Sustainability Index (SI)

The Sustainability Index is an Anomaly Detection index that indicates how much the ratio between production and waste deviates from the average. In other words, it represents how the current consumption differs from the mean.

The index is calculated for each consumable individually, thus indicating how the performance of each consumable deviates from the reference, but it can also be calculated globally as the average of the individual indices, referring to overall consumption.

To calculate the optimal reference consumption, training is conducted within a selected time period to obtain the average and standard deviation values for the phenomenon.

The start and end of the reference period can be set on the Machine Settings page, along with other specific parameters that need to be used for the index calculation, such as the Severity Coefficient (nSigma) and the cost of each consumable.

The formula used for calculating the SI index is as follows:

SI(x)=1−1/(1+e((consxRelMean−consx)/totCount)/consxRelStd+nSigma)SI(x)=1−1/(1+e((consxRelMean−consx)/totCount)/consxRelStd+nSigma)

where:

  • x represents the selected consumable;

  • SI(x) will be the Sustainability Index for consumable x;

  • consx refers to the quantity of consumable used;

  • totCount refers to the total number of units produced;

  • consxRelMean represents the expected mean value for that consumable;

  • consxRelStd represents the standard deviation for that consumable;

  • nSigma, also known as the Severity Coefficient, is a multiplier of the standard deviation and can be set by the user on the dedicated page (Machine Settings). The value associated with this parameter allows for adjusting the sensitivity of the index to variations. Increasing its value makes the index less sensitive to changes, while decreasing it increases the sensitivity of the SI index to changes in consumable performance.

The use of the standard logistic function allows for producing a value between 0 and 1, ensuring the management of values in a gradual manner.

Was this helpful?