I've seen two items that are nearly automated alert systems.  I have no idea how they might apply to your situation.  One is a Popilia japonica trap with a built in camera that can be checked online is being developed by a Portland, Oregon company and another is a device to remotely identify mosquitoes to species (Culicidae) using wing beat frequency designed by Darek Czokajlo.

Maybe in my case, we need to integrate with natural pheromone or live females to attract the male moths. 

In Switzerland and Germany (at least), the agricultural weather forecast includes warning for different pests based on climatic situations. That functions thanks to a lot of detailed knowledge: If you study insect species simultaneously in the fields (by pheromone or light trapping) and in the lab, the latter under controlled temperature and humidity / light conditions, you can determine the duration of the life cycle at defined temperatures and establish a value of day degrees for each generation. Registering the field temperatures automatically through modern micro climate devices, you can test your theory in praxis. Remember that not all day degrees are "active", this is, under and over the temperature threshold insects don't develop. They just survive in the shade or in the soil or elsewhere in the system. This is especially true if there is a daily amplitude and each temperature "window" is lasting just some hours. Knowing the species, the number of generations per year, the attractive plant parts which make a potential pest arrive in a crop alltogether allow to establish that forecast system you are looking for. It is the microclimate and climate conditions that trigger the development of insects in the fields. These are the elements that permit the weather forecast to include pest insect forecasts for farmers. If you want to know more, let's discuss it by e-mail [email protected]

there are many way to automate an alert system... I think (as for for an experimental plan) it depends of your pest, crop and social and economic context.

Different possibilities were earlier suggested to you. In some case video detection should be possible too.  it is a large thematic for applied science but really interesting.

You should look for this paper (for an example in controlled environment)  :

http://link.springer.com/chapter/10.1007%2F978-3-642-18333-1_16

But I'm aware for other possibilities (part of my job is now on monitoring tools and decision help process) so if you want to discuss you will be welcome ([email protected])