Alerting Logic for batch temperature readings

I need some help determining some alert logic, can’t get my head around a way to do this, hoping someone can help out.

Setup: Grafana 10 w/ Influx 1.8 Database
Metric: Temperature, readings taken every 15 mins, batch uploaded every 6 hours.
Source - NB-IOT Sensor
Alert Goal - Raise an alert when the temperature is consistently over 90 degrees for 18 hours.

Use Case: I am monitoring the temperature inside of a box that is outdoors. As you would expect, the temperature fluctuates based upon the outdoor temperature. When the box becomes occupied with a swarm of honeybees, the temperature should stabilize around 90-95 degrees consistently, without fluctuation.

The Challenge - My current alerting logic basically is written: If mean (avg) temperature of the last 18 hours is greater than 90F, send alert. The problem is that the temperature inside the boxes gets hot during the day, which shifts the average up, resulting in a false positive.

What I need it to do is alert only when the temperature is consistently above 90 degrees. I need to make use of the “pending period” I think. The challenge is that data is only uploaded every 6 hours (to save battery life of the IOT unit.) The temperature readings are taken every 15 mins though.

Any thoughts on how I can get this alerting to work as desired?

Normal day to day temperature fluctuations.

When occupied with bees:

I think I figured it out… I used classic condition, then used min().