limits, in which case they are visible as small dashes.

All procedures configure themselves largely independently. They observe the sensor signal curve of one or more processes. If the signal exhibits an unusual deviation, an alarm is triggered.

The ToolScope uses evaluations to determine whether or not a sensor value is unusual. One of the procedures (Fixed) works with fixed intervals of the signal in the Y direction over the entire learning curve. The other procedures determine the intervals in the Y direction based on statistical evaluations. Based on the last observed processes (learning curves) they calculate:

  • Which value is currently expected for the observed signal
  • The uncertainty of this value

The greater the expected uncertainty, the greater the permitted deviations from the expected value.

The statistical (Statistical and Six Sigma) procedures work intelligently. If it has previously only been possible to observe a process a few times, they permit larger deviations and thus avoid false alarms.

Based on these and other evaluations, the decision is made as to whether a process can be classified as a good process or whether an alarm should be triggered.

It is possible to give the procedure several sensors for selection, which it is intended to monitor. Minor deviations in individual signals do not then result directly in alarm outputs. The ToolScope can thus perform very sensitive monitoring yet at the same time avoid false alarms.

The tolerance-range-based procedures deactivate themselves when processes last for longer than 4000 seconds. Thus, only the first 4000 seconds of a process are monitored. If a data rate of more than 100 Hz is set, the monitoring procedure is deactivated sooner. When delivered, no more than 100 Hz are set.

Breakage monitoring can and will only report an error if the given sensor signal changes noticeably after the occurrence of an error.

To use one of these monitoring procedures, you must select "Tolerance ranges (time)" in the "Monitoring" selection field of the configuration settings. In the corresponding drop-down menu, you must then search for the desired monitoring procedure.

 

 

Fully automatic

Fixed

Adaptive

Dynamic

Six Sigma

Aim of limit calculation

Limits are calculated such that a change in the signal behavior (e.g. breakage) is detected.

Limits are calculated such that a change in the signal behavior (e.g. breakage) is detected.

Limits are calculated such that a change in the signal behavior (e.g. breakage) is detected.

Limits are calculated such that a known signal dynamic is not exceeded.

Limits are calculated strictly in accordance with statistical criteria (mean value ± x · standard deviation).

Permanent learning behavior?

Any process assessed as "good" is added to the learning data. The respective oldest process in the learning data is deleted internally. The limits "follow" the signal.

The configured number of processes is learned once. Then learning is finished. The limits remain fixed after the learning process.

Any process assessed as "good" is added to the learning data. The respective oldest process in the learning data is deleted internally. The limits "follow" the signal.

Any process assessed as "good" is added to the learning data. The respective oldest process in the learning data is deleted internally. The limits "follow" the signal.

The configured number of processes is learned once. Then learning is finished. The limits remain fixed after the learning process.