• Have a trapper that catches float values (once per minute, auto-mapped in simple graph).
• Attach a trigger that generates 'warning' when value is 0.

On a value sequence of, for instance, 1.0 1.5 0.0 0.0 0.0 0.0 2.5 you see in the graph that a direct line is drawn between the first occurrence of 0.0 and the last value of 2.5.

Visually, this looks like there were no more occurrences of the 0.0 value while in fact, there were another 3 of them...