This kind of design assignment should not be too difficult to break down into constituent parts, so you can replace one big problem with a bunch of smaller problems. Hopefully you already know how to solve some, or all, of the smaller problems.
If you start at the output and work backwards- you want 3 LEDs driven by some kind of circuit that controls their illumination. Let's say you have a "DC" voltage that represents the RMS AC input voltage that you've been asked to measure. By "DC" I mean rectified and low-pass filtered so that it has little ripple. Say the voltage is 10V for 240V RMS, 9.5V for 5% low, and 10.5V for 5% high. So you need to design a circuit that will illuminate the Red, Green or Yellow LED based on that voltage (3 states, so it can be defined with two comparison bits). That's one smaller problem.
A second problem is how to power the circuit. You know you have a step-down transformer, so you should be able to design a power supply. But wait- there's an issue here with the specifications. You're told to illuminate a yellow LED if the voltage is 5% or more below nominal, but it's going to be hard to do that at 0V. You may have to make a reasonable assumption here- say it will work down to 30% under nominal. So your power supply has to work with as low as 160V in, and still provide (say) 15V regulated for the circuit to work.
That's smaller problem number two.
The third problem is how to get a voltage representing the RMS voltage into a DC voltage. One approach is to use a rectifier or precision rectifier circuit and rectify and filter the output voltage of the transformer. It's easier to measure the average value of the rectified voltage than the RMS value and assume it's a sine wave (this is where your AC analysis might come in, there is a constant factor between the two for a sine wave). This is really three even smaller problems- rectify the voltage, filter the voltage and (perhaps) scale the voltage so that it meets our requirements in the first problem of 230VAC->10.0V output.
So, a total of five smaller design problems, and we've detected a deficiency in the specifications. This is a fairly representative assignment in terms of what you'll run into, in miniature, but all the elements are there.
One little enhancement I'll recommend- keep the current draw (especially of the LED circuit since it will draw the most) constant regardless of which LEDs are illuminated. If you can describe why that's a good thing, you may get bonus marks.
It looks pretty good, but there're a couple issues I can see.
One downside is you're just about doubling the current by using a 470 ohm pot for the detection part. I don't know how long you expect to have this circuit attached after the battery reaches the threshold for the red to turn on, but this circuit could excessively drain your batteries if you don't unplug it or charge it soon after seeing the red-light.
You should also make sure you're very careful when setting up/tuning your circuit. 1.2V * 6 = 7.2 which is higher than what this chip is rated for. I know you can get slightly higher than that with NiMH's so start at the low side when you're tuning the pot and work your way up. Another option is to add a resistor between the pot and the IC along with a 7V zener diode in parallel with the IC to protect it. Alternately, a higher Vdd IC could be sought.
Best Answer
Try the LM3914.
Example circuit. This outputs to 10 LEDs, but you can short groups of outputs together and operate it in "dot" mode to give you three different indicators at whatever thresholds you want.
It doesn't supply auto-cutout, but you could drive a transistor from one of the lower indicators or get a voltage supervisor IC of some kind.