Question 1

What does the SmartLimitWatcher do?

Accepted Answer

The tool sets dynamic limits for a critical process value (target variable). The software detects operating states that deviate from states for which the system has been trained with historical data.

Question 2

What is the significance of the target variable and support variables?

Accepted Answer

The target variable is the data source to be monitored, the support variables are used to describe the behaviour of the target variable. The target variable and the support variables must be in a measurable relationship (linear/non-linear). This relation must be appropriately captured in the available data. An unlimited number of support variables can be used.

Question 3

What settings does the SmartLimitWatcher Wizard offer after calculating a dynamic limit profile and which relevance has the adjustable sensitivity?

Accepted Answer

How sensitive the dynamic profile should work can be set via “high”, “medium”, “low”. Lower and upper alarm and warning limits can be activated and disabled separately. The sensitivity determines how big the process deviation may be before a warning or alarm is triggered.

Question 4

What happens after activating a dynamic limit of the SmartLimitWatcher?

Accepted Answer

After activation, the target variable is monitored by the SmartLimitWatcher. If a dynamic limit value is exceeded, a ticket is generated in moneo.

Question 5

What are the requirements for using the SmartLimitWatcher?

Accepted Answer

The SmartLimitWatcher functionality is activated by a separate license: DataScienceStarter LIC QM9103. A licensed moneo RTM module is mandatory.

Question 6

What’s the difference between the trained AI models?

Accepted Answer

There are linear and non-linear models. Decision tree and neural network belong to the artificial intelligence class, able to capture linear and non-linear process dependencies. Linear regression captures linear dependencies only.

Question 7

Does the SLW follow an unsupervised or supervised machine learning approach?

Accepted Answer

The SLW follows an unsupervised approach.

SmartLimitWatcher FAQs